CMT Level III asks you to integrate market evidence into defensible decisions about entries, exits, allocation, models, and risk. This guide contains twenty original practice items in four linked five-item groups, predominantly written responses, plus sixty substantive lessons. Every item repeats the facts needed to answer it independently. Coverage follows the supplied 2026 guide applicable as of October 16, 2026: nineteen questions and fifty-seven concepts address Application of Technical Analysis; one question and three concepts address Ethics. These are preview allocations, not official exam counts. Source s3 supports syllabus relevance; technical explanations apply established subject knowledge. Earlier guides and future revisions are separate versions.
What this guide covers
Twenty original CMT Level III practice items and 60 lessons on technical decisions, models, risk and ethics.
Application of Technical Analysis
1. Expected utility versus expected profit
Expected monetary value weights each outcome by its probability, whereas expected utility weights the investor's satisfaction with the resulting wealth. A risk-neutral investor evaluates a fair gamble by its monetary expectation. A risk-averse investor with concave utility can prefer a certain amount despite an equal expected payoff. A risk-seeking investor may prefer dispersion over certainty. Evaluate terminal wealth, not just isolated trade profits, because the same dollar loss can have different consequences for different accounts. Utility analysis explains preferences; it does not establish the accuracy of estimated probabilities.
Worked example: An investor with USD 100,000 can accept a certain USD 2,000 gain or a gamble offering equal chances of gaining USD 12,000 and losing USD 8,000. Both have USD 2,000 expected profit, but concave utility favors the certain gain.
Common mistake: Equal expected monetary profits do not make two choices equally attractive when investors differ in risk preferences or financial circumstances.
Official syllabus reference: [1]
2. Wealth growth and changing risk capacity
Risk appetite is willingness to accept uncertainty; risk capacity is the ability to absorb losses without undermining objectives. Portfolio growth can reduce the additional utility of another dollar, making preservation more important after goals are funded. However, declining appetite with wealth is an economic interpretation, not an invariant empirical law for every investor. Translate objectives into permissible dollar losses, drawdowns, and liquidity needs rather than inferring a mandate from account size alone. Reassess exposure when gains materially change the relationship between available wealth and required future spending.
Worked example: A trader grows an account from USD 200,000 to USD 500,000, enough to fund a planned purchase. Keeping the former aggressive exposure may threaten the now-funded objective, so the trader reduces the permitted drawdown rather than automatically increasing leverage.
Common mistake: A larger portfolio does not automatically justify larger percentage risks; funded objectives and loss capacity can change the appropriate exposure.
Official syllabus reference: [1]
3. Square-root-of-time volatility conversion
When returns are uncorrelated across equal periods and variance is stable, variance grows in proportion to time and standard deviation grows with its square root. Annualized volatility can therefore be converted to daily volatility by dividing by the square root of the assumed trading-day count. State that count explicitly. Serial correlation, changing volatility, jumps, and overlapping observations can invalidate simple scaling. This conversion describes dispersion under a model; it neither predicts direction nor guarantees that returns remain inside a particular range. Use consistent return definitions and time units.
Worked example: Assuming 252 trading days and stable independent daily returns, annualized volatility of 24% implies daily volatility of 24% divided by √252, approximately 1.51%. A ten-trading-day estimate is 1.51% times √10, approximately 4.78%, under the same assumptions.
Common mistake: Multiplying standard deviation directly by the number of days confuses variance scaling with volatility scaling and overstates modeled dispersion.
Official syllabus reference: [1]
4. Value at Risk as a loss quantile
Value at Risk identifies a loss quantile for a specified horizon, confidence level, exposure, and estimation method. A one-day 95% VaR of USD 10,000 places that amount at the model's 95th loss percentile. It does not describe the average loss beyond that point or the largest possible loss. Historical, parametric, and simulation methods can produce different answers from the same portfolio. State sign conventions and discrete-sample quantile rules, and inspect whether the data represent current positions and market conditions before interpreting the estimate.
Worked example: A risk report gives one-day 95% VaR of USD 10,000 for a current portfolio. The analyst interprets USD 10,000 as the estimated loss quantile, then separately examines scenarios producing USD 25,000 losses because VaR does not describe their severity.
Common mistake: A 95% VaR estimate does not mean the portfolio cannot lose more than the reported amount on any particular day.
Official syllabus reference: [1]
5. CVaR and expected tail severity
Conditional Value at Risk, often called expected shortfall under specified conventions, summarizes losses in the adverse tail beyond a chosen probability boundary. For a continuous distribution, it can be understood as the average loss within that tail. Discrete samples require an explicit rule for boundary observations and fractional probability weights. Unlike VaR, this measure responds directly to the magnitude of extreme losses. It remains sensitive to limited tail data and modeling assumptions, so combine it with scenario analysis rather than treating a historical tail average as a worst-case bound.
Worked example: Under a rule averaging the worst five of 100 equally weighted losses, tail losses of USD 2,000, USD 3,000, USD 4,000, USD 6,000, and USD 10,000 produce CVaR of USD 5,000. A much lower VaR would conceal this tail severity.
Common mistake: CVaR is an average of specified adverse outcomes, not an extra loss added to VaR or a maximum-loss guarantee.
Official syllabus reference: [1]
6. Leverage separates exposure from collateral
Leverage increases market exposure relative to equity. For a simple linear exposure, multiplying position size multiplies both gains and losses before financing, fees, and nonlinear effects. Margin is collateral supporting an obligation; it does not normally cap the loss on a futures position. Track notional exposure, economic sensitivity, collateral requirements, and liquidity separately. A strategy can satisfy initial margin while remaining vulnerable to variation-margin demands or gaps. Leverage should therefore be evaluated against stressed price moves and available funding, not only recent volatility or posted collateral.
Worked example: An account with USD 100,000 equity controls USD 300,000 of linear market exposure. A 4% adverse underlying move produces approximately USD 12,000 loss, or 12% of equity, before costs, even if the required initial collateral is much smaller.
Common mistake: Low initial margin does not make an exposure low risk; the underlying contract sensitivity determines how price movements affect equity.
Official syllabus reference: [1]
7. Risk-budget position sizing
Risk-budget sizing divides a permitted dollar loss by the estimated loss per unit. For shares, unit loss includes entry-to-stop distance and an appropriate allowance for complete-trade costs. For derivatives, apply the contract multiplier and any relevant nonlinear sensitivity. Round down to permitted trading increments, then check concentration, leverage, and liquidity limits independently. The calculated amount is planned exposure under stated execution assumptions. Stops can slip or gap, so a stop-based calculation should not be described as a guaranteed maximum loss or a substitute for stress testing.
Worked example: A USD 150,000 account permits 0.8% planned loss, giving USD 1,200. Entry is USD 40, the stop is USD 37, and costs are USD 0.20 per share. Dividing USD 1,200 by USD 3.20 permits 375 shares before other limits.
