OptiNod Academy
Precommitment and pass criteria — Set the rules before judgment hardens
Fix entry, exit, sizing and rejection criteria before a trade begins. That removes discretion at execution time and lets you measure its cost through rule adherence.
If a forward test merely asks whether money is being made, too little data can accumulate alongside growing confirmation bias. Fix entry, exit, sizing and rejection criteria before your opinion hardens, and bias has less room to enter.
Precommitment is a concept developed by economist Thomas Schelling in the 1970s: when you cannot trust the choices your future self will make under pressure, remove some of those choices now. At execution time, profit and loss are already moving and judgment is less steady. Make the decision instead when the trade's profit and loss is still zero.
Psychology describes a related principle as implementation intentions. Peter Gollwitzer formalized it in 1999: state in advance, “If situation X occurs, I will do Y.” Then the prescribed action can replace a fresh judgment at the moment it matters. When an entry signal appears or a stop is reached, the trader executes the rule decided earlier.
Discipline is often understood as willpower in the moment. If willpower is the only method, the remedy for failing to honor a stop is simply “try harder,” which may fail again on the next loss. Every bias discussed earlier in this series appears at execution time. Loss aversion, the disposition effect and FOMO chasing all pull on judgment after profit or loss starts moving. Willpower must fight them each time; precommitment moves the decision earlier so the fight is less necessary.
The cost shows up in the sample. Without fixed rules, a forward test that only asks whether the strategy “makes money” counts winning trades as evidence and dismisses losing trades as exceptions. That is how confirmation bias accumulates without a trade journal. With rules fixed beforehand, you can separate compliant and noncompliant trades and calculate expectancy for each. The difference is a measure of the cost of discretion: compliant-trade expectancy minus noncompliant-trade expectancy gives a concrete measure of the execution gap.
Rules must be fixed before judgment hardens
Precommitment has one essential condition: fix the rules before opening a position. Before entry, profit and loss is zero, so there is no personal reference point pulling you toward avoiding or seeking risk. A stop and target chosen then are less affected by the emotions of a live position.
As soon as a position opens, the reference point shifts to your entry price. When profit and loss move, familiar biases appear. In profit, you may want to close quickly to protect it. In loss, you may want to postpone the stop to get back to break-even. A rule invented then already reflects those feelings. “I will decide the stop after seeing the price” can merely dress risk seeking during a loss in the language of a rule.
Set every number for entry, exit and sizing before clicking to enter. A rule created afterward may look formal but cannot prevent the bias that shaped it. The first condition of precommitment is when the rule is made.
Fixing entry, exit and sizing removes discretion at execution
Bias affects three decisions at execution: when to enter, when to exit and how much to risk. Fix each in advance as a measurable condition, leaving less room to decide anew in the heat of the moment.
- Entry condition: Specify the signal bar and place a limit order at the planned entry price. This blocks a market-order chase after the signal has already appeared.
- Exit condition: Submit a stop and first target as an OCO order at entry. Base the stop on volatility and structure, not on a wish to recover your entry price.
- Sizing: Set risk per trade to a fixed share of the account, such as 1R, and do not raise or lower it based on recent winning or losing streaks. A fixed share prevents the bet adjustments driven by the gambler's fallacy.
- Rejection signal: Record when to cancel an entry, including a signal invalidation price and the maximum number of bars to wait.
The common feature is that the numbers are set before entry. With entry price, stop price and risk share written down, execution follows what is written rather than a new judgment. As the article on the disposition effect showed, bringing the exit decision forward to the entry reduces the opening for asymmetric emotion. Fixing sizing the same way also limits overconfidence after losses.
An entry whose three conditions cannot be written numerically is not fully precommitted.
Set pass and rejection criteria before the test so the sample can inform judgment
A common forward-testing mistake is to “run it for a while and go live if it makes money.” There is no ending rule. The test stops when the result looks profitable, often before the sample is large enough. Set numerical pass and rejection criteria before starting so the stopping point does not depend on the result.
