The Chartered Market Technician program builds from identifying technical concepts to applying them and then combining them in a defensible market analysis. At Level I, learn what a chart represents, how trends and patterns are described, and why investor behavior can affect price. At Level II, test competing interpretations: a breakout can look convincing while breadth weakens, or a trading rule can look profitable because its parameters were fitted to old data. At Level III, connect the evidence to a position decision, including an entry, a reason the idea would fail, position size, and a plan for changing conditions. The concepts below follow those three stages. Each example uses stated assumptions so you can check both the calculation and the judgment it supports.
Level I - Theory and History
1. Dow Theory and market confirmation
Dow Theory examines broad market movements through related averages. The CMT Program Guide includes confirmation between averages among its Level I sample topics. In an established primary uptrend, successive significant highs and lows provide the main trend evidence. A new high in one average gains credibility when the other also exceeds its corresponding prior high. If it does not, the result is a nonconfirmation: a reason to examine the trend more carefully, not an automatic sell signal. The averages can diverge temporarily because their constituent businesses face different pressures. Separate the primary trend, which concerns a broad sustained move, from a secondary reaction against it and smaller day-to-day fluctuations. To judge a reversal, look for more than one disappointing week: assess whether meaningful swing lows have failed and whether both averages support the change.
Worked example: Suppose an industrial average exceeds its previous significant high of 1,000 and reaches 1,030, while a transport average remains at 490 below its previous high of 500. The industrial high is observable, but the paired highs do not confirm it. If both averages still hold their significant rising lows, the primary uptrend remains possible. One weak week does not establish a primary reversal.
Watch the distinction: Treating nonconfirmation as proof of a bear market skips the trend test. It weakens confidence in the new high; it does not show that both averages have reversed.
Reference: [1]
2. What price and volume reveal
A price chart records transactions, while volume counts how much changed hands during the observation period. A close above a well-tested resistance area establishes a price breakout under a closing-price rule. Volume helps assess participation: unusually high turnover suggests more trading accompanied the move, whereas light turnover invites caution about its durability. Neither observation identifies the traders or their reasons. A large institution might buy, sell, or do both during a high-volume session, and reported aggregate volume cannot reveal that distinction. Compare volume with a relevant baseline, such as recent sessions with similar trading conditions, rather than an isolated number. Then check whether price stays above the old boundary, extends the move, or returns inside the prior range. That sequence tests the breakout more directly than a story about who supposedly caused it.
Worked example: A stock repeatedly stalls near 100, then closes at 104. If recent comparable sessions averaged 2 million shares and breakout-day volume is 1 million, participation is half that baseline: 1 million divided by 2 million equals 0.5. Record a close above resistance with light relative volume. A later close back below 100 would challenge the breakout; the chart alone cannot identify institutional buyers.
Watch the distinction: A breakout on low volume is still a breakout under the stated price rule. Calling it invalid automatically confuses weaker supporting evidence with absence of the price event.
Reference: [1]
3. Trend depends on the horizon
A trend must be named with its observation interval. On a weekly chart, an uptrend is commonly described by rising significant swing highs and swing lows. A daily decline can occur inside that weekly structure because shorter moves often retrace part of a longer advance. Choose comparable swings before classifying them: a minor intraday dip should not be compared with a major weekly trough. A practical sequence is to identify the chart interval, mark successive meaningful highs and lows, then state which level would challenge the current classification. Closing-price rules can make the test more consistent, but they must be chosen in advance. Multiple horizons can give different, simultaneously valid readings. That is useful for decisions: a longer-horizon investor may see a pullback toward support, while a short-horizon trader may face an active decline.
Worked example: Weekly highs of 80, 88, and 94 rise, as do weekly lows of 72, 78, and 85. Those two sequences support a weekly uptrend. If the last three daily closes are 93, 91, and 89, they show a daily decline within it. A break of the relevant weekly swing low at 85 would carry more weight for reassessing the weekly trend.
Watch the distinction: Three falling daily closes do not erase a pattern of rising weekly swings. The error comes from changing time intervals without stating which trend is being judged.
Reference: [1]
4. Technical and fundamental evidence
Fundamental analysis estimates economic value from assumptions about cash flows, growth, financing, and risk. Technical analysis examines what buyers and sellers have actually paid and how price has behaved over time. These answer different decision questions. A valuation above the current quote may suggest prospective return if the estimate proves sound, but it does not say when other participants will agree. Conversely, an upward price break can guide timing without proving that the asset is cheap. When evidence conflicts, state the valuation case and the market-action case separately. Examine what could invalidate each: revised earnings assumptions for valuation, or a defined trend or resistance break for technical timing. This avoids silently converting a long-term estimate into an immediate entry signal. Price can reflect information and expectations, but its movement alone does not reveal a unique intrinsic value.
Worked example: An analyst estimates value at 60 per share while the stock trades at 45. The estimated gap is 15, or 15 divided by 45, about 33.3% of the current price. Yet price remains below a falling trend line. The valuation supports a possible upside case if its assumptions hold; the chart has not supplied an upside timing signal under a trend-break rule.
Watch the distinction: Calling the stock a technical buy solely because estimated value exceeds price mixes two methods. A valuation gap can persist or widen while the observed downtrend continues.
Reference: [1]
5. Market efficiency and technical signals
Weak-form market efficiency holds that information contained in past prices is already reflected in current prices, making persistent excess returns from price history alone difficult to obtain. A technical rule therefore needs evidence beyond an attractive chart or an in-sample gross return. Define its signal, test it on data that did not determine the rule, and deduct realistic commissions, spread, slippage, and financing where relevant. Compare the remaining return with the risk taken and with a suitable alternative. Conditions also change: a pattern observed in one market regime may weaken when participants adapt or trading costs shift. Weak-form efficiency is a proposition about predictability, not a claim that every observed price sequence is random or that no rule can ever work. A usable signal must survive implementation costs and reasonable checks for chance and overfitting.
Worked example: Suppose a rule averages a hypothetical 0.4% gross gain per completed trade. Commissions and spread total 0.3%, and estimated slippage adds 0.2%. Net expectancy is 0.4% minus 0.3% minus 0.2%, or minus 0.1% per trade. Even if the gross estimate is reliable, it does not establish a profitable strategy at those costs. Testing other periods would address a separate robustness question.
Watch the distinction: A positive gross average is not a tradable edge by itself. Ignoring execution costs can reverse its sign, while fitting the rule to past data can exaggerate it.
Reference: [1]
6. Support as a zone of trading
Support is an area where past declines repeatedly met enough buying to halt or reverse them; resistance is an area where advances repeatedly met selling. Mark a zone around reactions rather than pretending every turn occurs at one exact price. The zone's usefulness depends on the quality and recency of the reactions, the size of subsequent moves, and whether price approaches it from above or below. After a decisive break beneath support, a rally back toward that area may encounter selling from holders seeking an exit or traders responding to the former boundary. This is a possible role reversal, not a mechanical law. State a break criterion, such as a close beyond the zone, then observe the retest. If price promptly recovers and holds above it, the supposed breakdown has weaker evidence.
Worked example: Price rebounds three times between 49.50 and 50.50, making that interval a plausible support zone. It later closes at 48, below the zone. On the next rally it reaches 50 and turns down. That sequence fits a break followed by a test of former support as resistance. A sustained recovery above 50.50 would instead challenge the resistance interpretation.
Watch the distinction: Drawing support at exactly 50 and expecting every trade there to reverse ignores normal price variation. A prior reaction identifies an area to monitor, not a guaranteed turning point.
Reference: [1]
7. Behavioral sources of trend
Trends can persist when information is incorporated gradually. Underreaction describes an initial price response that falls short of the eventual collective reassessment; later buying or selling extends the move. Herding offers another possible mechanism: participants take cues from others' actions, so a rising price attracts additional demand. Both mechanisms are consistent with momentum, but a chart cannot prove what any trader believed. New information, changing discount rates, liquidity, or unrelated news could produce a similar sequence. For analysis, separate the observed pattern from the explanation: record the news date, initial reaction, subsequent highs and lows, and whether volume or breadth changes. Then ask whether later evidence supports continuation or shows exhaustion. Behavioral finance is most useful when it produces a testable interpretation rather than a confident story assigned after the fact.
Worked example: Favorable news arrives when a stock trades at 20. It closes at 22 that day and reaches 27 several weeks later, a further gain of 5 divided by 22, about 22.7%. Gradual reassessment or herding could explain the extension. The price path establishes the extension, but without evidence about participants' decisions it cannot establish either motive as the cause.
Watch the distinction: Seeing a prolonged rise and declaring herding proven mistakes a plausible explanation for an observation. The same prices could reflect additional news or a revised valuation.