Common mistake: Sizing from the entry price alone ignores the stop distance and can assign identical quantities to trades with very different planned losses.
Official syllabus reference: [1]
8. Volatility targeting through exposure scaling
Volatility targeting adjusts exposure to keep estimated portfolio volatility near a chosen mandate level. For an unchanged risky sleeve and zero-risk cash, the exposure multiplier equals target volatility divided by estimated sleeve volatility. Apply leverage and liquidity caps after calculating the multiplier. Estimated risk can lag abrupt market changes, while frequent adjustments can raise costs and produce procyclical selling. The target is a model-based exposure objective rather than a guaranteed realized outcome. Use consistent annualization, reliable inputs, and an explicit policy for forecast changes and rebalancing.
Worked example: A risky sleeve has estimated annualized volatility of 16%, and its mandate targets 10%. Holding 62.5% of portfolio value in that sleeve and 37.5% in assumed zero-volatility cash gives modeled volatility of 10%, before estimation error and trading frictions.
Common mistake: A volatility target does not guarantee a drawdown ceiling, because jumps, estimation error, and return sequences can produce unexpectedly large losses.
Official syllabus reference: [1]
9. Covariance determines portfolio volatility
For two assets, portfolio variance equals w₁²σ₁² + w₂²σ₂² + 2w₁w₂σ₁σ₂ρ. The covariance term captures how the holdings move together. Portfolio volatility is the square root of that variance, not a weighted average of standalone volatilities except in special cases such as perfect positive correlation with positive weights. Match return frequencies and estimation windows across inputs. Correlation does not reveal every form of dependence, particularly joint tail behavior, so complement covariance calculations with explicit adverse scenarios and changing-correlation assumptions.
Worked example: Two equally weighted assets each have 20% annualized volatility and zero correlation. Portfolio variance is 0.5² × 0.20² twice, or 0.02. Its square root is 14.14%, demonstrating diversification despite neither asset having volatility below 20%.
Common mistake: Averaging constituent volatilities omits covariance and can materially misstate portfolio risk, even when capital weights are correctly specified.
Official syllabus reference: [1]
10. Diversification across independent risk sources
Diversification reduces dependence on individual outcomes by combining exposures with imperfectly aligned risks. Asset diversification spreads holdings across securities or classes; strategy diversification combines distinct return mechanisms; temporal diversification varies entry timing or horizons. These categories can overlap without providing independent protection. Several assets may all depend on the same equity, duration, liquidity, or currency factor. Examine economic drivers and return behavior together. Diversification can reduce concentration and some variability, but it does not eliminate marketwide losses or guarantee that historical relationships persist during stress.
Worked example: A portfolio holds five technology funds and appears diversified by ticker count. Holdings analysis reveals substantial overlap, while return analysis shows similar responses to growth-stock selloffs. Adding a genuinely different strategy may reduce dependence more effectively than adding a sixth similar fund.
Common mistake: Counting securities is not enough to measure diversification when their underlying holdings, factor exposures, or liquidity risks substantially overlap.
Official syllabus reference: [1]
11. Correlation instability and tail dependence
Correlation is an estimate conditional on the observations, horizon, and market environment used. Relationships can strengthen or reverse when funding pressure, common ownership, or macroeconomic shocks dominate ordinary trading. A low full-sample correlation can also conceal strong dependence during large losses. Compare rolling estimates, stressed periods, and economically plausible joint shocks rather than relying on one unconditional number. Selecting historically uncorrelated securities is useful only when combined with an assessment of shared drivers, estimation uncertainty, and the portfolio's ability to survive simultaneous adverse moves.
Worked example: Two strategies show correlation of 0.10 over five years, but both sell liquid equities when volatility rises. During a funding shock, both lose together. A stressed correlation of 0.80 produces a more relevant exposure check than their full-period average.
Common mistake: A low historical correlation is not a permanent property of a security pair or evidence that simultaneous large losses cannot occur.
Official syllabus reference: [1]
12. Inverse volatility versus equal risk contribution
Inverse-volatility weighting assigns capital in proportion to one divided by each asset's estimated volatility. Equal risk contribution instead allocates according to each holding's contribution to total portfolio volatility, incorporating the full covariance matrix. The two methods coincide under particular covariance structures, but they are not generally interchangeable in multi-asset portfolios. A correlated cluster can dominate risk even when its members receive small inverse-volatility weights. Inspect marginal and component risk contributions, concentration, and leverage after constructing weights, and remember that all depend on estimated relationships.
Worked example: Assets with volatilities of 10%, 20%, and 20% receive inverse-volatility weights of 50%, 25%, and 25%. If the two volatile assets are highly correlated and the first is diversifying, those capital weights need not equalize portfolio risk contributions.
Common mistake: Equalizing weighted standalone volatility does not generally equalize total portfolio risk contributions when correlations differ across a multi-asset portfolio.
Official syllabus reference: [1]
13. Drawdown and the asymmetric recovery hurdle
Drawdown measures the decline from an equity peak to a subsequent lower value, typically expressed as a percentage of the peak. Maximum drawdown is the deepest such decline over the observation period. Recovering from a fractional loss d requires a gain of d/(1−d) on the reduced capital. This asymmetry makes large losses disproportionately difficult to repair. Maximum drawdown depends on sample length and return order, so compare it alongside duration, recovery time, and stress scenarios rather than treating a historical maximum as a future limit.
Worked example: An account falls from USD 100,000 to USD 75,000, a 25% drawdown. Returning to USD 100,000 requires USD 25,000 profit on USD 75,000, or 33.33%. A subsequent 25% gain would leave the account at only USD 93,750.
Common mistake: A percentage gain equal to the preceding percentage loss does not restore starting wealth because the gain applies to a smaller base.
Official syllabus reference: [1]
14. Liquidity, slippage, and executable risk
Liquidity concerns the ability to trade a desired quantity promptly without excessive price impact. Quoted spreads describe only part of implementation cost; available depth, order size, volatility, and urgency also affect fills. A stop can trigger during a gap or thin market and execute beyond its nominal level. Estimate risk using realistic order and fill assumptions, including participation limits where appropriate. Liquid-looking historical bars may conceal intrabar execution difficulty. Distinguish a model's reference price from an executable price when assessing trade viability and portfolio capacity.
Worked example: A model buys at a displayed USD 30 offer, but only 500 shares are available there while the intended order is 10,000 shares. Filling the remainder at higher prices increases average entry cost and may invalidate the backtested reward-to-risk estimate.
Common mistake: A visible quote or daily closing price does not establish that the entire intended position could have traded at that price.