Pass criteria need both a minimum sample size and market conditions covered. A sample consisting only of a rising market inflates a long strategy's apparent performance. Bitcoin rose about 67% from its April 7, 2025 low of $74,508 to $124,474 on August 14. During that move, nearly any strategy that repeatedly bought without respecting pullbacks would have made money. Declaring it “passed” after 20 trades from that period does not validate its edge; it only confirms the background was a bull market.
The same rules can perform differently in a decline. Bitcoin fell about 36% from near $126,200 on October 6, 2025 to $80,600 on November 21, and a repeated-buying rule would have suffered consecutive losses. Requiring at least N trades spanning both rising and falling periods prevents a strategy from going live too early on a bull-only sample. It applies to live samples the kind of scrutiny that walk-forward analysis and robustness checks bring to backtests.
Rejection criteria are even more important. Unless you write “stop if maximum drawdown exceeds X%” or “reject if expectancy is negative after N trades” before the test, a losing run can always be excused as “not enough data yet.” A test without a rejection line can accumulate confirmation bias indefinitely. Only a line fixed in advance ends that process.
A forward test without numerical pass and rejection criteria can become evidence collecting disguised as validation.
Automation removes discretion at execution time
Writing down a commitment still leaves execution in human hands. If you move the mouse to cancel rather than honor a stop when price reaches it, the written rule does nothing. A further step is to separate execution itself from moment-to-moment discretion.
TradingView alerts can fire entry and exit signals without constant screen watching. Connect them by webhook to a forward test or an exchange bot, and the handoff from signal to order no longer requires another human decision. As the signal-to-execution gap showed, removing discretion at that point can reduce the gap created by behavior.
Automation does not solve everything. It reduces behavioral discretion while leaving mechanical costs such as slippage and fees. The costs differ: behavioral cost is unknown until measured; mechanical cost can be estimated in a backtest. Compare the two when deciding how much execution to automate.
The purpose of automation is not to promise better performance. It is to remove discretion from the execution step by design.
Adherence and expectancy of violations put a number on discretion
You cannot judge whether precommitment is working by feel. Add a rule-adherence column to the trade journal and calculate two figures.
The first is the adherence rate: the share of trades that followed entry, exit and sizing rules together. A 100% rate need not be the aim, but a low rate means the rules are not being followed in live trading.
The second is the expectancy difference between compliant and noncompliant trades. Split the trades by whether every rule was followed, then calculate average R for each group. For example, suppose 40 compliant trades out of 50 average +0.25R, while 10 violating trades average −0.9R. The numbers show how much those violations lowered overall expectancy. This difference measures the execution gap and the cost of discretion.
Categorize violations by cause to determine what to fix. Separate entries that chased after ignoring a limit order, exits whose stops were moved farther into loss, bets increased after a losing streak, and entries taken without a signal. Expectancy by violation type shows which bias costs the most and which rule to fix next.
Calculating adherence and violation expectancy regularly turns “I should follow my rules” into a ranked list of specific rules worth enforcing.
Two pitfalls
Assuming more rules are always better. Precommitment aims to reduce discretion, not to accumulate conditions. Adding indicator after indicator to an entry can turn confirmation bias into formal rules, and conditions tailored to the past may perform worse live. Rules need only be specific enough to remove judgment at execution.
Assuming automation removes psychology. Even when a bot places entries and exits, a person chooses when to turn it off or on. After a losing streak, “pause it for now” is a discretionary intervention that should be recorded as a violation. Automation reduces discretion in order execution but leaves discretion over stopping and restarting the system. Those decisions need precommitted rejection criteria too.
Check precommitment by measuring adherence
The biases covered in this series all appear at the point of execution. Once profit and loss move, loss aversion, the disposition effect and overconfidence can enter the decision. Precommitment moves decisions to before entry, when profit and loss is zero, reducing the opportunity for bias. The method has three parts: specify entry, exit and sizing numerically; fix pass and rejection criteria before testing; and automate execution through alerts and bots where appropriate.
Whether it works is shown by numbers, not by a promise to be disciplined. The expectancy gap between compliant and violating trades measures the cost of discretion, while expectancy by violation type determines which rule to lock down next. The final step is not resolving to follow rules. It is measuring whether you did and what violations cost.