Reference: [1]
8. Anchoring and loss aversion
Anchoring occurs when a reference number receives more weight than its relevance warrants. For a position holder, the purchase price is especially vivid, but it is a historical transaction rather than a reason the asset should trade there again. Loss aversion can reinforce the anchor: realizing a loss feels more painful than holding an uncertain position, even after the original trade case fails. Counter this by defining an invalidation condition before entry and reviewing new evidence against that condition. The planned exit should reflect the market setup and risk allowance, not a desire to recover the entry price. After entry, distinguish information that genuinely changes the thesis from discomfort about recording a loss. Stops may fill at a different price in a gap, so position sizing should allow for execution uncertainty as well as the planned price distance.
Worked example: A trader pays 100 per share and sets an exit at 88 if the setup fails. New evidence meets that condition, but the trader proposes waiting for 100 to break even. Following the stated plan at 88 produces a loss of 100 minus 88, or 12 per share before costs. Waiting for 100 expresses attachment to the entry price, not a revised market thesis.
Watch the distinction: The original purchase price does not become support just because a trader needs it to break even. Refusing the planned exit increases exposure after the stated thesis has failed.
Reference: [1]
Level I - Classical Techniques
9. Choose the chart that fits the question
Chart form determines which parts of a trading interval you can inspect. A conventional line chart joining closing prices makes longer trends easy to see but usually omits each period's open, high, and low. A bar chart displays those four prices, including the full range and where the session closed. A candlestick displays the same four values through a body between open and close and shadows extending to the high and low. Choose the form to match the question. Closing-price trend work may need only a line; rejection of an intraday advance requires the high and close. Candlestick color or body shape adds a quick visual cue but no extra underlying price data beyond the four values. In Level I chart study, keep intervals consistent before comparing shapes across securities.
Worked example: A session opens at 40, trades as high as 44 and as low as 38, then closes at 39. A close-only line contributes the point 39. A bar or candle shows the six-point range, 44 minus 38, and a close one point below the open. The retreat from 44 to 39 supports an intraday rejection reading that the line alone cannot show.
Watch the distinction: A candlestick is not a separate source of market information from an OHLC bar. Both display the same four prices; their visual emphasis differs.
Reference: [1]
10. Drawing a trend line
An upward trend line connects comparable rising swing lows on a chart with a consistent time scale and price scale. Two points define a line; a later reaction near its projection adds evidence that participants may be responding to that area. Do not force the line through every minor low or redraw it after each inconvenient move. A downward line is constructed from comparable falling swing highs. A touch or brief penetration is a test, while a break depends on a stated rule, such as a close below the line followed by failure to recover. Since line placement and chart scaling can change the apparent level, treat it as a zone of evidence alongside horizontal swings and broader trend structure. The line helps organize observations; it cannot supply an exact support price or guarantee the next reaction.
Worked example: Suppose equally spaced swing lows occur at 30 and 34. A straight projection adds four points by the next comparable swing, placing the line near 38. Price pulls back to 38 and rebounds, supplying a third reaction. Later it dips to 37 intraday but closes at 39. Under a closing-break rule, that session tested the line without confirming a break.
Watch the distinction: Calling any intraday trade below a trend line a confirmed reversal ignores the chosen observation rule. A close back above the line may leave the larger swing pattern intact.
Reference: [1]
11. Moving averages and lag
A simple moving average of length n adds the latest n observations and divides by n. Each observation has equal weight until it leaves the window. Replacing one old close with a new close changes the average by the difference between them divided by n, which explains why a long average responds slowly. Smoothing makes an underlying direction easier to inspect, but lag can delay a signal after a sharp turn. The average is a description of recent prices, not an estimate of intrinsic value. Specify the interval and whether the input is a close, typical price, or another series; a five-day average and a five-week average answer different questions. When using a price crossing or two-average crossover, define whether the signal requires an intraperiod touch or a completed close.
Worked example: Five hypothetical closes are 10, 11, 12, 13, and 19. Their sum is 65, so the five-period simple average is 65 divided by 5, or 13. The latest close is 19, six units above the average. If the next close is 20, the oldest 10 drops out and the average becomes 75 divided by 5, or 15.
Watch the distinction: An average below the latest price does not mean the asset is undervalued. The difference arises from smoothing past observations, especially after a large recent move.
Reference: [1]
12. Pattern boundaries and confirmation
A double top is a potential reversal structure with two comparable peaks separated by a reaction low. The intervening low forms the neckline. Until price breaks that boundary under a defined rule, the chart also fits an ongoing range or a pause in the preceding advance. Some analysts require a closing break to reduce sensitivity to intraday noise; consistency matters more than retroactively choosing the rule that fits the chart. A measured objective subtracts the peak-to-neckline height from the neckline. It estimates a possible move if the pattern works, not the probability of success or a guaranteed exit price. Watch for a return above the broken neckline, because that weakens the downside interpretation. Volume, trend context, and nearby support can further inform the trade plan without replacing the price condition that confirms the pattern.
Worked example: Price peaks near 70, falls to 62, and returns to about 70. At 64, it remains above the 62 neckline, so the downside pattern is unconfirmed. Suppose it later closes at 61. Under a closing-break rule, the break is confirmed. Height is 70 minus 62, or 8 points; subtracting 8 from 62 gives a conditional objective of 54.
Watch the distinction: Two similar highs alone do not confirm a double-top breakdown. Acting before the intervening low breaks treats a possible pattern as though its decisive price event had occurred.
Reference: [1]
13. Internal and external indicators
An internal indicator is calculated from the asset or index being studied, such as its moving average, momentum, or volatility. An external measure brings in information from a wider set of securities or markets. Market breadth, for example, counts how many constituent issues participate in a rise or fall. An index can advance because a few heavily weighted members surge even while most constituents decline. That makes breadth a useful check on participation, although weak breadth does not erase the index gain or predict an immediate reversal. First identify what each measure observes, then compare whether the signals agree. Also check construction: issue counts treat members equally, while a capital-weighted index does not. A difference between them can result from weighting rather than a calculation error.
Worked example: An index rises 2% and its own moving average turns upward. Of 100 constituents, only 35 advance, so the advance share is 35 divided by 100, or 35%. The average describes the index's price history; the constituent count shows narrow participation. The rally is real, but breadth gives a reason to qualify confidence in its breadth of support.
Watch the distinction: A rising index average does not imply that most stocks rose. A few large members can lift a weighted index while an issue-count measure remains weak.
Reference: [1]
14. Cycles, period, and phase
A cycle's period is the time between corresponding points, such as successive troughs. Its phase locates a series within that repeating pattern relative to a reference or another series. Two markets may share an estimated period yet peak at different times, making phase essential when comparing turning points. Estimate a period from several intervals rather than one apparent repetition, and inspect whether both period and phase remain reasonably stable. Financial cycles are noisy: trends, shocks, and changing behavior can shift or obscure peaks. A fitted rhythm is therefore timing context, not an independent command to trade. For a decision, specify a window in which a turn may occur and require observable price evidence, such as a break of a swing level. This keeps a proposed cycle from overriding what the market is currently doing.
Worked example: Suppose series A peaks at weeks 2, 10, and 18, while series B peaks at weeks 4, 12, and 20. Both have observed peak-to-peak intervals of eight weeks. B's peaks occur two weeks after A's, so their estimated phases differ by two weeks. A forecast that treats their peaks as simultaneous would miss that observed offset.
Watch the distinction: Equal estimated periods do not mean two cycles turn together. Ignoring phase can produce a mistimed comparison even when the interval between each series' peaks matches.
Reference: [1]
Level I - Advanced Techniques
15. Returns and dispersion
A simple holding-period price return equals ending price minus beginning price, divided by beginning price. State whether cash distributions and trading costs are included; a price-only return omits them. For a series of observations, an average summarizes the center, while dispersion describes how widely returns vary around it. Two assets can have the same average return but very different paths, including losses large enough to affect an investor's ability to stay invested. Standard deviation is one measure of dispersion: calculate each return's deviation from the mean, square those deviations, average them under the chosen sample or population convention, then take the square root. It treats upside and downside deviations alike, so it is not a complete account of loss risk. Compare return, variability, and drawdown in the same horizon and units.
Worked example: A price rises from 100 to 110. The price return is (110 minus 100) divided by 100, or 10%. Over two other hypothetical periods, asset A returns 8% and 12%, averaging 10%; asset B returns minus 20% and 40%, also averaging 10%. B's observations are much farther from their shared mean, so the equal averages conceal different variability.
Watch the distinction: Equal average returns do not establish equal risk. An average discards the spread and sequence of outcomes, both of which matter when losses or funding needs arise.
Reference: [1]
16. Volatility is not direction
Volatility describes the size or dispersion of price changes, not whether the next change will be positive or negative. A shrinking recent range or contracting volatility band indicates quieter observed movement under that measure's lookback and construction. A later expansion may accompany a break from consolidation, but contraction alone gives no direction. Define the range boundaries and a break rule before the move; otherwise it is easy to label whichever side breaks as the one the indicator supposedly predicted. Compare the band with actual price structure and consider false breaks that return inside the range. Volatility can also change quickly after news or a regime shift, so a quiet estimate should not be treated as a bound on future moves. The useful inference is about recent compression and a need to watch for expansion.