Official syllabus reference: [1]
15. Anchoring and the disposition effect
Anchoring gives excessive influence to a reference such as purchase price, even when current evidence has changed. The disposition effect describes a tendency to realize gains readily while retaining losses, often because realizing a loss feels like admitting failure. These biases can distort exits and capital allocation. Translate the investment thesis into observable invalidation conditions and compare the expected use of capital across current opportunities. A precommitted rule can reduce discretion, but it should itself have a defensible basis rather than mechanically preserving an arbitrary entry-price anchor.
Worked example: A trader bought at USD 70 and refuses to exit at USD 63 despite a failed breakout, saying the position will be sold when it returns to cost. The original price has replaced current trend and opportunity evidence as the decision reference.
Common mistake: Waiting to break even is not an analytical justification when the purchase price has no continuing relevance to the market thesis.
Official syllabus reference: [1]
16. Swing structure and trend invalidation
An uptrend is commonly characterized by rising swing highs and rising swing lows; a downtrend shows the reverse. A break of a meaningful prior low can weaken an uptrend without immediately proving a durable downtrend. Specify the chart horizon and swing-identification method because minor fluctuations and major reversals have different implications. Trendlines and moving averages summarize related information but do not replace the underlying sequence. Use structural changes to update scenarios and risk plans, while distinguishing an early warning from the additional evidence required by a reversal strategy.
Worked example: Weekly lows rise from USD 42 to USD 46 to USD 49, while highs also rise. A weekly close below USD 49 disrupts the latest higher low. The analyst reduces bullish conviction but waits for further structure before labeling a sustained downtrend.
Common mistake: One broken minor low does not automatically establish a major downtrend when the relevant horizon and broader swing structure remain different.
Official syllabus reference: [1]
17. Support-resistance role reversal
Support and resistance identify areas where buying or selling previously affected price movement. After a decisive breakout, former resistance may become support as traders reassess value and positioning. The reverse can occur after a breakdown. Treat these areas as zones rather than perfectly precise prices, and assess acceptance through closes, retests, volume, or other predefined evidence. A retest that fails to hold weakens the role-reversal hypothesis. Historical reactions support a working framework, but they do not create an obligation for future market participants to transact there.
Worked example: A share repeatedly stalls near USD 80 before closing at USD 83. It later retreats to USD 80.50 and rebounds on stronger volume. The analyst treats the old resistance zone as potential support while defining what renewed trading below it would invalidate.
Common mistake: A previously important price is not guaranteed to reverse its role; subsequent acceptance and rejection behavior determine the usefulness of that hypothesis.
Official syllabus reference: [1]
18. Classical pattern measurement objectives
Classical chart objectives translate a pattern's measured dimensions into a conditional projection. For a rectangle, a common convention adds the range height to the breakout boundary for an upside objective. Other formations use their own stated measurement points. Identify the boundary, height, breakout direction, and arithmetic explicitly before evaluating the trade. An objective is a planning reference, not an estimated fair value or guaranteed destination. Intervening resistance, execution costs, stop distance, and the probability of failure still determine whether the setup is attractive.
Worked example: A rectangle spans USD 32 to USD 38. Its USD 6 height projects an upside objective of USD 44 from the USD 38 boundary. Buying at USD 41 leaves only USD 3 of projected upside, despite the unchanged pattern measurement.
Common mistake: A valid pattern objective does not make every entry attractive, because a late entry can consume much of the prospective reward.
Official syllabus reference: [1]
19. Percentage breakout filters
A percentage filter requires price to move a specified proportion beyond a reference boundary before recognizing a signal. Under a stated 3% upside rule, multiply resistance by 1.03; for a 3% downside rule, multiply support by 0.97. Specify whether intraday trading or a completed close must satisfy the condition. The filter seeks to reduce marginal boundary crossings but delays entry and can worsen reward-to-risk. A fixed percentage also represents different numbers of recent volatility units across securities, so its suitability depends on the instrument and strategy.
Worked example: Resistance is USD 60, and a strategy requires a closing breakout at least 3% above it. The threshold is USD 61.80. An intraday high of USD 62 followed by a USD 61 close does not satisfy this closing-price rule.
Common mistake: A 3% filter is a specified trading convention, not a universal confirmation threshold or a statement that all false breakouts are excluded.
Official syllabus reference: [1]
20. ATR filters and volatility-adjusted distances
True range is the greatest of high minus low, absolute high minus previous close, and absolute low minus previous close. ATR smooths true range over a specified lookback and method. An ATR multiple expresses a buffer relative to recent price movement, allowing distances to adapt across instruments and volatility regimes. State whether the ATR is fixed at entry or updated later. Increasing ATR can widen a proposed stop and require smaller position size. ATR measures movement magnitude rather than trend direction, and its historical value cannot bound future gaps.
Worked example: A breakout boundary is USD 100 and entry requires a close 0.5 ATR above it. With ATR of USD 4, the threshold is USD 102. If a later setup has ATR of USD 6, the same rule requires USD 103.
Common mistake: ATR does not indicate whether the next move will be upward or downward; it scales distance using recent price variability.
Official syllabus reference: [1]
21. Reward-to-risk and required win probability
Reward-to-risk divides the prospective net gain by the prospective net loss under specified exits. With a binary payoff and no other outcomes, the break-even win probability is loss divided by gain plus loss. Costs should be incorporated consistently into both outcomes. A high payoff ratio can coexist with negative expectancy if successful exits are sufficiently rare. Real strategies often have partial exits, gaps, and varying outcomes, so use the ratio as a trade-planning measure and compare it with empirical outcome distributions rather than assuming an objective will be reached.
Worked example: A trade offers USD 6 net gain against USD 3 net loss, a 2:1 ratio. Its binary break-even win probability is 3/(6+3), or 33.33%. A credible 25% win probability would still imply negative expected value despite the attractive-looking ratio.
Common mistake: A large reward-to-risk ratio does not establish positive expectancy unless the probability and distribution of realized outcomes also support the trade.
Official syllabus reference: [1]
22. Failed patterns and trapped-position logic
A failed pattern occurs when price behavior contradicts the formation's anticipated resolution, such as a breakout returning into its prior range. Traders who entered on the original signal may then exit, potentially reinforcing movement in the opposite direction. This trapped-position explanation is a hypothesis about market behavior, not proof of imminent reversal. Separate the rule for abandoning the original trade from the rule for initiating a contrary position. Reentry and reversal require independently defined evidence, execution conditions, and risk limits to prevent an analytical warning becoming an impulsive trade.
Worked example: A stock breaks above USD 50, then closes back at USD 48.80 inside its former range. The long strategy exits under its failure rule. A short strategy still waits for a failed retest of USD 50 rather than reversing solely because the long lost money.
Common mistake: Invalidating a long setup does not automatically validate a short setup; the opposite trade may require different evidence and risk parameters.
