Anchoring Bias in Trading: When Yesterday Sizes Today
Anchoring bias fixes a trader on a reference number, like a prior session's P&L, and lets it set the next session's size. Learn to catch it in your log.
By Imperial Analytics
A trader who made money yesterday tends to add contracts today; a trader who lost tends to either shrink to nothing or double up to win it back. In both cases the prior session's number, not the current trade's plan, is setting the risk. That pull toward a reference number is anchoring bias. This primer defines it, separates it from the biases it is often confused with, shows how a prior result anchors the next session's sizing, and gives a way to set size so the anchor never reaches the decision.
What anchoring bias is
Anchoring bias is the tendency to fix on an initial reference value and adjust too little away from it when making a judgment. Tversky and Kahneman described it in 1974 as anchoring and adjustment: people start from a salient starting number, even an arbitrary one, and their final estimate lands near it because the adjustment away is insufficient.1
In the original experiments, subjects who were shown a plainly arbitrary number before estimating a quantity produced estimates pulled toward that number, even though it carried no information.1 The starting point stuck. Later work on what Ariely called arbitrary coherence showed the same mechanism outside the lab: once a reference number is in mind, later judgments organize themselves around it in a way that feels reasonable but traces back to an anchor that never deserved the weight.2
For a trader the anchor is rarely a random number. It is usually the last session's profit or loss, the current account balance, or the entry price of an open position. Each is a real number, which is what makes it convincing. But a real number about the past is still an arbitrary starting point for a decision that should be made on the trade in front of you, and anchoring is the reason the mind treats it as the natural place to begin.
How it differs from its near neighbors
Anchoring bias fixes on a reference number. Loss aversion weights losses more heavily than equal gains, recency bias overweights the latest outcomes, confirmation bias filters evidence toward a held belief, and sunk cost defends money already spent. Anchoring is the number the mind cannot leave, not the pain, the recency, the filter, or the escalation.
The distinction matters because the fix differs for each. Loss aversion is about the asymmetric sting of a loss; it explains why a trader cuts winners early and holds losers. Recency bias is temporal; the last few trades dominate the read of a strategy regardless of any reference number. Anchoring is narrower and more specific: a single number is fixed in place, and every later judgment adjusts too little from it.
The overlap with the sunk cost fallacy is worth naming because both can keep a trader in a losing position. Sunk cost is the commitment to money already spent. Anchoring is the fixation on the entry price as the number the trade must return to before it can be closed. They often fire together, but the mechanism to correct is different: sunk cost asks whether you would enter now, while anchoring asks why a past price has any claim on the current decision. Confirmation bias filters incoming signals to fit a thesis; anchoring does not need a thesis, only a number.
How a prior session's result anchors the next session's sizing
The most expensive anchor for a futures trader is the prior session's result. A green day anchors size upward, because profit reads as license to add contracts, while a red day anchors it toward shrinking to nothing or oversizing to win the loss back. The size comes from the last number on the account, not the trade in front.
Consider the green-day case. The trader closed yesterday up a meaningful amount, and today the same setup appears. The plan calls for a fixed risk unit, but the recent profit sits in mind as a cushion, and the position goes on larger than the plan allows. The size was not chosen from the trade; it was adjusted up from yesterday's number. When the setup fails, the loss is larger than the plan ever sanctioned, and it lands on a day the trader was not prepared to give back.
The red-day case splits in two directions, and both are anchored to the same number. One trader, anchored to yesterday's loss, shrinks size to a fraction of plan and misses the qualified setup at the size that would have made it worth taking. The other, anchored to the deficit, sizes up to erase it in one trade, which is the sizing decision a daily loss limit exists to stop. In every version the error is the same: the sign and magnitude of the prior session set the size, when the only inputs that should are the current trade's stop distance and the account's fixed risk rule.
Data note
The sizing figures in the example below are illustrative round numbers chosen to show the mechanism, not results from any account or sample. Imperial Analytics surfaces a behavioral pattern from a trader's own data only when it meets the sample minimum in the AI Operating Charter, which is twenty trades in the matching condition.
Put numbers on it to see the shape. Suppose a trader's plan is a constant one-contract risk unit per setup. After a day up several hundred dollars, the same setup goes on at three contracts because the profit felt like room to work with. After a day down the same amount, the next setup goes on at four contracts to make the deficit back in one trade. Neither three nor four came from the trade; both were adjusted off the prior day's result. The plan said one, and the anchor overrode it in both directions.