Worked example: A hypothetical market's recent high-to-low span shrinks from 12 points to 4 points, and its volatility band contracts. The span has fallen by 8 points, or about 66.7% of the former 12-point span. That supports a statement that recent movement is quieter. It does not favor an upward break over a downward one; the subsequent price break supplies direction.
Watch the distinction: A narrow band does not forecast a rally. It measures reduced dispersion in its lookback window, while the next expansion can occur in either direction.
Reference: [1]
17. Rules make signals testable
A systematic trading idea needs rules precise enough that independent analysts would identify the same trades. Define the instrument universe, observation interval, entry trigger, exit condition, position size, and treatment of costs. A prior-period high must exclude the current observation; otherwise the trigger can become ambiguous or use information unavailable when the decision was made. Test completed-bar signals using the prices actually available for execution, then subtract costs on each transaction. Record losses and drawdowns as well as average gains. A rule can be reproducible without being profitable, and an attractive backtest can result from selecting parameters after seeing the data. Use an untouched period to assess that risk. The CMT practice questions can reinforce terminology, while a written rule makes your own signal interpretation auditable.
Worked example: Replace “buy strong stocks” with this hypothetical rule: buy after a weekly close exceeds the highest close of the preceding 20 weeks, then exit after a weekly close falls below the lowest close of the preceding 10 weeks. Charge 0.2% of traded value at entry and again at exit. A round trip therefore faces 0.4% in stated transaction costs before any slippage.
Watch the distinction: A backtest of “strong stocks” cannot be reproduced until strength, timing, exits, and costs are specified. Vague rules leave room to choose favorable trades after seeing outcomes.
Reference: [1]
18. Cross-asset relative performance
A relative-strength line can be formed by dividing one asset's price by a benchmark price at each observation. If the ratio rises, the asset has outperformed that benchmark over the interval, even if both prices fell. If it falls, the asset has underperformed. Use comparable, consistently adjusted price series and the same dates; splits or mismatched observations can create misleading ratio jumps. The units of the two prices affect the numerical level of the ratio but not the interpretation of its change when units remain consistent. Absolute return and relative performance answer separate questions: the first describes the investor's gain or loss in the asset, while the second describes its result against an alternative. For allocation, inspect both because leadership during a falling market may still involve a monetary loss.
Worked example: A stock drops from 50 to 45, a return of (45 minus 50) divided by 50, or minus 10%. Its index falls from 100 to 80, or minus 20%. The stock-to-index ratio rises from 50 divided by 100 = 0.50 to 45 divided by 80 = 0.5625. The stock lost value but outperformed the index over the same interval.
Watch the distinction: A rising relative-strength line does not guarantee a positive asset return. It can rise simply because the benchmark declined faster than the asset.
Reference: [1]
19. Risk before position size
Position size follows from the loss you are prepared to take if a trade thesis is invalidated. For a long position, subtract the planned stop price from the entry price to obtain planned risk per share, then divide the monetary risk allowance by that distance. Round down to a permitted whole-share size. Choose the stop from the chart or trading rule first; moving it merely to fit a preferred share count defeats the risk plan. Check that the resulting purchase fits the available capital and any portfolio exposure limit. The arithmetic estimates loss at the stop price, but a gap, thin market, commissions, or slippage can produce a larger realized loss. Wider stops require smaller sizes at a fixed risk allowance, while tighter stops increase the chance of ordinary price noise reaching the exit.
Worked example: Assume a hypothetical loss allowance of 1,000 currency units, entry at 52 per share, and a protective stop at 47. Planned risk is 52 minus 47 = 5 units per share. Dividing 1,000 by 5 permits 200 whole shares before costs, requiring 10,400 units of purchase capital. If the stop moved to 42, risk would be 10 per share and size would fall to 100 shares.
Watch the distinction: Buying 200 shares first and then choosing a convenient stop reverses the decision sequence. The invalidation level and risk allowance determine size, subject to capital and execution limits.
Reference: [1]
Level I - Ethics
20. Material nonpublic information
Material information is information a reasonable investor would consider significant to an investment decision. Nonpublic information has not been broadly made available to the market. A technician who receives both through a confidential relationship cannot use it for a trade or pass it to someone else to trade. The restriction applies even when an independently observed chart also looks bullish: the attractive price pattern does not remove the information advantage. Identify what was learned, how it was obtained, whether it is material, and whether it is public before acting on the affected security. Keep the information confidential and follow the applicable professional procedures for handling it. After genuine public dissemination, other duties and restrictions may still matter, so a later trading decision requires its own assessment. The ethics issue concerns access and use of information, not the analyst's preferred technique.
Worked example: A client privately tells a technician the terms of an unannounced merger that could materially affect a listed company. The technician also sees a bullish breakout. The merger terms are material and nonpublic under the stated facts, so the technician must refrain from trading the affected security or passing the tip to another trader. The breakout remains a chart observation but supplies no exception to that restriction.
Watch the distinction: Claiming the trade was “based on the chart” does not cure possession of material nonpublic information. A separate public signal does not make confidential merger terms usable.
Reference: [1]
Level II - Theory and History
21. Behavioral bias in applied analysis
Availability bias makes an event seem more likely when it is easy to recall. A recent crash can dominate an analyst’s judgment even when earlier periods with similar charts produced many different outcomes. To test the judgment, define the comparison conditions before examining results: the preceding trend, volatility, breadth, valuation context if relevant, and the time horizon for the forecast. Then inspect every qualifying episode, including quiet recoveries and prolonged declines. Record how often each outcome occurred and how severe it was. The sample may still be too small or drawn from a different market regime, so historical frequencies are evidence rather than precise probabilities. In applied technical analysis, the useful question is whether a vivid event has displaced the analyst’s stated decision rule. The CMT Program Guide includes behavioral finance in Level II theory.
Worked example: A crash occurred last month. An analyst calls another crash virtually certain after seeing a similar breadth decline. In a hypothetical review of 20 earlier periods meeting the same predefined conditions, three ended in crashes, five in moderate declines, and 12 in recoveries. The crash frequency is 3 / 20 = 15% in that sample. The recent crash warrants attention, but the broader record does not support certainty.
Watch the distinction: A memorable outcome is not automatically a representative one. Selecting only past crashes with similar charts hides the recoveries needed to evaluate the analyst’s forecast.
Reference: [1]
Level II - Classical Techniques
22. Adjust charts before comparing trends
A corporate action can change a quoted price without creating a comparable change in a holder’s economic position. With a two-for-one split, each old share becomes two shares, while the per-share quote is expected to be roughly halved, apart from ordinary market movement. A raw historical chart may therefore show an artificial break in a trend line, moving average, or return calculation. Identify whether the series is raw, split adjusted, or adjusted for both splits and distributions before comparing observations. For a price trend across a split, put prices on a consistent share basis. For investment return, also account for cash distributions under the chosen return convention. Volume may require corresponding adjustment when comparing historical share counts. The Level II study guide can be used alongside chart exercises, but every calculation still needs a stated data convention.
Worked example: A stock closes at 100 immediately before a hypothetical two-for-one split and at 50 immediately after it, with no other price movement. A raw calculation gives (50 − 100) / 100 = −50%. On the new share basis, the earlier close is adjusted to 100 / 2 = 50. The comparable price change is (50 − 50) / 50 = 0%; the holder owns twice as many shares.
Watch the distinction: Treating the raw 50-point gap as selling pressure would falsely signal trend failure. Check the corporate action and adjustment convention before interpreting the gap.
Reference: [1]
23. Point-and-figure reversal logic
Point-and-figure charts filter price movement into boxes. An X column records advances, and an O column records declines; a new column starts only after the specified reversal distance is met. With a one-point box and a three-box reversal, an X column whose top is 40 needs a move of at least three points down before an O column begins. Smaller declines are ignored for column construction, although the market price has changed. Box size sets the price resolution, while reversal count sets how much movement is needed to change direction. Both must be fixed before reading a signal. A larger box or reversal count filters more noise but also delays recognition of a turn. Check the chart’s scaling and treatment of intraday highs and lows, since different construction conventions can produce different plotted columns.
Worked example: Assume one-point boxes, a three-box reversal, and an X column ending at 40. A decline to 38 spans 40 − 38 = 2 boxes, so the X column remains. A further decline to 37 spans 3 boxes and starts a new O column. Under this convention, the O boxes are marked at 39, 38, and 37; the 40 box remains in the X column.
Watch the distinction: A one-box decline after an X column does not automatically create an O column. Apply the stated reversal count to the column extreme.
