CMT Level III common mistakes
CMT Level III · Understand the traps and how to avoid them.
Treating a chart pattern as a standalone buy or sell signal without stating the confirmation rule, invalidation level, and measured-move target in the written answer.
Level III answers are graded on the reasoning chain, not the label. Candidates memorize pattern names from Level I/II and assume naming the pattern earns the mark, so they omit the trigger, the stop, and the objective that make the call testable.
What to do instead
Adopt a fixed four-part template for every pattern question: identification criteria, entry trigger, invalidation (price and time), and target/risk-reward. Practise rewriting each answer so a reader could execute it without seeing the chart. If any of the four parts is missing, the answer is incomplete.
Applying a trend-following tool (moving average crossover, MACD, ADX) to a range-bound market, or a mean-reversion oscillator (RSI, stochastic) to a strong trend, without first classifying the market regime.
Candidates learn indicators in isolation and reach for the one they know best. They skip the regime test because it feels like extra work, and the resulting signal is systematically wrong for the environment.
What to do instead
Make regime classification the first sentence of any indicator answer: trending, range, or transitional, with the evidence used (higher highs/lows, ADX level, price relative to a long moving average, volatility state). Then justify why the chosen tool suits that regime and state what would change the classification.
Ignoring volume and open interest confirmation, or citing volume only after the fact without saying what volume behaviour would have invalidated the price signal.
Price is visually dominant and easier to describe. Volume and open interest require a second data series and a judgement about whether expansion or contraction supports the move, so candidates default to price-only narrative.
What to do instead
For every breakout, divergence, or trend-continuation answer, add one explicit volume/open-interest sentence: what confirmation looks like, what non-confirmation looks like, and the action taken in each case. In futures, distinguish rising open interest (new positions) from falling (liquidation) and tie that to the interpretation.
Presenting a single indicator or single timeframe as the basis for a market opinion, instead of building a multi-timeframe, multi-factor case with stated weights.
Time pressure encourages one-line answers. Candidates also confuse confidence with conviction and believe a clean single signal is stronger than a messy composite.
What to do instead
Structure answers as: higher-timeframe context, intermediate setup, lower-timeframe trigger, then corroborating non-price evidence (breadth, sentiment, relative strength, intermarket). Explicitly note where factors conflict and how the conflict lowers conviction or position size.
Describing a divergence, Elliott wave count, or Fibonacci projection as if it were a forecast, without assigning a probability, an alternative count, or a level at which the thesis is abandoned.
Subjective methods invite narrative thinking. Candidates commit to one interpretation because hedging feels like weakness, and examiners mark down unqualified certainty about inherently probabilistic tools.
What to do instead
Always give a primary and an alternate scenario with the price level that switches between them. Attach an explicit confidence qualifier and the invalidation point. For wave counts, state the rules that constrain the count (wave 2 retracement limit, wave 3 not shortest, wave 4 overlap restriction) and show the count is rule-compliant.
Treating risk as position-level stop distance only, omitting portfolio-level exposure, correlation, and the contribution of the new trade to total portfolio risk.
Technical analysts are trained on single-instrument charts. Portfolio risk aggregation is a separate skill set, and candidates answer the risk question with a trading answer.
What to do instead
For any recommendation, compute the risk in three layers: per-trade risk (entry to stop times size), sector/factor concentration after the trade, and marginal contribution to portfolio volatility given correlations. State position sizing as a fraction of risk budget, not a share count.
Using a stop-loss placement that ignores volatility and market structure, such as a fixed percentage or a round number, and then reporting risk without stress-testing a gap through the stop.
Fixed-percentage stops are easy to remember and easy to grade mentally. Candidates forget that stops are not guaranteed fills and that volatility regimes shift the appropriate distance.
What to do instead
Derive stops from structure (swing high/low, prior consolidation edge) plus a volatility buffer such as an ATR multiple. Then show the realised-loss scenario under a gap or slippage assumption and recompute risk-reward on the worse fill. Note the difference between normal and stressed liquidity.
Reporting a single risk statistic (usually standard deviation or a Sharpe ratio) as the complete risk picture, without drawdown, tail, or path-dependent measures.
Textbook risk is variance-centric. Candidates reproduce the metric they memorized rather than matching the measure to the question being asked, such as 'can this portfolio survive a bad year' versus 'how bumpy is the ride'.
What to do instead
Match measure to purpose: volatility for dispersion, maximum drawdown and drawdown duration for path risk, VaR/CVaR or scenario loss for tail risk, beta and factor exposures for systematic risk, and Sortino or Calmar when the concern is downside. Say explicitly what each measure misses.
Confusing the direction of a hedge or overlay, or sizing it by notional rather than by risk, so the stated hedge does not actually neutralise the exposure described.
Derivative mechanics (long versus short the hedge instrument, contract multiplier, beta adjustment) are error-prone under time pressure, and candidates skip the arithmetic check that the hedge works.
What to do instead
Write the hedge as an equation: exposure to be neutralised, hedge instrument beta or delta, contract value, and number of contracts, then verify the post-hedge net exposure is near zero. Sanity-check direction verbally ('I am long the portfolio, so I sell futures') before moving on.
Ignoring behavioural and governance risk in the recommendation, such as overconfidence after a winning streak, recency bias in regime labelling, or a stop that the client will not actually tolerate.
Candidates treat the answer as a pure analytics exercise and forget that Level III portfolio settings include the client's constraints, the manager's incentives, and the discipline required to execute the plan.
What to do instead
Close each recommendation with an implementation paragraph: behavioural failure points, the pre-committed rule that overrides discretion, suitability against the client's stated risk tolerance and liquidity needs, and how the plan is monitored and reviewed.