Official syllabus reference: [1]
23. Candlestick geometry needs market location
Candlesticks show the relationship among open, high, low, and close within one period. Real-body size, shadow length, and closing location summarize the session's price rejection and acceptance. A long lower shadow after a decline near support can support a recovery hypothesis, while the same shape inside congestion may offer little directional information. Evaluate preceding trend, chart location, liquidity, and subsequent confirmation before assigning significance. A candle compresses intraperiod trading into four values, so it does not uniquely reveal the order of every transaction or participant motivation.
Worked example: A share opens at USD 25, falls to USD 23, and closes at USD 25.10 near its high after testing established support. The rejection is more informative than an identical candle appearing midway through a broad, directionless trading range.
Common mistake: Recognizing a candle shape without examining its preceding trend and location confuses visual classification with a justified trading interpretation.
Official syllabus reference: [1]
24. Independent confirmation of candlestick signals
Confirmation adds evidence that a candlestick interpretation is consistent with broader market behavior. Examples include a subsequent close above a reversal candle's high, alignment with established support, or participation visible in volume. Prefer evidence that addresses a different dimension of the hypothesis rather than several indicators derived from nearly identical price transformations. Confirmation can reduce premature entries but usually sacrifices some price advantage. Define the confirmation condition in advance and reassess the resulting stop distance and payoff ratio, since a stronger signal can still become an unattractive trade at a later entry.
Worked example: A bullish rejection forms at weekly support, and the next session closes above its high on expanding volume. The analyst gains price, location, and participation evidence, then recalculates risk because the confirmed entry is higher than the initial candle's close.
Common mistake: Adding several closely related oscillators does not necessarily provide independent confirmation, and waiting for confirmation can materially change trade economics.
Official syllabus reference: [1]
25. Candlestick invalidation and stop placement
A candlestick-based stop should connect to the signal's underlying claim, such as rejection of prices near a meaningful low. A renewed move through that low can contradict the rejection thesis, but the exact execution rule may include a volatility buffer or closing condition. Distinguish analytical invalidation from a protective order designed for sudden adverse movement. Calculate position size after choosing the defensible stop distance. A tighter stop improves the quoted payoff ratio while increasing sensitivity to ordinary noise, and a wider stop requires less size for the same planned dollar loss.
Worked example: A bullish rejection candle bottoms at USD 41. A strategy places its protective stop at USD 40.60 using an explicit buffer, rather than exactly at the visible low. With entry at USD 43, sizing uses the USD 2.40 distance plus costs.
Common mistake: Placing a stop as close as possible does not automatically improve a strategy; it can increase premature exits despite improving the displayed ratio.
Official syllabus reference: [1]
26. Point-and-figure box and reversal settings
Point-and-figure charts emphasize price movement by recording rising columns of Xs and falling columns of Os. Box size determines the price increment, while the reversal setting specifies how many boxes of opposite movement create a new column. Time is irregular because a quiet interval may add no marks. Different box sizes, reversal counts, scaling methods, and price inputs can materially change the chart. State these construction choices before interpreting breakouts or counts. A cleaner-looking formation does not eliminate the need to evaluate execution risk and broader market conditions.
Worked example: With USD 1 boxes and a three-box reversal, a rising column reaching USD 30 requires a sufficient decline to USD 27 to begin a falling column under the stated convention. A move only to USD 29 does not create that reversal.
Common mistake: Equal horizontal spacing between point-and-figure columns does not mean equal elapsed time, because new columns depend on price movement.
Official syllabus reference: [1]
27. Point-and-figure horizontal count objectives
A horizontal point-and-figure count uses the width of a selected congestion area to estimate a price objective under a stated convention. A common arithmetic framework multiplies counted columns by box size and reversal amount, then projects that distance from a specified reference. Count boundaries and reference points vary by method, so make them explicit rather than presenting the result as uniquely determined. Objectives depend strongly on chart construction and selected congestion. Treat the count as a conditional scenario to integrate with trend, nearby barriers, and risk-based trade design.
Worked example: Under a specified method, eight congestion columns, USD 0.50 boxes, and a three-box reversal produce a USD 12 count distance. Projecting from a USD 40 reference gives USD 52. The analyst records the chosen count boundaries so another analyst can reproduce it.
Common mistake: A point-and-figure objective is not independent of box size, reversal convention, or the analyst's specified count boundaries and projection reference.
Official syllabus reference: [1]
28. Volume confirmation and price acceptance
Volume measures trading activity and can help assess participation in a price move. Expanding volume on a breakout may strengthen the case that more participants are engaging at new prices, whereas weak activity can warrant caution. Interpretation depends on the instrument, trading venue, seasonality, and the comparability of the volume series. High volume alone does not identify whether informed buyers or sellers will prevail. Combine activity with price progress, closing location, and follow-through; substantial volume with little progress can indicate intense opposing supply and demand rather than straightforward confirmation.
Worked example: A share breaks above USD 90 on twice normal volume but closes back below USD 90. The activity shows participation, yet the failed close weakens acceptance above resistance. The analyst avoids treating high volume as sufficient bullish confirmation by itself.
Common mistake: High volume does not independently establish direction; its meaning depends on what price accomplishes during and after the activity.
Official syllabus reference: [1]
29. Multiple-timeframe signal reconciliation
Multiple-timeframe analysis separates the horizon defining the investment thesis from the horizon used for entry and execution. A weekly uptrend can coexist with a daily pullback, while an intraday rally can occur within a daily downtrend. Decide which horizon controls eligibility, invalidation, and holding expectations. Alignment can strengthen a continuation setup; conflict may identify a countertrend trade requiring different assumptions. Avoid switching the controlling horizon after entry merely to excuse adverse movement. Any horizon change should represent a new, explicit decision with revised risk and opportunity analysis.
Worked example: A weekly uptrend remains intact while daily momentum falls during a pullback. A trader waits for a daily recovery signal to enter in the weekly direction, using daily structure for execution while retaining a weekly holding thesis and defined risk.
Common mistake: A short-term reversal does not automatically reverse the longer-term trend, and changing horizons after a loss can conceal a broken thesis.
Official syllabus reference: [1]
30. Moving-average lag and regime sensitivity
Moving averages smooth price to summarize trend, but they necessarily respond after information enters their calculation. A simple average gives equal weight within its window; an exponential average emphasizes recent observations under a specified smoothing factor. Shorter settings respond faster and usually generate more whipsaws, while longer settings respond slowly and may surrender more of a reversal. Crossovers and slope filters should therefore be assessed against the intended horizon and trading costs. No setting removes the fundamental trade-off between responsiveness and noise across every market regime.
Worked example: A ten-day average turns upward during a brief rally, while a fifty-day average continues downward. A fast crossover strategy may enter earlier than a slow trend filter, but repeated sideways fluctuations can create additional entries, exits, and accumulated costs.
Common mistake: An exponential moving average still lags price; placing more weight on recent observations does not make it a leading indicator.