Reference: [1]
24. Distinguish pullback from trend failure
An uptrend is a sequence of higher swing highs and higher swing lows on a stated observation interval. A decline from a high is initially a pullback; it challenges the trend when it violates a level chosen to define that trend. The analyst should identify the last meaningful swing low before seeing the next price bar, then specify whether an intraday breach or a close below it counts. Closing-price rules filter some temporary penetrations, although they respond later. A broken low establishes that the former higher-low sequence has failed under the chosen rule. It does not, by itself, establish a sustained downtrend, which would require further evidence such as lower highs and lower lows. Keep the time frame consistent: a daily break may coexist with an intact weekly trend.
Worked example: Suppose successive daily swing lows are 42, 48, and 53. Price reaches 65, pulls back to 55, and then closes at 52. The move to 55 stayed above the last low, so it remained a pullback under a close-below-53 rule. The 52 close is 1 point below 53 and breaks that rule’s trend support. Reassess the long; do not infer a complete downtrend from this one close.
Watch the distinction: Calling every decline from a peak a reversal ignores the swing structure. Equally, keeping an uptrend label after its predefined swing-low rule breaks makes the rule meaningless.
Reference: [1]
25. Head-and-shoulders trade levels
A head-and-shoulders top has a higher central peak between two lower shoulder peaks, with intervening troughs defining a neckline. The pattern is a proposed distribution structure until price breaks the neckline under a specified confirmation rule. Measure the vertical distance from the head to the neckline at the relevant point, then project that distance below the break area for a conditional objective. If the neckline slopes, use its value near the break rather than assuming it is horizontal. Set an invalidation condition as well: a recovery above the broken neckline weakens the bearish interpretation, while a move above the right shoulder may offer a wider but costlier risk boundary. Volume, trend context, and nearby support can affect confidence. The measured objective is a planning aid, not a probability statement or a guaranteed fill.
Worked example: Assume a horizontal neckline at 80, shoulders near 90, and a head at 100. A close at 79 meets the chosen below-neckline confirmation rule. Pattern height is 100 − 80 = 20 points, giving a measured objective of 80 − 20 = 60. If price quickly closes back above 80, the break needs reassessment even though the 60 objective remains unvisited.
Watch the distinction: Shorting merely because three peaks resemble the shape skips confirmation. The neckline break, invalidation level, and risk must be defined before acting.
Reference: [1]
26. Candles need location and follow-through
A bullish engulfing pattern occurs when a rising candle’s real body covers the preceding falling candle’s real body. It describes a shift in control during two sessions, not the full history of the trend. Its location matters: near a support zone after a decline, the pattern may indicate that selling met stronger demand. In the middle of a sideways range, the same candle offers less information. Compare the bodies using opens and closes; the shadows need not be engulfed under the usual body-based definition. Then look for follow-through, such as a subsequent close above the engulfing candle’s high, and define what would invalidate the interpretation. A close below the pattern’s low would weaken the rebound case. Wider market conditions and volume can support or challenge the reading, but no single candle establishes a durable reversal.
Worked example: After a decline toward support at 30, Monday opens at 32 and closes at 30. Tuesday opens at 29 and closes at 33, so its body from 29 to 33 contains Monday’s body from 30 to 32. The pattern is bullish in this location. A later close above Tuesday’s high would add confirmation; a close below 29 would undermine the setup.
Watch the distinction: An engulfing body is not confirmation that a downtrend has ended. Its location, subsequent price action, and invalidation level determine its practical value.
Reference: [1]
27. Momentum divergence
Negative momentum divergence appears when price makes a higher high while an oscillator makes a lower high at corresponding swing points. The oscillator measures the pace or character of recent price movement, so divergence says the new price high has less measured momentum behind it. It does not say that buying has stopped or that price must fall. Strong trends can show several divergences while continuing upward. Match comparable highs, keep the oscillator settings unchanged, and note whether the peaks are separated by a meaningful swing rather than adjacent noise. Next identify a price-based trigger, such as a close below the intervening swing low, and a level that would negate a short thesis. Divergence can justify closer monitoring, a tighter risk plan, or reduced conviction in a new long without forcing an immediate reversal trade.
Worked example: Price reaches 100 with an oscillator reading of 72, pulls back, then reaches 106 while the oscillator peaks at 64. Price’s second high is 6 points higher, while the oscillator’s is 8 units lower. That is negative divergence. If the intervening price low is 96, an analyst requiring a close below 96 has a warning at 106 but no short trigger yet.
Watch the distinction: Selling solely because an oscillator makes a lower high treats a warning as a trade signal. Price may keep rising while momentum moderates.
Reference: [1]
28. Breadth and the Traders Index
The Traders Index, or TRIN, compares the advance-to-decline issue ratio with the advance-to-decline volume ratio: (advancing issues / declining issues) ÷ (advancing volume / declining volume). A value below 1 means advancing volume is proportionally stronger than the count of advancing issues; above 1 means declining volume has the stronger relative weight. One way to see the mechanism is to compare average volume per advancing issue with average volume per declining issue. TRIN is their inverse ratio, so it captures whether volume is concentrated on the advancing or declining side. Define the exchange universe and observation period consistently, and avoid treating one reading as a complete market forecast. Extremely small denominators can make the measure unstable. Breadth measures participation; they qualify an index move without canceling the observed price change.
Worked example: In a hypothetical session, 600 issues advance on 900 million shares and 400 decline on 300 million shares. The issue ratio is 600 / 400 = 1.5; the volume ratio is 900 / 300 = 3. TRIN is 1.5 / 3 = 0.5. Average volume is 1.5 million shares per advancer versus 0.75 million per decliner, consistent with stronger advancing-side volume.
Watch the distinction: A TRIN reading of 0.5 does not mean half the stocks advanced. It compares issue breadth with volume breadth, using the same advancing and declining universe.
Reference: [1]
29. Cycle claims and unstable phase
A cycle claim needs more than a few visually similar turning points. Period is the elapsed time between comparable points, such as successive troughs; phase describes where a cycle sits relative to a reference or another series. To assess usefulness, mark troughs using the same rule, calculate each interval, and inspect both the average and the spread. A fitted average can hide a large shift in timing. Also check whether peaks and troughs remain aligned across separate parts of the sample, since a changing phase can make a seemingly stable period poor for trade timing. Market cycles may be influenced by changing participants and conditions, so an observed rhythm is uncertain. Use a cycle window to focus attention, then require an independent price condition, such as a defined trend-line break or swing reversal, before taking a position.
Worked example: Suppose trough intervals are 18, 19, and 31 weeks. Their mean is (18 + 19 + 31) / 3 = 22.7 weeks, approximately, but the last interval exceeds 19 by 12 weeks. Forecasting the next trough exactly 19 weeks after the latest one ignores that instability. A reasonable plan marks a broad observation window and waits for a separate price trigger.
Watch the distinction: A repeating pattern found by eye can be a fitting artifact. A precise forecast needs stable timing evidence, not just a memorable pair of past troughs.
Reference: [1]
Level II - Advanced Techniques
30. Correlation and changing relationships
Correlation summarizes how two return series move together within a particular sample. It ranges from −1 to +1 when defined in the usual Pearson form, but its estimate depends on the return interval, sample dates, and market regime. A low historical correlation can improve diversification because the positions have not usually moved together. It cannot guarantee protection during stress, when common funding needs, sector exposures, or broad risk aversion may make both fall. Examine rolling estimates, adverse periods, and the economic drivers behind the relationship. Also distinguish correlation from loss magnitude: two assets may have modest correlation yet each carry large volatility. Portfolio risk depends on both individual variances and their covariance, so position sizes matter alongside the coefficient. Correlation describes co-movement in observed data; it does not identify the cause.
Worked example: Two assets have a historical return correlation of 0.2 over a hypothetical three-year sample. During one stressed month, Asset A falls 8% and Asset B falls 6%. The joint decline does not by itself calculate a new correlation, but it shows that the earlier estimate did not prevent simultaneous losses. Review rolling and stress-period relationships before assuming the pair will diversify a new portfolio.
Watch the distinction: A correlation of 0.2 is not a promise that losses will offset. It is a sample estimate, and common exposures can become more important during stress.
Reference: [1]
31. Assess volatility across horizons
Volatility measures dispersion of returns over a stated interval. Under an assumption of independent returns with stable variance, variances add across periods, so standard deviation scales with the square root of time. If daily volatility is σ, the corresponding estimate over N trading days is σ√N. This is a standard-deviation estimate, not a guaranteed price range or a directional forecast. Check that the input is daily rather than annualized, and use trading days if the estimate is based on trading-day returns. The scaling rule can mislead when volatility clusters, returns are correlated, jumps occur, or the regime changes. For an options-implied measure, its underlying horizon and annualization convention also matter. Translate volatility into a planning range only after stating the distributional model and the starting price.
