Trade Decision Attribution: Separate Signal, Sizing, Execution and Exit Skill

Trade Decision Attribution: Separate Signal, Sizing, Execution and Exit Skill

A profitable trade does not prove that every decision was good. The signal may have worked while oversizing, poor execution or an early exit reduced the result. A losing trade can also contain a sound process. Decision attribution uses consistent counterfactual portfolios to identify which stage added or removed value.

বাংলায় পড়ুন: Trade Decision Attribution: Signal, Sizing, Execution ও Exit Skill আলাদা করুন

Learning objective: break total trading P&L into signal, sizing, execution, exit and cost components, build a worked counterfactual example, and review repeated decisions without confusing luck with skill.

1. Total P&L answers the wrong diagnostic question

Net P&L tells the trader what happened to capital. It does not explain why. Consider four possibilities:

• good signal, excessive size;
• poor signal, lucky exit;
• good entry, weak exit discipline;
• correct direction, expensive execution.

Treating all profit as skill and all loss as error creates outcome bias. The review needs to compare the actual trade with versions in which one decision changes at a time.

CFA Institute describes decision attribution as a bottom-up approach that can examine entry, speed of sizing, position size, additions, trims and exit. It becomes useful when applied consistently over many observations.

2. Define the policy trade first

Attribution needs a frozen reference process. Before trading, record:

• the signal and decision timestamp;
• standard position size;
• intended entry rule;
• invalidation and stop policy;
• intended exit rule or evaluation horizon;
• expected fees and financing;
• benchmark or market return if relevant.

This creates the policy trade: what the written strategy would have done without discretionary changes. The actual trade can then be compared with it.

3. Five useful attribution buckets

Signal or selection

Did the chosen asset move favourably over the strategy's stated horizon, compared with cash or a relevant benchmark?

Sizing

Did holding more or less than the standard size add value? Correct larger positions help; incorrect larger positions hurt.

Execution

How much value was lost or gained between the decision price and actual fills? Include delay, spread and market impact. The implementation shortfall guide explains this calculation.

Exit

Did the actual exit add value relative to the predefined rule? An early exit can reduce loss or surrender a winner. A late exit can capture trend or give back profit.

Explicit costs

Commission, exchange fees, tax, borrow charges and financing reduce the result even when every market decision is unchanged.

4. Worked example

A written strategy calls for buying 100 shares at £50.00 and exiting at the rule-based price of £54.00. The trader instead chooses 150 shares, gets an average fill of £50.30, exits early at £53.20 and pays £15 in total fees.

Actual net P&L is:

150 × (£53.20 − £50.30) − £15 = £420

Build the attribution in a fixed order:

Component	Calculation	Contribution
Signal on standard 100 shares	100 × (£54.00 − £50.00)	+£400
Extra sizing	50 × (£54.00 − £50.00)	+£200
Execution versus £50.00	150 × (£50.00 − £50.30)	−£45
Early exit versus £54.00	150 × (£53.20 − £54.00)	−£120
Fees	Actual charges	−£15
Actual net P&L	Sum of components	+£420

The trade made money, but the result does not mean the complete process was optimal. The signal and larger size added value in this observation; execution, exit and fees removed £180.

5. Counterfactual order matters

Sizing and return interact. In the example, the extra 50 shares are credited using the policy entry and exit. Another valid framework could assign some interaction to execution or exit.

CFA Institute's performance-attribution research notes that interaction effects can be shown separately or assigned according to the investment process. There is no honest reason to change the order from trade to trade. Choose a method, disclose it and reconcile every component to actual net P&L.

6. One trade cannot establish skill

In the worked example, oversizing helped only because the trade won. That does not prove sizing skill. Across many trades, compare:

• average and median contribution by decision type;
• hit rate and tail loss for discretionary size changes;
• execution cost by liquidity and urgency;
• exit contribution by setup and holding period;
• results in trending, ranging and high-volatility regimes;
• contribution after fees and financing.

CFA Institute's timing-versus-sizing review distinguishes being correct about direction from assigning larger weights to the decisions that prove correct. Both require a sufficiently large sample.

7. Review process, not hindsight perfection

Do not compare every exit with the highest price later reached. That benchmark was not knowable or executable as a rule. Compare with a pre-specified exit: close below a moving average, fixed time, target, trailing stop or portfolio rebalance.

Likewise, do not call an entry poor merely because price dipped afterwards. Execution should be measured against an available benchmark at the time; signal quality should be measured over the planned horizon.

8. Build a practical attribution journal

Add these fields to each completed trade:

1. policy quantity and actual quantity;
2. decision price and average fill;
3. policy exit and actual exit;
4. benchmark return over the holding period;
5. signal, sizing, execution, exit and fee contributions;
6. reason for every discretionary override;
7. market regime and event context;
8. confidence level recorded before the outcome.

Review monthly or after a meaningful sample. Change one process rule only when the same weakness repeats across comparable trades.

Common mistakes

1. Calling every profitable trade a good decision.
2. Comparing exits with the unknowable later high or low.
3. Changing the attribution order to improve the story.
4. Ignoring interaction between size, entry and exit.
5. Excluding fees, financing and borrow costs.
6. Mixing signal horizon with execution horizon.
7. Judging sizing skill from one large winner.
8. Aggregating unrelated strategies and regimes.

Practical checklist

• Policy trade written before execution?
• Signal horizon defined?
• Standard and actual size recorded?
• Decision price and every fill stored?
• Policy and actual exits recorded?
• Attribution order fixed across trades?
• Components reconcile to net P&L?
• Discretionary overrides tagged?
• Results grouped by comparable setup and regime?
• Process change supported by repeated evidence?

The Money Man View

The purpose of attribution is not to invent a perfect trade after the fact. It is to find repeatable strengths and leaks in the decisions that were genuinely available. A trader who separates signal, sizing, execution and exit can improve one component without accidentally abandoning another that already works.

Educational risk disclosure

This tutorial uses a simplified hypothetical example and was prepared with AI-assisted research and writing under The Money Man editorial workflow. It is educational information, not personalised investment advice. Attribution results depend on benchmark and counterfactual choices, past decision quality may not persist, and trading can cause substantial loss.