The position sizing explodes after a good run and the trader cannot explain why. Not a deliberate decision — no updated thesis, no systematic rationale, no change to the risk plan. Just a felt sense that the account can absorb more right now because the gains "aren't really mine yet." They came from the market. Playing with house money. This is a real, named cognitive error, and it operates entirely beneath the level of deliberate thinking. By the end of this article you will be able to recognize when your account has been mentally sub-divided into separate risk budgets, name the mechanism driving it, and apply a single-ledger test that forces you to treat every dollar identically regardless of where it came from.

What Mental Accounting Actually Is

Mental accounting is the tendency — documented in behavioral economics — to treat money differently depending on its perceived origin, its intended category, or the account it is held in, even when the dollar amounts are objectively identical. The concept was developed and named by Richard Thaler, who observed that people do not treat money as fungible the way rational economic models assume. Instead, people sort money into informal mental "accounts" — each with its own rules about how freely it can be spent, risked, or lost.

The mechanism is not irrational by accident. It serves a real purpose in ordinary budgeting: keeping a "vacation fund" mentally separate from everyday spending prevents you from draining it on small impulses. The problem for traders is that the same mental architecture, applied to a trading account, creates systematic and measurable errors in risk allocation.

The most important trading expression of mental accounting is the house-money effect. After a winning period, gains are placed in a separate mental sub-account — "house money," not real money — and the brain applies a far more permissive set of risk rules to that sub-account. The result is that position size, stop placement, or setup threshold loosens specifically in response to recent gains, not in response to any change in actual edge or forward probability. The trader is effectively running two risk frameworks simultaneously: a conservative one for "real money" and a permissive one for "found money."

This is distinct from loss aversion and the disposition effect, which describes the asymmetric pain of losses versus gains — a different distortion that produces the pattern of cutting winners early and holding losers too long. Mental accounting does not require a loss to be present at all. It operates when the account is ahead, producing a specific and opposite error: over-exposure, not over-caution.

Why the Brain Does This

The mental sub-division of money appears to be a form of cognitive simplification. Managing a single unified pool of capital with consistent risk rules across all conditions is genuinely demanding. Sorting recent gains into a "less serious" category reduces that cognitive load — but at the cost of accuracy.

There is also a pain-asymmetry component. If a gain is categorized as "house money," its loss does not register with the same intensity as the loss of "real" capital. This lowers the emotional cost of risky behavior and therefore lowers the behavioral inhibitor that would otherwise apply the brakes. The brain has effectively pre-discounted the loss before it occurs, making the risk feel smaller than it actually is.

A third driver is the break-even effect, another expression of mental accounting. A trader sitting at a session loss will sometimes take a large, low-probability bet specifically to return to flat — because a break-even outcome feels qualitatively different from a small loss, even when the dollar difference is negligible. The mental account for the session has a "break-even" threshold that changes the risk calculation discontinuously at that point, in a way that has no relationship to actual edge or forward probability.

None of these drivers announce themselves. They operate through what feels like natural decision-making, which is exactly what makes them persistent.

The Evidence Base

Behavioral economists Richard Thaler and Eric Johnson named the house-money effect in a 1990 paper published in Management Science (vol. 36, no. 6, pp. 643–660). Their central finding was that prior winnings create a psychological cushion: after a gain, subsequent losses tend to be mentally coded as reductions in the gain rather than as real losses from principal — as if the winnings function as a buffer that makes losing feel less painful.

In laboratory experiments with student subjects, Thaler and Johnson found that participants who had just won and saw that prior gain presented separately before a new risky choice were significantly more risk-seeking than participants offered the same mathematical gamble framed as a single integrated choice with no visible prior gain. The underlying expected values were identical. The framing of "prior gain now held separately" was sufficient to produce a large and measurable shift in risk preference — driven entirely by how the windfall was mentally categorized, not by any change in the actual odds.

The same paper documented the mirror pattern: the break-even effect. When subjects faced a prior loss and the loss was made salient, they were attracted specifically to outcomes that offered a path back to flat — even when the expected value of that path was inferior to safer alternatives. The break-even threshold in the mental account was doing real work on behavior.

These findings are laboratory results using students as subjects. What happens when the same dynamic appears in a live market with professional money?

Joshua Coval and Tyler Shumway examined exactly this in a 2005 paper in the Journal of Finance (vol. 60, no. 1, pp. 1–34). Their dataset covered proprietary floor traders on the Chicago Board of Trade — futures transactions on personal accounts across a full trading year. These were not retail investors; they were professional traders whose income depended directly on their results.

The pattern that emerged was the break-even effect in live-market form. Traders who had lost money in the morning session took measurably more risk in the afternoon than traders who had been profitable. According to the CFA Institute digest of the paper, the probability of a trader with morning losses making a price-moving trade was approximately 15% higher than that of a trader with a profitable morning.

The consequence was not contained to individual accounts. The afternoon prices set by traders carrying morning losses reversed more quickly than prices set by traders who had been profitable — the market itself bore a detectable signature of loss-driven risk-taking. The behavioral bias was not just a psychological curiosity; it was moving prices and then correcting.

What It Costs

The direct cost of the house-money effect is straightforward: position sizes in the "house money" phase are systematically larger than the trader's own risk rules warrant. If the risk framework specifies, for example, that no single position should risk more than two percent of the total account, but the trader has mentally sub-divided the account so that recent gains are governed by a different, looser standard, then the actual risk per position exceeds the intended rule — not because the rule was revised, but because part of the account has been removed from the rule's scope without conscious decision.