Worked example: Assume hypothetical daily return volatility of 1% and 25 trading days of independent returns with stable variance. The estimated 25-day volatility is 1% × √25 = 5%. At a starting price of 100, 5% corresponds to 5 price units as a rough one-standard-deviation scale under the model. It does not imply that price must stay between 95 and 105.
Watch the distinction: Multiplying daily volatility by 25 gives 25%, which confuses standard deviation with variance accumulation. Use √25 under the stated assumptions.
Reference: [1]
32. Trigger, filter, and value rules
A systematic rule set assigns distinct jobs to different conditions. A trigger identifies the event that would initiate a trade, such as a close above resistance. A filter decides whether the environment qualifies, such as a rising longer-term trend or sufficient liquidity. A value or sizing rule decides how much exposure is acceptable given price, stop distance, and risk allowance. Keep these roles separate so a signal can be audited: record whether the trigger fired, whether the filter passed, and what size the risk rule allowed. Define the order of evaluation and the data available at the decision time to avoid using future information. A filter can improve trade selection but can also remove good trades; its effect should be tested after costs. Sizing controls planned loss under normal execution, while gaps and slippage can exceed it.
Worked example: A hypothetical system enters long after a daily close above 120, provided the weekly trend is rising. It permits 800 of planned loss. Price closes at 121, but the weekly trend is falling, so the trigger fires and the filter rejects the trade; the sizing rule is never used. If the filter later passes with entry 121 and stop 117, size would be 800 / 4 = 200 shares before costs.
Watch the distinction: Calling the 121 close a complete buy signal overlooks the separate weekly filter. A reproducible system records each rule’s result before placing an order.
Reference: [1]
33. Out-of-sample evidence
Optimization chooses parameters that performed well on a development sample. The more combinations tested, the easier it is to find a rule that fits noise. Out-of-sample evaluation reserves data that did not influence the rule, parameter choices, or discretionary edits, then measures performance after realistic costs. The test is strongest when observations are ordered in time and the later period remains untouched until the strategy is fixed. Compare returns with drawdowns, trade count, turnover, and performance across nearby parameter values. A single favorable holdout result is still uncertain; repeated redesign after seeing it effectively turns the holdout into training data. Walk-forward tests can repeat the sequence of fitting on past data and evaluating on later data. The aim is a rule whose behavior is understandable and reasonably stable outside the conditions used to build it.
Worked example: A strategy is tuned on an earlier sample and earns a hypothetical 18% annualized there after estimated costs. Its rules are frozen before a later untouched test, which earns −3% annualized after costs with comparable exposure. The difference is 21 percentage points. That gap does not prove the rule can never work, but it weakens the implementation case and calls for scrutiny of fitting and market conditions.
Watch the distinction: Changing the parameters after seeing the holdout loss and testing the same period again no longer supplies an independent check. Preserve a fresh evaluation period.
Reference: [1]
34. Trade excursions and stop design
Maximum adverse excursion, or MAE, is the largest unrealized loss reached during a trade relative to entry. Maximum favorable excursion, or MFE, is the largest unrealized gain. For a long position, subtract the lowest price reached from entry for MAE, and subtract entry from the highest price for MFE. These measures reveal the path hidden by the final profit or loss. Compare MAE across winning and losing trades to examine whether a proposed stop removes many eventual winners, while checking whether that stop limits damaging losers enough to justify the trade-off. MFE helps assess how much favorable movement a trailing exit captures or gives back. Use executable price assumptions and account for gaps; an excursion calculated from bars does not guarantee a stop fill. A single trade illustrates the calculation but cannot establish an optimal stop.
Worked example: A hypothetical long enters at 50, reaches a low of 46, later reaches 60, and exits at 57. MAE is 50 − 46 = 4 per share; MFE is 60 − 50 = 10 per share; realized gain is 57 − 50 = 7. A stop at 47 would have been crossed on the way to 46, so this trade would have exited before the later gain.
Watch the distinction: A profitable final exit does not mean the trade never faced a loss. Stop analysis requires the path and execution assumptions, not just entry and exit prices.
Reference: [1]
35. Risk-adjusted performance
Risk-adjusted ratios answer different questions because their denominators differ. Sharpe divides return above a risk-free rate by total return volatility. Sortino divides return above a stated target by downside deviation measured against that target. The Information ratio divides active return over a benchmark by tracking error, the volatility of active returns. Calmar compares annualized return with maximum drawdown, the largest observed peak-to-trough loss. Match periods and units before comparing ratios, and state the target or benchmark used. A high Sharpe can coexist with a damaging drawdown; a high Calmar may reflect a short sample that missed a severe episode. Information ratio is most useful when benchmark-relative performance matters. None describes liquidity, tail risk, or whether results survive transaction costs on its own.
Worked example: Assume hypothetical annual return of 12%, risk-free return of 2%, total volatility of 10%, downside deviation about a 2% target of 5%, and maximum drawdown of 20%. Sharpe is (12% − 2%) / 10% = 1. Sortino is (12% − 2%) / 5% = 2. Calmar is 12% / 20% = 0.6. An Information ratio cannot be calculated without benchmark return and tracking error.
Watch the distinction: Using total volatility as the Information ratio denominator confuses it with Sharpe. Information ratio needs active return and tracking error against a specified benchmark.
Reference: [1]
36. Intermarket confirmation
Intermarket analysis asks whether movements in related markets support a trade thesis. A strengthening currency can make a commodity quoted in that currency more expensive for buyers using other currencies, creating possible downward pressure on its price. A producer’s shares may also weaken if expected revenue falls. This is a plausible transmission channel, not a fixed rule: global demand, supply disruptions, hedging, production costs, and company-specific news can dominate. State the currency pair and quotation convention before interpreting a move, since saying that a currency rose can be ambiguous. Compare timing as well: a stock move preceding the commodity change may reflect separate information. Use cross-asset evidence to increase or reduce conviction, then define a price level or event that would change the position. Several markets moving together provide context without proving one caused another.
Worked example: Suppose a currency strengthens 4% against a trade-weighted basket, a commodity quoted in that currency falls 6%, and a producer’s shares fall 9% over the same month. The moves fit a currency-pressure thesis. They do not explain the full 9% stock decline: the analyst should also check production costs, hedges, and company news before attributing the share move to the currency.
Watch the distinction: Three aligned charts do not establish a single cause. The relationship can reverse when supply, demand, or company factors outweigh the currency channel.
Reference: [1]
37. Risk budget across correlated trades
A position risk budget starts with the planned loss between entry and invalidation, multiplied by position size. When trades share a sector, factor, or funding source, a common shock may trigger several exits together. Add their planned losses under a simultaneous-stop scenario and compare that total with the portfolio allowance. This is a scenario calculation, not a forecast that all stops will fill precisely. Gaps can make realized loss larger, while offsets may reduce it. Examine overlapping exposures through holdings, return correlations, and the economic reason the trades might fail together. A portfolio of several securities can still be a single concentrated bet. If combined exposure is too large, reduce sizes, remove a trade, or choose a genuinely different risk source. Revisit the calculation as prices and stop levels change.
Worked example: Three hypothetical bank-stock trades each risk 500 from entry to stop. Individually each fits a 600 per-trade limit. If a banking shock triggers all three, planned loss is 500 + 500 + 500 = 1,500. Against a 1,000 portfolio loss allowance, the set exceeds the budget by 500. Equal reductions to about 333 of planned risk each would bring the combined amount to roughly 999 before gaps.
Watch the distinction: A per-trade limit does not cap portfolio loss when positions are correlated. Aggregate the plausible simultaneous losses before treating each order as acceptable.
Reference: [1]
Level II - Application of Technical Analysis
38. Confirm an entry with competing signals
A chart entry becomes more defensible when evidence plays distinct roles. A break above resistance supplies the trigger. A rising higher-time-frame trend gives directional context. Breadth or sector participation shows whether the move is supported beyond one security. These observations may conflict, so decide in advance whether weak participation blocks the entry, reduces size, or merely calls for closer monitoring. Then identify the price that invalidates the breakout thesis and calculate the risk per share from the intended entry. Check whether the stop lies beyond ordinary noise and whether the resulting size fits the loss allowance. A close beyond resistance records a price event; it does not ensure follow-through. An exit rule should also cover a failed break back into the prior range, even if the initial stop has not been reached.
Worked example: A stock closes at 76 above resistance at 74, remains above a rising weekly average, and its sector shows broad advances. A hypothetical plan enters at 76 and exits on a close below 72, giving planned risk of 76 − 72 = 4 per share. With an 800 loss allowance, size is 800 / 4 = 200 shares before costs. Narrowing breadth would lower conviction or size under a stated filter.
Watch the distinction: Three bullish observations do not replace an invalidation rule. Without a stop condition and size, the analyst cannot say how much the trade risks.