Official syllabus reference: [1]
31. Relative-strength ratios and benchmark choice
A relative-strength ratio divides one asset's price or total-return index by a benchmark's corresponding series. A rising ratio indicates outperformance, which can occur during absolute gains or losses. Use matched currencies, timestamps, and dividend treatment to avoid attributing data differences to genuine leadership. Benchmark selection matters: outperforming a weak sector does not necessarily imply outperforming the broader market. Analyze ratio trends alongside absolute price trends when the mandate requires both leadership and capital appreciation. Relative strength is distinct from the bounded Relative Strength Index oscillator.
Worked example: A fund falls from 100 to 96 while its benchmark falls from 100 to 90. Their normalized ratio rises from 1.00 to 1.0667, indicating 6.67% relative-ratio improvement despite the fund's 4% absolute loss over the period.
Common mistake: A rising relative-strength line does not establish an absolute profit and should not be confused with an RSI reading.
Official syllabus reference: [1]
32. Rate of change and momentum ranking
Rate of change expresses the percentage movement from a chosen earlier price: current price divided by prior price, minus one. Cross-sectional momentum ranks assets by comparable historical returns, while time-series momentum evaluates an asset against its own past or a stated threshold. Lookback choice influences responsiveness and can change rankings materially. Ensure comparable total-return treatment and handle missing histories consistently. Momentum leadership does not directly measure volatility or portfolio diversification, so use separate risk controls when translating ranks into capital weights and avoid confusing a strong rank with assured continuation.
Worked example: Asset A rises from USD 50 to USD 56, giving 12% rate of change, while B rises from USD 80 to USD 88, giving 10%. A ranks higher over that lookback even though B's dollar price increase is larger.
Common mistake: Ranking by dollar price changes instead of comparable percentage returns can favor high-priced securities without identifying stronger proportional performance.
Official syllabus reference: [1]
33. RSI divergence as a warning, not an exit rule
The Relative Strength Index compares smoothed gains with smoothed losses over a specified lookback and maps their relationship onto a bounded scale. A bearish divergence occurs when price reaches a higher high while RSI forms a lower high, suggesting reduced upside momentum. It does not prove that price must reverse, and strong trends can maintain elevated readings. Interpret divergence alongside structure, trend, and a predefined trigger. A trading rule should specify what price event converts the warning into action rather than assuming the oscillator alone supplies an executable reversal signal.
Worked example: A share makes highs at USD 70 and USD 75 while RSI peaks at 76 and 68. The analyst notes weakening momentum but retains the trend position until the strategy's price-based exit condition occurs, rather than shorting immediately on divergence.
Common mistake: An overbought RSI reading or bearish divergence does not by itself establish a downtrend or determine the timing of a reversal.
Official syllabus reference: [1]
34. MACD level, signal line, and histogram
MACD measures the difference between a faster and slower exponential moving average. Its signal line smooths that difference, and the histogram measures MACD minus its signal line. A positive MACD places the faster average above the slower one; a shrinking positive histogram indicates that MACD's advantage over its signal line is narrowing. This can warn of deceleration without proving a negative trend. Specify settings and distinguish level, crossover, and divergence interpretations. Because all components derive from price, agreement with other price smoothers may add limited independent evidence.
Worked example: MACD remains positive at 1.20 while its signal line rises to 1.10, shrinking the histogram to 0.10. The analyst describes weakening incremental momentum, rather than claiming the faster average has already fallen below the slower average or that price must decline.
Common mistake: A declining positive histogram does not necessarily mean MACD is negative or that the underlying price trend has already reversed.
Official syllabus reference: [1]
35. Breadth divergence and participation risk
Market breadth measures how widely securities participate in a move. An advance-decline line accumulates advancing issues minus declining issues, while other measures track new highs, new lows, or participation above moving averages. An index reaching new highs with weakening breadth may depend on fewer constituents, creating a concentration warning. Breadth divergence can persist and is not a precise timing signal. Keep the universe and data treatment consistent, and recognize that capitalization-weighted index behavior can legitimately differ from equal-count breadth during periods of concentrated leadership.
Worked example: A capitalization-weighted index reaches a new high, but only 40% of its constituents remain above their fifty-day averages, down from 70%. The analyst identifies narrower participation and reviews concentration, while waiting for the strategy's actual risk trigger before changing exposure.
Common mistake: Weak breadth does not guarantee an immediate index decline, because a small group of heavily weighted constituents can sustain the index.
Official syllabus reference: [1]
36. Trend and mean-reversion regime selection
Trend-following seeks continuation after directional evidence, while mean-reversion strategies seek movement back toward a reference after an extension. A market with persistent directional movement can reward the first and repeatedly hurt the second; a range can produce the opposite pattern. Define regime filters from information available at the decision time, and evaluate transitions rather than assuming clean categories. Filters introduce their own lag and estimation error. A regime label should explain why a strategy's mechanism may apply, not retrospectively excuse every losing trade or authorize untested discretionary switching.
Worked example: A strategy buys range lows during stable congestion but stops doing so after repeated lower lows and a declining long-term average. Its predefined filter changes eligibility because the evidence now favors directional persistence rather than repeated returns to the old midpoint.
Common mistake: A regime identified perfectly after the fact may be impossible to recognize in real time without lag, ambiguity, or false transitions.
Official syllabus reference: [1]
37. Cross-asset confirmation and conditional relationships
Cross-asset analysis examines whether movement in one market supports or challenges an interpretation in another. Rates, currencies, commodities, credit, and equities can share economic drivers, but their relationships vary with the nature of the shock. A bond selloff associated with stronger growth can imply something different from one driven by inflation or funding stress. State the hypothesized transmission mechanism and compare contemporaneous evidence rather than relying on a permanent sign rule. Cross-asset confirmation improves scenario reasoning only when matched to the relevant horizon, currency exposure, and market regime.
Worked example: Equities rise while credit spreads narrow and economically sensitive commodities strengthen. The analyst treats this combination as consistent with improving risk appetite, but checks whether currency changes or supply disruptions explain the commodity move before using it as independent confirmation.
Common mistake: A historical cross-asset correlation is not a universal economic law; different shocks can reverse the relationship or alter its timing.
Official syllabus reference: [1]
38. Sector rotation from relative leadership
Sector rotation reallocates exposure as leadership changes among groups. A technical process may combine sector-to-market relative-strength trends, absolute trend eligibility, and momentum rankings. Require comparable return series and specify how quickly leadership changes must persist before trading. Ranking alone can produce heavy concentration in related sectors, while rapid switching raises turnover. Translate leadership evidence into weights through explicit risk and concentration rules. Rotation describes observed market behavior; assigning a business-cycle narrative afterward does not establish that the sector sequence was predictable or that the next phase must follow.
