The worst stretch of trading I ever had came directly after the best month I ever had. For years I filed that under bad luck, and it took embarrassingly long to accept that the two were connected. I was up big, I had quietly started treating the gains as chips that belonged to the table rather than money that belonged to me, and I sized and behaved accordingly until the market collected. Almost every trader I know has a version of this story, and the mechanism behind it is old enough to predate every market most of us trade.
The term comes from casinos. A gambler who is up for the night starts betting bigger and looser, because in his head he is now betting the casino's money rather than his own. Richard Thaler and Eric Johnson took that piece of casino folklore into the lab decades ago and found it held up with ordinary subjects and small stakes. People who had just experienced a gain became measurably more willing to gamble with it than people risking the same amount of their own starting money. The gain gets mentally tagged as belonging to someone else, so losing it never quite registers as a real loss.
The accounting error underneath it
Behavioral economists call the underlying bug mental accounting. Money is fungible, in the sense that a dollar of profit and a dollar of deposit buy the same groceries, hedge the same risks, and compound at the same rate, but your brain does not file them in the same place. Deposits sit in the account labeled my money, guarded by loss aversion and every cautious instinct you have. Recent profits sit in a separate account labeled winnings, and that account runs on looser rules. Losing winnings feels like giving something back, a much softer emotion than watching your original stake shrink. Traders are unusually exposed to this because the interface shows you both buckets side by side all day. Realized gains for the month sit right there on the screen, visually separated from your starting balance, practically begging to be treated as a different kind of money.
The result is a pattern anyone who has looked at enough equity curves will recognize. A trader has a great month, then round-trips the whole thing over the following weeks, and the giveback is rarely one bad trade. It is a sequence of slightly larger, slightly sloppier trades that each felt fine at the time.
What it looks like from inside
From inside, house money behavior feels like confidence, which is what makes it hard to catch. Size creep is the classic version. You risked one unit per trade all the way up, and now that you are up twenty units, risking two feels conservative even though nothing about your edge changed. Setup drift follows close behind, because trades you would never take with your own money become acceptable with the market's money, so the B-grade setups start getting filled alongside the A-grade ones. Stops drift wider for the same reason, since a bigger cushion makes any single loss feel survivable. And trade frequency climbs, because when losing feels cheap, the bar for acting drops.
There is a second, quieter problem stacked on top of the psychology. Winning streaks are not evenly distributed through time. They cluster in regimes that suit your style, trending markets if you trade momentum, rangebound chop if you fade extremes. So your maximum confidence and your maximum size tend to arrive late in the regime that produced them, right around when the conditions that made you look smart are getting old. The house money effect makes you press hardest exactly when the environment is most likely to turn, which is a big part of why the giveback after a hot streak is so often fast and so often total.
The sweep and the cap
I have never met anyone who fixed this with awareness alone. Knowing about the bias does roughly nothing in real time, because in real time it just feels like a good run that you would be foolish not to press. What has actually worked, for me and for the traders I have compared notes with, is removing the decision entirely. Two mechanical rules:
- The profit sweep. At a fixed interval, and monthly is the obvious choice, move a fixed percentage of net new gains out of the trading account entirely. By out I mean somewhere that takes real friction and at least a day to reverse, a bank account or a cold wallet, not a sub-account on the same exchange and not a stablecoin balance sitting next to your open positions. The exact percentage matters less than the consistency. A third of net gains is a reasonable default, half if you know yourself to be aggressive. If the month was flat or down, nothing moves.
- The size cap. Position size gets calculated from your account balance at the start of the month, not from the running balance. Whatever you make mid-month does not change your unit size until the calendar rolls over and the sweep has happened first. And if you want to raise risk per trade beyond what the new balance implies, cap the increase to one fixed step at a time, then hold that size for a full month before stepping again, regardless of how the first week at the new size goes.
The sweep works because it physically relabels the money. Once profits land in the boring account, your brain files them under mine, and the protective instincts you already own start covering them. You are recruiting the bias to defend the gains instead of trying to out-discipline it, which is a much easier fight. The cap works because a winning streak is almost never statistically meaningful at the timescale where it feels meaningful. Ten good trades tell you very little about whether your edge grew, but they tell your nervous system a great deal, and the cap makes sure only the calendar gets to act on that feeling.
The failure modes are predictable, so plan for them up front. The most common is sweeping into an account you can refill in five minutes, then wiring the money back after two losing trades in the new month because you suddenly need the buffer. Distance and delay are the entire point, and if the reversal is easy the sweep is decoration. The second is deferral, telling yourself you will start sweeping once the account is bigger. The balance at which sweeping starts to feel comfortable does not exist, and the people who most need the rule are exactly the ones running hot right now. The third is quietly recalculating size off the running balance mid-month after a big win, which is the whole bias sneaking back in through a spreadsheet cell.
One thing that makes the pattern easier to see is reviewing performance in discrete monthly buckets rather than as a single running equity number. A running total hides round trips, while a month-by-month table makes them uncomfortable in a useful way. It is part of why the performance scorecards in Blockcircle break results out the way they do, though an ordinary spreadsheet does the job fine.
None of this makes a good month less good. You keep most of the gains, you keep your full edge, and you give up a little compounding speed in exchange for not handing your best month back to the market. Pick the sweep percentage now, while you are calm, put the transfer on a calendar reminder for the first of the month, and let the rule do the deciding on the days you are least equipped to.