Connect behavior to observable decisions
Behavioral finance studies how judgment can depart from a purely rational model. In the 2026 CMT Program Guide, Level I public objectives include prospect theory, loss aversion, belief-perseverance biases, information-processing biases, and emotional biases. The Level II outline includes behavioral finance within its broad Theory and History area and applies behavioral analysis to risk decisions. The public guide does not assign a standalone percentage to “behavioral finance,” nor does it promise particular bias names will appear as exam questions. Use the assigned curriculum to determine depth.
This legacy guide teaches a decision-audit method. It complements Statistical Analysis because bias control and statistical validation often solve the same practical problem from different angles.
The bias is in the process, not the price chart
An investor buys at 100, watches price fall to 88, and refuses to sell because “I need to get back to 100.” The purchase price is an anchor. It may matter for taxes or accounting, but it is not independent evidence of current value or trend. A better question is: If I had no position today, would I initiate one at 88 under my current rules? If the answer is no, holding requires a reason other than the purchase price.
Loss aversion is the tendency for losses to feel more painful than comparable gains feel pleasant. Prospect theory describes choices relative to a reference point and helps explain why someone may become willing to take greater risk after a loss. This does not mean every person behaves the same way or that one fixed numerical “pain multiplier” applies. It predicts a possible pattern of choices, which a decision journal can reveal or reject for an individual.
Confirmation bias appears when the trader records supportive evidence but dismisses contradictory observations. Suppose the thesis says a stock is in an uptrend. The analyst highlights a rising moving average but ignores lower swing highs and widening downside volume. The cure is to write the failure condition at entry and score each new observation against it. A prewritten invalidation rule reduces the power of a story to adapt after the fact.
Worked case: a stop that moves for the wrong reason
A trader enters 100 shares at 50 with a planned exit at 46. The initial price risk is (50 − 46) × 100 = 400 before fees and slippage. Price falls to 46. The trader lowers the stop to 42 only because selling now would realize a loss. New potential loss from entry is (50 − 42) × 100 = 800, twice the original amount. No fresh evidence improved the trade; the rule changed in response to discomfort.
A defensible alternative is to take the planned exit, or to rewrite the entire thesis only if genuinely new, documented evidence warrants a different trade with its own size and risk limit. This does not claim stops always prevent larger losses. Gaps can execute beyond a stop, and some strategies use different exits. The point is to avoid silently doubling planned risk to escape the feeling of loss.
If the trader instead owns 200 shares, a four-point risk implies 800 of initial exposure. Position size and stop distance interact. A price decision cannot be separated from the portfolio risk it creates.
Crowd information: signal and trap
Survey enthusiasm, options positioning, and strong price momentum can tell an analyst that a trend has broad participation. A crowded trade might continue longer than a contrarian expects. It may also become fragile when expectations are extreme. Neither conclusion follows from “the crowd is bullish” alone. Define the historical reference, sample composition, and price trigger. For example: “I will label sentiment extreme only above the 90th percentile of its own ten-year history, and I will require a break of support before changing exposure.” That 90th-percentile threshold is an illustrative research rule, not an official CMT standard.
Herd behavior can be rational in part: participants may respond to the same public information. Calling everyone else irrational is itself an overconfidence error. To test a contrarian idea, compare past sentiment extremes with subsequent outcomes across all observations, including cases where the trend persisted. Check whether the indicator was known in real time and whether the market regime changed.
A six-line decision journal
Before acting, record the date, thesis, observed evidence, strongest opposing evidence, action threshold, and invalidation. Afterward, record whether you followed the process independently of profit or loss. A good decision can lose money; a poor decision can make money by chance. Score process and outcome separately.
For a chart-driven trade, an entry such as “breakout above 75” is incomplete. Add why 75 is meaningful, whether the close or intraday high counts, where the idea fails, and what size keeps a loss tolerable. The journal then exposes whether you used hindsight to change a rule. For a study problem, practice naming the cognitive mechanism and writing the corrective procedure, rather than merely memorizing a definition.
Common mistakes
Do not label every losing trade a behavioral error. Markets are uncertain. Do not use “anchoring” to dismiss a carefully reasoned valuation reference. Do not claim that loss aversion is always exactly two times gain pleasure; estimates depend on context and measurement. Do not infer that sentiment extremes precisely time reversals. Finally, beware hindsight bias: a chart is much easier to “read” after the subsequent bars are visible.
Self-check
You planned to risk 1% of a 20,000 account, or 200. Your entry is 40, invalidation 38. Ignoring fees, what is the maximum whole-share position, and what bias appears if you later move the exit to 36 solely to avoid a loss? Answer: Price risk is 2 per share, so 200/2 = 100 shares. Moving to 36 doubles risk to 4 per share, or 400 for 100 shares; the reference-point discomfort is consistent with loss aversion.
Next, take your last three practice scenarios. For each, write the strongest disconfirming fact before reading the explanation. That habit is more valuable than simply labeling biases after an outcome is known.