Worked example: Technology loses relative momentum while industrials improve and remain above their long-term average. A rotation process trims technology only when its predefined ranking and trend rules change, then checks whether the new industrial exposure duplicates existing cyclical risk elsewhere.
Common mistake: A sector's improving rank does not automatically justify a full allocation when absolute trend, concentration, and portfolio risk remain unfavorable.
Official syllabus reference: [1]
39. Bollinger Band width and volatility expansion
Bollinger Bands place upper and lower boundaries around a moving average using a chosen multiple of rolling standard deviation. Narrowing width indicates reduced dispersion under that calculation, often called a squeeze. It does not determine the direction of a subsequent expansion. A band touch can represent trend strength or a possible extension depending on the strategy and regime. Specify lookback, standard-deviation convention, and signal rule, then use price structure or another justified trigger for direction. Band widths based on price levels also require careful interpretation when comparing different securities.
Worked example: A stock's bands contract around USD 50 after several quiet weeks. The analyst prepares both upside and downside breakout scenarios rather than assuming a rally. A later close above established resistance supplies the directional condition, while the squeeze supplies the volatility context.
Common mistake: A volatility squeeze does not inherently predict an upward breakout, and touching an upper band is not automatically a sell signal.
Official syllabus reference: [1]
40. Implied volatility and horizon-consistent ranges
Implied volatility is the volatility input consistent with observed option prices under a specified pricing framework. It reflects market pricing of uncertainty and risk compensation, rather than a direct forecast of direction. An approximate horizon standard deviation can be derived using square-root-of-time scaling when its assumptions are appropriate. State whether calendar days or trading days define the quoted convention; do not mix them silently. Actual return distributions can be skewed, fat-tailed, and time-varying, so an implied-volatility range is a model-based reference rather than a guaranteed probability interval.
Worked example: Using an explicitly assumed 365-day convention, 30% annualized implied volatility gives an approximate thirty-day standard deviation of 30% × √(30/365), or 8.60%. The analyst describes the estimate as dispersion and does not infer a bullish or bearish move.
Common mistake: A high implied volatility reading indicates expensive uncertainty under the pricing model, not a directional forecast or a guaranteed realized move.
Official syllabus reference: [1]
41. Cycle period, phase, and amplitude
A cycle's period is the time required for one full repetition, amplitude describes its magnitude around a reference, and phase describes its position or offset relative to a benchmark cycle. Separating these properties prevents a timing model from being mistaken for a price-target model. Market cycles are noisy and can change length or disappear. Estimate cycles without using future turning points, and evaluate whether apparent periodicity persists outside the fitting sample. Combine timing windows with price confirmation rather than assuming a mathematically fitted trough forces the market to turn.
Worked example: A fitted twenty-day cycle projects a trough window next week, but its amplitude has recently weakened. The analyst watches for a price reversal in that window rather than assigning a fixed upside objective or buying solely because the projected date arrives.
Common mistake: A cycle's phase identifies timing within a repetition; it does not specify amplitude, guarantee a turning point, or establish trade direction alone.
Official syllabus reference: [1]
42. Rebalancing bands and turnover trade-offs
Rebalancing restores a portfolio toward target weights after price changes or revised forecasts. Calendar rules act at scheduled intervals, while tolerance bands act when deviations exceed specified limits. Frequent rebalancing controls drift more tightly but can increase costs, taxes where applicable, and market impact. Wide bands reduce trading while allowing larger departures from intended risk. Define whether turnover sums purchases and sales or uses a one-sided convention. Evaluate the mechanism against the mandate: restoring diversified risk differs from deliberately allowing momentum winners to run under a tested policy.
Worked example: A portfolio targets 60% equities with a five-percentage-point tolerance band. Equity weight rising to 63% causes no trade; reaching 66% triggers the specified review or rebalance. The analyst estimates costs before restoring the target and reports the turnover convention.
Common mistake: Rebalancing is not automatically beneficial at every small deviation; its risk-control value must be weighed against implementation costs and the strategy's design.
Official syllabus reference: [1]
43. Economic hypotheses before signal optimization
A quantitative trading hypothesis states why a measurable condition might predict returns or improve risk management. It identifies the universe, observation timing, expected mechanism, holding horizon, and circumstances that could invalidate the idea. Translating the hypothesis into rules makes the analysis reproducible and limits retrospective storytelling. Economic plausibility does not replace empirical testing, but it helps distinguish a potentially persistent effect from a data accident. Record major alternatives and development choices so that impressive results can be evaluated against the amount of searching that produced them.
Worked example: An analyst hypothesizes that sustained institutional buying creates medium-term continuation. Before testing, the analyst specifies a liquid universe, a six-month return signal, monthly observation timing, next-session execution, and a holding rule, then evaluates whether results survive realistic costs.
Common mistake: A profitable optimized rule does not explain its own mechanism, and inventing a narrative afterward does not remove data-mining concerns.
Official syllabus reference: [1]
44. Trigger, filter, and value rules
A trigger specifies the event that initiates an action, such as a completed close above resistance. A filter determines whether that action is eligible, such as requiring a positive longer-term trend. A value rule assigns a score or quantity used to compare or size opportunities, such as momentum rank or estimated volatility. These functions can interact but should remain explicit in implementation. Separating them makes rejected trades and model changes explainable. A filter that improves historical performance still needs independent validation because added conditions can overfit and reduce the effective sample.
Worked example: A system enters after a twenty-day high breakout, only when price exceeds its two-hundred-day average, and sizes eligible trades inversely to volatility. The breakout is the trigger, the long-term trend is the filter, and volatility provides a sizing value.
Common mistake: Changing a filter is a model change requiring evaluation; it is not merely a harmless reporting choice because the trigger remains unchanged.
Official syllabus reference: [1]
45. Point-in-time data and survivorship bias
Point-in-time data reproduce the information actually available at each historical decision. Survivorship bias arises when a backtest includes only securities that remain in today's universe, excluding failed or delisted names. Look-ahead bias includes revised fundamentals, future constituent lists, or closing information used before it was available. Corporate-action adjustments also require consistent handling so returns and execution prices remain coherent. Audit timestamps, membership histories, delisting returns, and price adjustments before interpreting model performance. A sophisticated signal cannot repair a dataset that gives the strategy knowledge unavailable to a real trader.
Worked example: A historical stock-selection test uses today's successful index members throughout the prior decade. Replacing that list with each date's actual membership and including delisted securities reduces the return, revealing that part of the original result came from future survival information.
Common mistake: Using a long price history for current securities does not make a backtest point-in-time if historical universe membership contains future information.
Official syllabus reference: [1]
46. Walk-forward testing and untouched holdouts
Walk-forward testing repeatedly develops or calibrates a model on earlier data and evaluates its frozen rules on a subsequent segment. The process approximates how decisions would evolve through time. A final untouched holdout provides an additional check only while it remains independent of design choices. Inspecting failures and revising rules using that holdout converts it into development evidence. For overlapping labels or holdings, prevent information leakage across boundaries. Record every tuning and evaluation step, since repeated informal checks can consume independence even when the dataset is still called out-of-sample.