Reference: [1]
39. Turn volatility into a trade constraint
A volatility-based stop tries to allow room for ordinary movement while defining when the trade no longer fits the risk plan. First state the volatility measure and how it produces the stop distance, such as a chosen multiple of average true range. Next compare that stop with chart structure: a mathematically convenient level inside a support zone may be hit without invalidating the setup. Calculate risk per share as entry minus stop for a long, then divide the monetary loss allowance by that amount and round down to a tradable size. Include likely commissions and slippage when the allowance is tight. Volatility changes, so a stop derived from an old estimate may no longer match current movement. Planned loss is conditional on execution at the stop; gaps can exceed it. Position size should follow the chosen risk distance, not dictate a stop that lacks chart meaning.
Worked example: Assume entry at 100, a hypothetical volatility-based stop at 94, and a permitted loss of 1,200. Planned risk is 100 − 94 = 6 per share, so the maximum whole-share size before costs is 1,200 / 6 = 200. At 300 shares, planned loss would be 300 × 6 = 1,800, exceeding the allowance by 600. Reduce size or reject the trade.
Watch the distinction: Buying 300 shares and moving the stop closer solely to fit the budget may place it inside ordinary noise. Validate the stop first, then size.
Reference: [1]
Level II - Ethics
40. Suitability and client constraints
A technically attractive setup must still fit the client’s stated objectives, risk tolerance, and investment constraints. A breakout describes market action; it does not amend a mandate. Read the proposed instrument as well as the chart: leverage changes the size and speed of losses, while liquidity, concentration, and holding period can make a trade unsuitable even without leverage. Compare the product’s possible adverse outcomes with the client’s need for capital preservation and any explicit prohibition. If the instrument fails a hard constraint, do not recommend it to that client. An unleveraged instrument is a separate candidate requiring its own analysis of trend, downside, and portfolio exposure. Document the relevant facts and communicate the recommendation’s basis and risks clearly. The Level II CMT curriculum includes both applied risk management and ethics.
Worked example: A leveraged exchange-traded product breaks above resistance, but a client mandate prohibits leverage and prioritizes capital preservation. Even if the chart signal is valid, that product fails the mandate, so the technician does not recommend it. A conventional unleveraged fund tracking the same market could be assessed separately for volatility, concentration, entry risk, and fit; it is not automatically suitable merely because it lacks leverage.
Watch the distinction: Chart strength cannot override a client’s explicit restriction. Replacing the leveraged product with another instrument also requires a fresh suitability assessment.
Reference: [1]
Level III - Application of Technical Analysis
41. Choose risk for the investor
Start with the payoff distribution, then ask how the investor values each outcome. A risk-neutral investor compares expected monetary payoffs, calculated by multiplying each payoff by its probability and adding the results. A risk-averse investor places extra value on avoiding a damaging low outcome; a risk-seeking investor may prefer the uncertain payoff even when its expected value is lower. This is a question about utility, the value the investor assigns to wealth, rather than a universal ranking of trades. Existing wealth, liquidity needs, and the size of a possible loss can change that ranking. For a position recommendation, compare the proposed payoff with a feasible safer alternative, check whether the downside fits the investor's risk capacity, and size the exposure accordingly. The CMT Program Guide includes assessing different investor attitudes toward a gamble in Level III.
Worked example: A hypothetical choice offers 120 monetary units or zero, each with probability 50%, against a certain 55 units. Its expected payoff is 0.5 × 120 + 0.5 × 0 = 60 units, five more than the certain amount. A risk-neutral investor chooses the gamble on that measure. An investor who values the certainty of 55 more than the chance of 120 can rationally choose the certain payoff.
Watch the distinction: Choosing the gamble for every investor because 60 exceeds 55 ignores the zero-payoff outcome. Expected value does not measure how painful that outcome is to a particular investor.
Reference: [1]
42. Estimate tail loss and its severity
Value at risk, or VaR, marks a loss boundary for a specified probability and holding period. Conditional value at risk, or CVaR, describes the average loss in the worse tail beyond that boundary. State the loss convention because discrete outcomes can put observations exactly on a quantile. An empirical calculation sorts losses from smallest to largest, identifies the chosen cutoff, and averages the designated worst outcomes for CVaR. Both measures depend on the sample or model, the assumed holding period, and how positions behave under stress. VaR says where a tail begins under the stated distribution; it does not reveal how severe losses become after crossing that point. CVaR supplies that additional severity information, though neither measure captures an event absent from the assumed distribution. Compare the results with liquidity and leverage before relying on them for a portfolio decision.
Worked example: Assume 100 equally likely one-day portfolio losses, measured in thousands of dollars. The smallest 95 losses are no greater than 4, at least one equals 4, and the worst five are 6, 7, 8, 9, and 10. Using the 95th observation as the 95% VaR boundary gives $4,000. Averaging the worst five gives CVaR of (6 + 7 + 8 + 9 + 10) / 5 = $8,000.
Watch the distinction: A 95% VaR of $4,000 does not cap losses at $4,000. It identifies a boundary under the assumed distribution; the tail can contain much larger losses.
Reference: [1]
43. Diversify the source of risk
Diversification depends on how exposures respond to the same shock, not simply on how many securities appear in a portfolio. Holdings in one industry may share sensitivity to funding costs, regulation, or demand. Their prices can therefore fall together even when each company has a separate ticker. To assess a proposed addition, group holdings by likely risk drivers, inspect their weights, and compare historical co-movement across ordinary and stressed periods. Then ask whether the new holding changes the portfolio's exposure to those drivers. A low historical correlation is useful evidence, but it is sample-dependent and may rise during stress. Diversification can reduce security-specific risk; it cannot remove broad market risk or guarantee a small loss. Position sizing should reflect concentrated exposures as well as each holding's individual stop or volatility estimate.
Worked example: Suppose four bank stocks and one utility each hold 20% of a hypothetical portfolio. Banks collectively account for 4 × 20% = 80%. If a banking shock cuts all four bank positions by 10% while the utility is unchanged, the portfolio loses 0.80 × 10% = 8%, before other effects. Splitting the bank allocation among five banks would increase the ticker count without removing that shared exposure.
Watch the distinction: Five holdings are not automatically well diversified. If most portfolio value responds to the same industry shock, losses can remain strongly linked despite the number of securities.
Reference: [1]
44. Build a pattern trade with two stop rules
A pattern's measured objective estimates a possible move from its dimensions; it is a planning reference, not an expected fill. Define the breakout price, entry, objective, and invalidation point before comparing stop rules. A percentage stop sets a fixed fraction below entry for a long trade. An average true range, or ATR, stop uses a multiple of recent range, so it adapts to observed price movement. For each choice, calculate risk per share as entry minus stop and potential reward as objective minus entry. Divide reward by risk, then calculate share size from the monetary risk allowance. The higher reward-to-risk ratio is not automatically the better setup: a tight stop can sit within normal price noise, while a wider stop can leave too little room under the portfolio budget. Check both against nearby chart structure and likely execution costs.
Worked example: A hypothetical pattern has a measured objective of 115 and a breakout entry of 103; ATR is 2 price units. Potential reward is 115 − 103 = 12 per share. A 3% stop is 103 × 0.97 = 99.91, risking 3.09 and yielding 12 / 3.09 ≈ 3.88 to one. A two-ATR stop is 103 − 4 = 99, risking 4 and yielding 3 to one.
Watch the distinction: Selecting the 3% stop solely because its ratio is higher can be misleading. A stop inside routine price movement may trigger before the pattern has meaningfully failed.
Reference: [1]
45. Plan for a failed breakout
A breakout thesis rests on price accepting levels beyond an established boundary. Before entry, specify what would count as acceptance and what return into the range would invalidate it. A close-back-inside rule is one possible test; it avoids changing the rule after an uncomfortable reversal. Volume can strengthen the warning if substantial trading accompanies the return, but volume alone does not define failure. Once the condition occurs, the long trade's original reason has weakened, so follow its exit plan and reassess from a flat position. A failed upside break can attract sellers, yet a short trade needs its own trigger, stop, size, and objective. It should not arise automatically from the disappointment of the long. A quick recovery above the boundary would also challenge a bearish interpretation. The Level III study guide provides a related path for integrated trade decisions.
Worked example: Assume a range from 48 to 52, with a rule to buy a close above 52 and exit on a close back inside. Price closes at 54, then at 51 on heavy volume. Since 48 < 51 < 52, the exit condition is met: close the long under the stated rule. A short remains conditional; for example, it could require a further break below 48 and a defined stop.
Watch the distinction: Holding the long because 54 was once a valid breakout ignores the later invalidation event. A past signal cannot override the exit condition established for that trade.
Reference: [1]
46. Combine candles with wider structure
A hammer has a small real body near the top of its range and a long lower shadow. It shows that sellers pushed price down during the session but could not hold it there by the close. That one session does not establish a lasting reversal. First locate the candle within the wider trend: a hammer after a decline at prior support has a clearer potential role than one in the middle of a sideways range. Next examine the following close or a break above the hammer's high for evidence that buyers continue to act. Define invalidation below the hammer's low or another justified structural level, allowing for the instrument's normal volatility. A nearby resistance level may limit the room for a rebound. This sequence joins candle information with trend, support, follow-through, and a risk plan.