The other anchors in a trading session
Within a session the anchors are the entry price, a round number, and the first result of the day. A position is held to its entry, a target sits at a round level because it is round, and the day's first win or loss frames the sizing that follows. Each is a number given weight it did not earn.
The entry-price anchor is the most familiar. A trade goes red, and the trader resolves to exit at breakeven rather than at the level the plan named, because the entry price has become the number the position must return to. The plan's invalidation is still the correct exit; the anchor is what postpones it. Round-number targets work the same way. A profit target lands on a round level not because the structure argues for it but because a round number is a comfortable place to anchor an exit.
↳ Note
The account balance is a fact about the past. The only number that should set the next trade's size is the risk on the next trade.
The first result of the day is the subtler one. A trader who opens with a win carries that number forward and sizes the rest of the session against it; a trader who opens with a loss does the same in the other direction. The session is young, the sample is one trade, and yet that single outcome becomes the reference the rest of the day adjusts from. Naming these anchors is the first step, because a number you have named as an anchor is easier to refuse than one operating silently.
How to catch it in your trade log
Anchoring on prior results leaves a mark when planned size is recorded before the session and actual size beside it. Log the intended risk unit for each setup, flag any trade whose size deviated, then sort the deviations by the prior session's result. Upsizes clustering after green days and oversizes after red days are the signature.
The bias thrives when size is decided in the moment, so the correction is to fix the intended size in advance and make the deviation measurable. Before the session, write the risk unit the plan assigns to each setup, expressed in R so it is comparable across trades. During the session, record the size actually taken next to the planned size, and tag each trade as planned or improvised at the moment it is placed, the same binary split covered in how to tag plan and improvised trades.
With those fields in place, the log answers the question directly. Pull every trade where actual size deviated from planned size, and line each one up against the sign of the prior session's result. If the upsizes cluster after winning days and the oversizes cluster after losing days, the prior session was setting the size. Read the pattern only once it clears the sample floor, the discipline behind what makes a behavioral pattern claim trustworthy and the sample size a claim needs; a handful of trades is a story, twenty in the matching condition is a count.
How to set size so the anchor cannot drive it
Three moves keep the anchor out of sizing: fix the risk unit as a constant fraction of the account, reviewed on a schedule not after a single session; size each trade from its own stop distance in R, not from yesterday's result; and keep the account-balance number out of the size decision. Each replaces a remembered number with a rule.
The constant risk unit handles the prior-session anchor. When the amount risked per setup is set as a fixed fraction of the account and revised only on a fixed schedule, a single green or red day has no channel through which to change it. The size is a rule, not a reaction, so there is no number from yesterday to adjust from. Sizing from the stop handles the individual trade: the risk unit divided by the distance to the stop gives the contract count, which ties the size to the trade in front of the trader rather than to any past result.
Separating the balance from the decision handles the rest. The running P&L should not be visible at the moment size is chosen, because a number in view is a number that anchors. For the red-day revenge case specifically, a daily loss limit is the structural floor that removes the option to size up into a deficit at all. Set this way, the size of the next trade comes from the plan, the stop, and the fixed rule, and the prior session's number never gets a vote.
Frequently asked questions
- q: Is anchoring bias the same as loss aversion? a: No. Loss aversion is the asymmetric weight given to a loss relative to an equal gain, which explains cutting winners and holding losers. Anchoring is fixation on a specific reference number, such as an entry price or a prior session's P&L, that later judgments adjust too little from. They can appear together but correct differently.
- q: Why is the prior session's result such a strong anchor? a: Because it is a recent, salient, real number about the trader's own money, which makes it feel like the natural starting point for the next decision. Sizing should come from the current trade's stop and a fixed risk rule, but the mind reaches for the last number on the account instead.
- q: How do I detect anchoring in my own trading? a: Record the planned size for each setup before the session and the actual size beside it, then sort the deviations by whether the prior day was green or red. Upsizes clustering after winning days and oversizes after losing days are the signature, once the sample clears twenty trades in the matching condition.
- q: What is the single most effective guard against it? a: Size each trade from its own stop distance and a fixed risk unit, and keep the running account balance out of view when choosing size. When the size is a rule applied to the trade in front of you, there is no past number left to anchor to.