Worked example: A model is calibrated on three years, traded unchanged for the next six months, and then recalibrated using only available history. After several such folds, a reserved final year is examined once. Revisions based on that year require new independent evaluation.
Common mistake: A dataset does not remain an untouched test merely because its filename says holdout after its results have influenced model design.
Official syllabus reference: [1]
47. Multiple testing and selection inflation
Testing many strategies or parameter combinations increases the chance of finding unusually favorable historical results by luck. Selecting the highest Sharpe ratio from a large search therefore creates an optimistic estimate even if each individual calculation is correct. Correlation among trials affects the amount of independent searching but does not remove the concern. Document the search space, compare simpler alternatives, examine stability, and use genuinely independent evaluation. Statistical corrections and resampling can assist, yet none makes repeated retrospective selection equivalent to a single prespecified test with the same apparent performance.
Worked example: An analyst tries 500 moving-average pairs and reports only the best pair's Sharpe ratio. A nearby broad range performs poorly. The impressive winner is less persuasive than a modest result from a prespecified rule that remains stable on new observations.
Common mistake: Correctly calculating a selected model's performance does not correct the optimism introduced by searching many alternatives and reporting only the winner.
Official syllabus reference: [1]
48. Parameter plateaus and robustness
Robustness asks whether a model's behavior survives reasonable changes in parameters, datasets, instruments, regimes, and implementation assumptions. A broad parameter plateau is generally more reassuring than an isolated sharp optimum because small estimation changes do not destroy the result. However, neighboring settings often share trades, so their similarity is not fully independent evidence. Test economically meaningful perturbations and inspect whether the same mechanism remains plausible. Robustness supports implementation confidence but does not guarantee future profitability or justify ignoring deterioration once the strategy is deployed.
Worked example: Lookbacks from seventy to one hundred days produce similar net results, while a single eighty-three-day setting is marginally best. The analyst prefers a simple representative setting and checks other periods, rather than treating the exact historical optimum as essential.
Common mistake: A broad parameter plateau is useful evidence, but highly overlapping neighboring tests do not count as many independent confirmations of an edge.
Official syllabus reference: [1]
49. Net performance, market impact, and capacity
Implementation analysis converts a theoretical signal into expected executable results. Include commissions, spreads, slippage, financing, borrow availability and cost where relevant, and market impact appropriate to order size. Costs often rise with urgency, volatility, and participation in available volume, making capacity nonlinear. A model profitable at small size can become unattractive when scaled. Distinguish fixed assumptions from estimates supported by execution data, and stress them. Gross performance is useful for understanding the signal, but capital allocation should consider whether a realistic implementation retains sufficient net benefit.
Worked example: A daily strategy earns an estimated 0.18% gross per complete trade. Spread, commissions, and slippage total 0.12%, leaving 0.06%. Doubling size raises impact enough to increase total costs to 0.20%, turning the same signal's estimated net return negative.
Common mistake: Multiplying a small-account backtest by a larger capital amount assumes capacity and costs scale linearly, which may fail in actual execution.
Official syllabus reference: [1]
50. Expectancy combines frequency and payoff
Expected value per trade combines the probability of each outcome with its payoff. Under a simple win-loss model, expectancy equals win probability times average win minus loss probability times average loss, with costs subtracted if not already included. Expressing results in R can normalize different planned risks, but inconsistent R definitions distort comparisons. Win rate alone cannot establish an edge, and the sample estimate has uncertainty. Inspect tail outcomes, serial dependence, and how position sizing changes dollar results before translating a positive average trade into an implementation decision.
Worked example: A strategy wins 40% of trades with average gain 2R and loses 60% with average loss 1R. Gross expectancy is 0.20R. Complete-trade costs of 0.08R reduce estimated net expectancy to 0.12R, despite more losing trades than winners.
Common mistake: A high percentage of winning trades does not imply positive expectancy when occasional losses or trading costs outweigh the smaller gains.
Official syllabus reference: [1]
51. Profit factor and aggregate outcome balance
Profit factor divides aggregate profits from winning trades by the absolute aggregate losses from losing trades. Values above one indicate positive net trade profit over the measured sample when all relevant costs are included consistently. It differs from the average win-to-loss ratio because trade frequencies also matter. Very few losses can produce an unstable or extreme estimate, and zero observed losses makes the usual denominator problematic. Profit factor omits return timing, capital usage, and drawdown path, so interpret it alongside sample size, exposure, and independent validation.
Worked example: Thirty winners average USD 300, producing USD 9,000 total profit. Seventy losers average USD 100, producing USD 7,000 total loss. Profit factor is 1.29, even though the win rate is only 30% and the average payoff ratio is 3.00.
Common mistake: Profit factor is not the average win divided by the average loss; aggregate totals must incorporate how often each outcome occurs.
Official syllabus reference: [1]
52. CAGR and geometric compounding
Compound annual growth rate is the constant annual rate linking beginning and ending wealth over a stated number of years, assuming external cash flows are absent or appropriately handled. Calculate it as ending wealth divided by beginning wealth, raised to one over years, minus one. Arithmetic average returns do not reproduce compounded wealth when returns vary. CAGR summarizes endpoint growth but conceals the path, drawdowns, and interim liquidity demands. For portfolios with contributions or withdrawals, select an appropriate time-weighted or money-weighted framework rather than applying the simple endpoint formula uncritically.
Worked example: An account rises 20% in year one and falls 20% in year two, taking USD 100,000 to USD 96,000. The arithmetic average is zero, but CAGR is √0.96 minus one, approximately −2.02% per year.
Common mistake: Averaging annual percentage returns arithmetically does not generally equal the annual compound growth rate or reproduce the account's ending wealth.
Official syllabus reference: [1]
53. Sharpe ratio and dependence-aware annualization
The Sharpe ratio compares mean excess return over a specified risk-free reference with the standard deviation of those excess returns. Match measurement frequency and define the return series consistently. Square-root annualization of a periodic Sharpe ratio relies on assumptions including limited serial dependence; smoothed or autocorrelated returns can make the usual shortcut misleading. The ratio treats upside and downside variation symmetrically and may underrepresent tail risk. A higher historical value can reflect leverage-independent efficiency, selection bias, or understated volatility, so inspect the underlying return process and sample uncertainty.
Worked example: A strategy has annual excess return of 8% and annual excess-return volatility of 10%, giving a Sharpe ratio of 0.80 under the stated convention. Before comparing it with another strategy, the analyst checks whether both use comparable frequencies and valuation methods.