Worked example: A hypothetical stock declines from 95 toward weekly support near 80. It then forms a hammer with a low of 79, a high of 82, and a close near 81. The next session closes at 84, above the hammer's 82 high. That follow-through supports a rebound thesis. A close below 79 would instead challenge the proposed support and require reassessment.
Watch the distinction: Calling every hammer a buy signal ignores its location and subsequent price action. A similar shape inside a noisy range may carry little directional information.
Reference: [1]
47. Point-and-figure objectives in context
Point-and-figure charts filter price movement into columns of Xs and Os using a chosen box size and reversal requirement. A horizontal count converts the width of a congestion pattern into a conditional price objective under a stated counting convention. Because conventions can vary, identify the method, box size, reversal amount, breakout level, and count before computing. The objective estimates the scale of a possible move, while prior supply marks a nearer price level where selling could appear. A trade plan can take partial profits, tighten a stop, or wait for evidence of acceptance at that obstacle. The count does not establish when the objective will be reached, and a changed box size can produce a different count. Keep the entry and invalidation tied to the chart's actual breakout and reversal behavior rather than treating the computed level as a guarantee.
Worked example: Suppose a hypothetical horizontal method counts four columns across a base, uses one price unit per box and a three-box reversal, and projects 4 × 1 × 3 = 12 units above a breakout at 56. Its objective is 56 + 12 = 68. Prior supply lies near 65, only 9 units above entry. Plan a response at 65 before assuming the remaining 3-unit move to 68.
Watch the distinction: Treating 68 as a guaranteed exit price confuses a count-based objective with an order fill. Resistance at 65 or a failed breakout can interrupt the move.
Reference: [1]
48. Rank markets by relative strength
Relative strength compares an asset's price with a benchmark by dividing one by the other at each date. A rising ratio means the asset is outperforming over that interval, even if both prices are falling. An absolute uptrend answers a different question: whether the asset itself is gaining value. For new long capital, examine both. A candidate with a rising absolute trend and a rising relative ratio has stronger evidence than one that merely loses less than a falling market. Use comparable adjusted price series, consistent dates, and a benchmark relevant to the allocation decision. Ratio changes can also reflect weakness in the denominator, so inspect both underlying charts. Ranking identifies preference among candidates; it does not by itself specify an entry, stop, position size, or acceptable valuation.
Worked example: At the first date, assets A and B and their shared benchmark all trade at 100, so each relative ratio is 1. Later the benchmark is 105, A is 110, and B is 108. A's ratio is 110 / 105 ≈ 1.048; B's is 108 / 105 ≈ 1.029. Both have absolute gains, 10% and 8%, but A shows the stronger relative gain.
Watch the distinction: A rising relative ratio does not always mean the asset price rose. The ratio can improve when the asset falls less than its benchmark, leaving an absolute loss.
Reference: [1]
49. Read breadth beneath an index high
An index can reach a new high while fewer constituent stocks participate because larger or more influential members carry the advance. Breadth measures participation through observations such as advancing versus declining issues, new highs, or the share of members above a trend measure. Compare breadth at successive index peaks using the same universe and method. If price makes a higher high while participation contracts, the breakout is still a price fact, but its support across the market has narrowed. That conflict can justify smaller size, a closer review of the invalidation level, or waiting for broader confirmation. Breadth divergence is a warning rather than a timing rule: narrow leadership can persist, and a later expansion in participation can resolve the concern. Avoid treating a single day's count as decisive without considering the surrounding trend.
Worked example: At an earlier index high, 700 issues advance and 300 decline, giving an advance share of 700 / 1,000 = 70%. At a new index high, 430 advance and 570 decline, so the share is 43%. The index has broken higher, but participation is 27 percentage points lower. A long thesis remains possible, with conviction or position size reduced until breadth improves.
Watch the distinction: Declaring the index breakout false solely because breadth weakened overstates the indicator. Narrow participation raises risk, but the observed new price high and its own exit levels still matter.
Reference: [1]
50. Separate trend from momentum loss
Trend describes the direction of successive price swings; momentum describes the pace of change over a chosen interval. A rate-of-change measure compares today's price with the price a fixed number of periods earlier. Its decline can mean prices are still rising, only more slowly. Compare the momentum horizon with the trend horizon and locate the latest swing low that defines the trend thesis. If that low remains intact, a trend-following rule need not force an exit. Slowing momentum may nevertheless call for smaller exposure when the potential reward no longer justifies the same risk. A separate exit condition, such as a close below a prior weekly swing low, gives a clear decision point. Avoid moving that condition simply to defend a prior forecast. Momentum can recover; a trend can also fail before an oscillator reaches any particular reading.
Worked example: Suppose weekly swing lows rise from 72 to 78 to 84. Twelve-week rate of change falls from 18% at one reading to 5% at the next, so the measured advance has slowed by 13 percentage points. If the latest weekly close is 90, above the 84 swing low, the stated trend rule remains intact. The manager may reduce size while retaining 84 as the defined invalidation level.
Watch the distinction: A fall in rate of change from 18% to 5% is not a 13% price decline. Both readings are positive; the later gain over its comparison period is smaller.
Reference: [1]
51. Connect cycle evidence to a price trigger
A cycle estimate proposes a window when turning points may be more likely under a fitted rhythm. Calculate intervals between comparable troughs, examine how much they vary, and project a range rather than a precise date. Phase matters: two series may have similar periods but reach troughs at different times. Cycles can shift or disappear, so the calendar window should frame observation, not create an automatic order. Choose an independent price trigger, such as a close above a prior swing high, and define an invalidation level below the prospective trough. If price keeps making lower lows during the window, the downtrend still controls the chart. A later trigger can be assessed on its own merits, including volume, volatility, and the distance to resistance. Repeatedly redrawing the cycle after each miss reduces its usefulness as decision evidence.
Worked example: Hypothetical troughs occur near weeks 20, 39, and 60. The observed intervals are 19 and 21 weeks, averaging (19 + 21) / 2 = 20 weeks, so week 80 is a tentative next window. At week 80 price makes a new low; no reversal trigger has occurred. If a later close exceeds a defined swing high at 50, that price event can support entry with risk measured below the new trough.
Watch the distinction: Buying simply because week 80 arrived treats an estimated period as a confirmed turn. Continued lower lows show that the required price evidence is still absent.
Reference: [1]
52. Judge a model fit for implementation
A model can describe its fitting data well while offering little value in new conditions. Assess implementation in stages. First identify which observations and parameters were used to build it, then examine performance on data kept out of fitting. Deduct realistic trading costs and inspect the path of returns, especially maximum drawdown and time spent recovering. Next vary parameters slightly: a useful rule should not depend on one exact setting to remain viable. Compare results across relevant market regimes and check whether the required turnover and liquidity fit the intended account. An attractive average return cannot compensate for a drawdown that violates the mandate. Neither out-of-sample success nor parameter stability guarantees future performance; together they make a stronger case than a single optimized result. CMT practice questions can help rehearse these distinctions before writing a model recommendation.
Worked example: Assume all returns are hypothetical annualized figures after stated costs. Model A earns 14% in fitting data, 1% on untouched later data, and suffers a 30% maximum drawdown. Model B earns 10% and 8%, with a 15% drawdown and similar results at nearby parameter settings. If both fit the mandate, B has better evidence for implementation: its later result holds up more closely and its drawdown is half A's.
Watch the distinction: Selecting A because its fitted return is 14% rewards optimization on data the model has already seen. The weak later result and larger drawdown are directly relevant.
Reference: [1]
53. Allocate a fixed risk budget
A fixed loss budget can turn a ranking of setups into explicit position limits. Allocate percentages of the total budget according to the strength and independence of the opportunities, ensuring the percentages sum to 100%. For each trade, divide its monetary allowance by the entry-to-stop distance per share, then round down to a tradable whole-share size. This creates comparable planned losses despite different prices and stop widths. Check the total cash needed, possible gaps, transaction costs, and correlated exposures before placing orders. Three trades can each fit an individual allowance yet still suffer together from one sector shock. The budget controls planned loss at stated stop prices; it cannot guarantee those fills. If the combined exposure is too concentrated, reduce the allocations or leave part of the risk budget unused rather than forcing every ranked idea into the portfolio.
Worked example: A hypothetical total planned-loss budget is $1,200. Rankings assign 50%, 30%, and 20%, producing allowances of $600, $360, and $240. If stop distances are $3, $4, and $2 per share, sizes are $600 / $3 = 200 shares, $360 / $4 = 90 shares, and $240 / $2 = 120 shares. Planned losses sum to $1,200 before costs and gaps.