The compounding effect is more damaging. A trader who over-sizes during periods of recent gains tends to give back gains in clusters. The same behavioral state that inflates size also tends to loosen checklist discipline and setup thresholds — patterns that overlap with overconfidence after wins. The result is not just larger individual losses; it is a systematic pattern where the biggest losing decisions are concentrated precisely in the periods when the account had most recently built equity.

There is also a process-record cost. A position sizing decision driven by the house-money effect will typically be documented — if documented at all — with a rationale that does not name the real driver. The post-trade record will show "increased size / favorable conditions" rather than "increased size because gains felt less real." This corrupts the feedback loop. When you review that trade later, you see a plausible-sounding rationale for a process error, which prevents accurate diagnosis.

The break-even effect creates a different but related cost: it produces low-probability bets at exactly the moment when the session P&L has already indicated that something is not working. A trader who applies more risk to recover losses, rather than less, is doing the opposite of what a systematic risk framework would prescribe.

The Single-Ledger Test

The structural fix is to enforce a single ledger — one set of risk rules applied to the entire account balance at all times, with no sub-accounts, no categorical treatment of gains versus deposits, and no session-level break-even thresholds. This is not complicated. It is demanding to maintain against the cognitive pull of mental sub-division.

The single-ledger test is a pre-decision check. Before sizing any position, answer two questions in sequence:

Question one: What is my total current account balance, including all open unrealized P&L? This number — not the starting balance, not the "deposit" portion, not the "won money" portion — is the only number the risk rule applies to. The rule specifies a fixed fraction of this total. If the rule says risk two percent per position, the calculation uses the current total balance.

Question two: If this gain had come from a different source — a wire transfer from another account, a gift, any source other than recent trading wins — would my intended position size change? If the answer is yes, the size is being driven by the source of the capital rather than by the risk rule. That is mental accounting in operation. The corrective action is to apply the rule as if the capital had arrived from the neutral source, which produces the rule-compliant size.

The second question is the more important one. It creates a controlled separation between the origin of capital (irrelevant to forward risk) and the appropriate sizing of a new position (governed only by the rule and the total balance). Most traders, when they apply it honestly, find that the answer to the second question is "yes" far more often than they expected — especially after strong runs.

The pre-commitment structure here connects to a broader discipline covered in the sunk-cost framework: the open-fresh test asks whether you would open a position fresh with no history, and the single-ledger test asks whether you would size it identically with no history of how the capital arrived. Both tests share the same logic — strip the psychological history, apply the rule to the forward facts.

For the break-even effect specifically: the pre-committed rule is simpler. If the session is down, position sizing does not increase. It either holds constant or decreases, depending on whether the drawdown has crossed a pre-defined threshold. The session P&L is reported daily but does not directly govern within-session risk allocation. What governs is the total account balance and the rule.

Dynamic Sizing Done Right

A common objection is that professionals do adjust position size dynamically — and this is true. The discipline of dynamic position sizing addresses exactly when and how to scale. The critical distinction is this: legitimate dynamic sizing is driven by systematic changes to the total account balance (the math of compounding), by the strength or clarity of a specific setup, or by a formal review of whether the approach is performing as expected. None of these drivers reference where the capital came from, what the session P&L looks like, or how good a recent run has felt. Mental accounting creep masquerades as dynamic judgment but uses categorically different inputs — psychological ones.

A trader developing this discipline is not learning to size smaller. They are learning to size consistently by the rule, then to update the rule only when the systematic rationale for doing so is present and documented. The distinction between "I'm running well so I'll size up" and "the account has grown enough to justify a systematic step-up in my standard size" is precise and consequential.

Simulator Exercise: The Single-Ledger Speed Run

This drill is designed to make mental accounting visible in operation. Run a standard Speed Run in Abu Terminal. Before the session begins, record your starting bankroll in a separate note — a physical pad, a scratch file, anywhere external to the simulator. Label it "total account."

As the run progresses, maintain a running tally in that note. Every event, update two figures: the cumulative gain or loss from the starting bankroll, and the current total. Keep them clearly separated on the page — one line for "won/lost from starting point," one line for "current total."

At each decision point where you are about to size a position, stop and ask: am I about to size this based on the current total, or based on how the cumulative gain looks? If you notice any pull toward sizing relative to the gain line rather than the total line, that pull is mental accounting. Write "MA" next to that decision in your note before proceeding.

Now add the source-neutrality check. At each decision point, ask: if the current total balance had arrived as a single lump sum from an external source — not from a series of trades — would the position size change? If the answer is yes, adjust to what you would do with the neutral-source capital, then proceed.

After the run, count your MA notations. Any decision where the cumulative-gain line was pulling the size is a mental accounting event. Review those decisions specifically: were they larger than your rule warranted? Did the setup quality justify the size, or did the gain history? The goal is not zero MA events on the first run — it is to see them clearly enough that you can recognize the pull in real time before it completes.

Run the drill across multiple sessions. The pattern to watch for is directional: MA events should cluster after strong positive runs. If they are random across the session regardless of P&L shape, the error is a different one — likely a systematic sizing inconsistency rather than house-money effect specifically.

Mental Accounting: Why House Money Feels Different