Common mistake: A high Sharpe ratio does not prove low tail risk, and blindly annualizing autocorrelated periodic returns can overstate risk-adjusted performance.
Official syllabus reference: [1]
54. Information ratio and active risk
The information ratio divides average active return relative to a specified benchmark by tracking error, the volatility of active returns. It measures consistency of benchmark-relative performance rather than total portfolio risk. Benchmark choice can materially change both numerator and denominator, so use one aligned with the mandate and maintain comparable currency and return treatment. A portfolio can have a strong information ratio while losing money if its benchmark performs worse. Inspect whether active return reflects intentional skill, factor exposures, or an unsuitable benchmark before drawing conclusions about manager quality.
Worked example: A portfolio earns 9% while its benchmark earns 6%, and annualized tracking error is 4%. The information ratio is 3%/4%, or 0.75. Its total volatility may still be high because tracking error measures differences from the benchmark.
Common mistake: Tracking error is not total portfolio volatility, and a positive information ratio does not necessarily imply a positive absolute return.
Official syllabus reference: [1]
55. Sortino and Calmar reveal different risk preferences
The Sortino ratio divides mean return above a specified target by downside deviation relative to that target, using compatible time units and a stated annualization convention. It evaluates unfavorable return variability. The Calmar ratio divides CAGR by the absolute maximum drawdown over the evaluation window, emphasizing the deepest cumulative equity decline. Compare strategies over matching periods, with consistent return and cost assumptions. Rankings can differ because downside deviation measures shortfalls across observations, while maximum drawdown depends on their sequence. Neither ratio alone establishes implementation suitability or future risk.
Worked example: For matching three-year samples, assume a zero target. A: annualized mean return 15%, downside deviation 10%, CAGR 12%, maximum drawdown 24%. B: 12%, 10%, 10%, 10%, respectively. Sortino favors A: 15/10 = 1.50 versus 12/10 = 1.20. Calmar favors B: 10/10 = 1.00 versus 12/24 = 0.50.
Common mistake: Treating Sortino and Calmar as interchangeable ignores their different denominators; a strategy can exhibit moderate downside variability yet suffer a prolonged, deep drawdown.
Official syllabus reference: [1]
56. MAE and MFE diagnose intratrade behavior
Maximum adverse excursion measures the largest unfavorable movement during a trade, while maximum favorable excursion measures the largest favorable movement before exit. Normalize consistently by currency, percentage, or a fixed original R unit. Comparing excursions with final outcomes can reveal premature exits, recovery dependence, or missed profit opportunities. However, summaries omit the full sequence of price moves and can be distorted by bar-resolution limits. Proposed stop or target changes require a complete strategy simulation because altered exits can affect fills, reentries, position availability, and subsequent trades.
Worked example: A trade eventually gains 2R after first moving 0.8R against entry and later reaching 3R favorable excursion. A proposed 0.5R stop would have interrupted its recovery, while the 3R MFE suggests an opportunity that still requires executable target analysis.
Common mistake: Excursion summaries do not establish the profitability of new stops or targets without reconstructing timing, execution, and the complete strategy.
Official syllabus reference: [1]
57. Trade-sequence simulation and dependency
Resampling trade or return sequences explores how outcome ordering can change drawdowns, recovery time, and capital requirements. Randomly permuting fixed trades preserves their set of outcomes but changes path-dependent statistics. Sampling with replacement also changes which outcomes appear. Both methods can misrepresent strategies with serial dependence, changing exposure, or regime clustering. Block resampling can preserve some local dependence, though block choices introduce assumptions. Simulated ranges describe the supplied data and model, not every future possibility. Add scenarios for missing shocks and evaluate whether the simulated sequence remains operationally feasible.
Worked example: A strategy's historical profits are unchanged when trades are reordered, but several simulations place losses together and create much deeper drawdowns. The analyst also tests blocks of consecutive trades because independent shuffling would erase the observed tendency for losses to cluster.
Common mistake: Randomly shuffled histories do not reveal all future risks and can understate drawdowns when the real process contains persistent loss clustering.
Official syllabus reference: [1]
Ethics
58. Reasonable basis and analytical limitations
A defensible recommendation requires diligent analysis and a reasonable basis proportionate to the claim being made. For technical research, examine data quality, model assumptions, implementation costs, and evidence supporting the interpretation. Distinguish observable facts from forecasts and opinions, and communicate material limitations so recipients can assess the conclusion. Using a vendor model does not eliminate the need to understand its relevant strengths and weaknesses. The supplied CMT guide establishes ethics relevance and adopted references; detailed application should be distinguished from any claim that the syllabus itself verifies current standard wording.
Worked example: An analyst receives a vendor's exceptional backtest but cannot determine whether it includes delisted securities or costs. Before recommending implementation, the analyst investigates those assumptions and presents unresolved limitations rather than describing the result as demonstrated achievable performance for clients.
Common mistake: A reputable vendor's impressive result does not automatically provide a reasonable basis for an analyst's recommendation without appropriate investigation.
Official syllabus reference: [1]
59. Conflict disclosure and recommendation independence
A conflict exists when personal or institutional interests could interfere with duties or reasonably appear to compromise objectivity. Personal holdings, compensation arrangements, and business relationships can be relevant to a recommendation. Provide meaningful, timely, prominent disclosure where appropriate, and manage the conflict so analytical judgments remain independent. Disclosure alone does not authorize misconduct or replace other duties, including client priority where applicable. Evaluate both actual incentives and reasonable perceptions rather than asking only whether the analyst believes the recommendation is unbiased. Avoid assuming jurisdiction-specific legal requirements from a general ethics outline.
Worked example: An analyst recommends a thinly traded share already held in a personal account. The report prominently explains the holding, and the analyst follows applicable firm controls governing research and personal transactions rather than treating disclosure as permission to trade ahead of clients.
Common mistake: Disclosing a conflict does not automatically make every related action acceptable or remove the obligation to preserve independence and client interests.
Official syllabus reference: [1]
60. Hypothetical performance and fair presentation
Performance communication should enable readers to distinguish actual investment experience from hypothetical, backtested, or simulated results. Identify the return basis, relevant fees, material assumptions, and limitations that could affect interpretation. Avoid presenting selected favorable periods or model revisions in a way that creates a misleading impression of achievable experience. A general risk warning cannot correct a false description of the underlying record. Clear presentation also separates gross from net returns and explains important comparability limits. These principles address communication quality without assuming that the supplied syllabus establishes detailed reporting-standard compliance.
Worked example: A presentation shows a model's 22% historical return before transaction costs. The analyst labels it hypothetical, states the gross basis, explains signal timing and selection assumptions, and presents material limitations instead of placing it beside client returns under an undifferentiated performance heading.
Common mistake: Calling results illustrative does not cure a misleading presentation when their hypothetical status, fee basis, or material assumptions remain unclear.
Official syllabus reference: [1]