Watch the distinction: Allocating 50%, 30%, and 20% of capital would be a different calculation. These percentages divide permitted loss; position values depend on each entry price and stop distance.
Reference: [1]
54. Target portfolio volatility
Volatility targeting changes exposure in response to an estimate of portfolio variability. Under simple proportional scaling, divide the target volatility by estimated current volatility to obtain an exposure multiplier. Apply that multiplier to the positions if their composition stays the same and the risk estimate scales approximately with exposure. A multiplier below one calls for less exposure; above one suggests more, subject to leverage and mandate limits. Decide what to reduce before trading: selling every position proportionally preserves weights, while cutting the most volatile holding changes the portfolio's composition and correlations. Re-estimate volatility after large market moves, because yesterday's estimate may understate a new regime. Volatility describes dispersion, not the direction of returns or the size of every possible loss. Also check liquidity and trading costs so frequent adjustments do not consume the intended benefit.
Worked example: Suppose a portfolio has a hypothetical annualized volatility estimate of 18% and a 12% target. The simple multiplier is 12% / 18% = 0.667. If holdings and their relationships remain stable, keeping about two-thirds of current exposure gives an estimated 18% × 0.667 ≈ 12% volatility. The remaining third can be held outside those risky exposures; changed correlations would require a fresh estimate.
Watch the distinction: A 12% volatility target does not mean the portfolio cannot lose more than 12% in a year. Volatility is a dispersion estimate, not a loss ceiling.
Reference: [1]
55. Account for leverage in downside plans
Leverage magnifies an asset move relative to the investor's equity. Begin with asset exposure divided by starting equity to find the leverage multiple. For a small move before financing and costs, multiply the asset's percentage return by that multiple to approximate the equity percentage return. The same multiplication applies to losses, making a chart stop more consequential in equity terms. In a downside plan, convert the stop distance into a monetary loss on the full exposure, then compare that loss with the account's risk allowance. Consider whether a gap could bypass the stop and whether a margin requirement could force liquidation earlier. Borrowing costs and changing equity alter the result over time. Leverage can make an otherwise sensible technical setup unsuitable if the position size turns an ordinary underlying move into an unacceptable account loss.
Worked example: A hypothetical account holds $200,000 of assets financed with $100,000 of equity and $100,000 borrowed, so starting leverage is 2 times. A 5% asset decline loses $200,000 × 0.05 = $10,000. Ignoring interest and costs, equity falls to $90,000, a $10,000 / $100,000 = 10% decline. A planned 5% chart stop therefore exposes 10% of starting equity before execution uncertainty.
Watch the distinction: Calling the loss 5% of equity uses the underlying asset percentage as the account percentage. The account owns twice its equity in assets, so the effect is doubled.
Reference: [1]
56. Adapt stops as a trade develops
An initial protective stop limits the planned loss if a new trade fails quickly. A trailing stop changes as the trade develops, typically moving in the favorable direction by a stated distance or volatility multiple. For a long position, calculate the trailing level from the highest qualifying price observation and do not lower it merely because price pulls back. Specify whether the rule uses intraday highs, closing prices, or another reference; mixing them changes the result. A trailing stop can protect part of a gain while leaving room for normal fluctuations, but a narrow distance may exit a sound trend prematurely. Assess the stop against price structure and the account's risk budget. A stop order may fill beyond its level during a gap, so the calculated locked-in gain is conditional on execution at the stop price.
Worked example: A hypothetical long enters at $40 with an initial stop at $36, a planned $4 loss per share. After its highest close reaches $50, a rule placing the stop $4 below that close moves it to $46. If filled at $46, the trade earns $46 − $40 = $6 per share before costs. A gap below $46 could yield less, and a later lower close does not lower the stop.
Watch the distinction: Moving a long trade's trailing stop downward after a pullback defeats the rule's protective purpose. Under this stated rule, only a new higher qualifying close can raise it.
Reference: [1]
57. Weigh a volatility forecast against price structure
A volatility forecast can test whether a chart objective and intended holding period belong together. Translate the forecast into price units for the same horizon as the trade, then compare the objective's distance with that estimated movement. State the model: a one-standard-deviation move is a measure of dispersion under assumptions about the return distribution, not a maximum move or a promised range. A target many estimated standard deviations away may still be possible, but reaching it within the stated period requires a stronger move than the model regards as typical. Check whether the objective came from a longer-term chart pattern, whether the forecast horizon is too short, and whether volatility has recently changed. Entry and stop decisions still need price structure and acceptable loss. Do not enlarge the volatility estimate casually just to make a preferred objective appear plausible.
Worked example: Assume a hypothetical stock trades at $100 and a chosen one-week model estimates a one-standard-deviation move of about $4. A $115 objective is $15 away, or $15 / $4 = 3.75 such estimated moves. The objective may suit a longer holding period, but it is ambitious for one week under this model. Review the horizon and assumptions before making $115 the basis of a near-term reward calculation.
Watch the distinction: Treating the $4 estimate as a hard trading band would wrongly rule out $115. It describes modeled variability; unusual moves can occur, especially when the assumptions fail.
Reference: [1]
58. Respond to conflicting intermarket evidence
Intermarket evidence helps test whether a price move is supported by related conditions. An equity breakout gives a direct signal for the asset being traded. Widening credit spreads may indicate investors require more compensation for credit risk, while defensive-sector leadership can indicate caution within equities. These observations challenge confidence without mechanically canceling the breakout. Begin by identifying which relationship is relevant to the trade and its horizon; correlations can change, and a sector's move may have its own cause. Then set a conditional response: smaller initial size, a specific invalidation level, and observable conditions for adding or reducing exposure. Breadth improvement and stabilizing credit behavior could strengthen the thesis; a price return below breakout support could end it. Keep the cross-asset checks separate from the actual order trigger so conflicting evidence produces a disciplined decision.
Worked example: Suppose an index closes at 104 above resistance at 102, while credit spreads widen and a defensive sector outperforms. The price trigger supports a long, but the two contextual signals argue for caution. Under a hypothetical plan, enter half the intended size at 104, add only after broader participation and stable spreads appear, and exit if a later close falls back below 102. None of those conditions predicts the outcome.
Watch the distinction: Calling the index breakout invalid solely because credit spreads widen gives one relationship too much authority. The conflict affects conviction and sizing; a defined price failure remains a separate event.
Reference: [1]
59. Write a complete position decision
A defensible position recommendation connects evidence to an executable decision. State the prevailing trend and whether the asset is outperforming a relevant benchmark. Identify the price event that triggers entry, then set an invalidation level where the underlying thesis would no longer hold. Estimate the distance from entry to stop, divide the monetary loss allowance by that distance, and check the resulting position's cash requirement, volatility, and correlated portfolio exposure. A measured objective supplies a conditional upside reference; compare its distance with planned risk and with likely obstacles along the route. Specify when to reduce, exit, or review the trade as new evidence arrives. The stop level describes planned risk rather than guaranteed execution, and the objective is a scenario rather than a forecast certainty. This sequence is especially useful for a Level III written response because each conclusion follows from stated evidence.
Worked example: A hypothetical sector leads its benchmark and closes at $76 above $75 resistance. Its pattern objective is $87; a close below $72 invalidates the trade. Risk is $76 − $72 = $4 per share, so a $1,000 loss allowance permits $1,000 / $4 = 250 shares before costs. Potential reward is $87 − $76 = $11, giving $11 / $4 = 2.75 to one. Review resistance before $87 and exit on the stated failure.
Watch the distinction: Writing only “buy at 76, target 87” omits the reason the trade fails and the amount at risk. A target cannot substitute for invalidation and sizing.
Reference: [1]
Level III - Ethics
60. Disclose conflicts in recommendations
A personal holding can create an incentive to present favorable technical evidence selectively. Before communicating a recommendation, identify financial interests that a reader would reasonably consider relevant, including ownership of the security being discussed. Disclose the interest clearly so the audience can judge the analysis with that context. Then apply the same analytical standard that would be used without the holding: describe the pattern, confirmation, contrary evidence, risk level, and uncertainty fairly. Disclosure does not turn weak research into sound research, and a sound chart does not erase the conflict. If the analyst's judgment is compromised, the recommendation should not be presented as an independent assessment. Keep the decision about personal trading separate from the communication process and comply with applicable conduct requirements. Level III includes ethics alongside integrated technical analysis.
Worked example: A technician already owns 500 shares when the stock closes above a resistance line and is preparing a bullish research note. The note should disclose that relevant holding, explain the breakout and its invalidation level, and address contrary evidence such as weak breadth. Readers can then evaluate the recommendation knowing the technician may benefit if price rises. The breakout itself does not remove the conflict.
Watch the distinction: A valid chart signal does not make an undisclosed personal holding irrelevant. Readers need the conflict information to assess possible bias in the analyst's selection and presentation of evidence.
Reference: [1]