Win Rate: Half of a Number
Win rate is the share of trades that made money: winners divided by total trades. It is a count, so it ignores size. Paired with the average win against the average loss it tells you whether a method earns; on its own it tells you nothing at all.
A count of how often you were right, and nothing else. The half it leaves out decides the money.
How it works
Winning trades divided by total trades. If three of ten closed in profit, the rate is three in ten. It is a count, and counts ignore magnitude.
On its own it tells you almost nothing. It counts how often you were right and not one word about how much you made — the half that decides the outcome.
Because it says nothing about how big the winners were. Someone right nine times in ten who loses more on the tenth than the nine made holds a fine rate and a shrinking account.
The half nobody quotes
It only means something paired with reward to risk — the average win against the average loss. The two together fix break-even.
At one unit of reward for one of risk, break-even before costs needs one trade in two. At two for one, one in three; at three for one, one in four. Reverse it — one of reward for two of risk — and two wins in three merely stand still.
A ninety per cent winner can lose money every month. High rates are manufactured by banking small profits and holding losses until they recover: the count rises, the average falls. Selling premium and averaging down both have that shape.
And twenty trades cannot measure it at all. Twenty is an anecdote with arithmetic attached. It cannot separate a working method from a broken one.
A genuine edge still produces long losing runs. Direction runs on this site’s shared 576-bar history average 2.01 bars with a longest of 11, across 286 runs, before any entry rule is applied. Trading psychology is mostly the study of what those runs do to a person — Mark Douglas’s whole subject.
In practice
Costs raise the rate you need before you start. A round trip here costs 0.0098 price units: 2% of a median bar’s range, 45% of the smallest, charged on winners and losers alike.
And participation changes the fills and the rate with them. Thin volume widens the spread you pay, so a rule measured in a quiet session is not the one you traded in a busy one.
A slow system takes years to produce a usable figure. A method trading twice a month gives no trustworthy sample this year, which is why backtesting exists and why overfitting is the trap inside it.
One gapped loss undoes a long run of wins. An opening gap fills your stop loss where the market opens, not where you put it, so the loss is larger than the count assumed.
Move the stop wider and the rate rises and the edge falls. Trailed by a multiple of the 14-bar average true range, the median position survived 3 bars at one average range, 10 at two, 22 at three and 32 at four across 562 trials each — and 91% to 100% were stopped out.
So the number can be gamed. Widen the stop, tighten the take profit, and the rate rises while the account worsens.
Every round trip costs 2% of a bar. Add it to the clean arithmetic and the rate you need is higher than the one you calculated.
Record both halves in units of risk
Log every result as a multiple of what you risked, not as a sum of money. A trade that lost what you were prepared to lose is minus one; one that made twice that, plus two.
That single change makes both halves comparable. A trading journal kept this way lets you read the rate and the average result side by side, at any instrument or size.
Then average the whole column. That average is expectancy — the result per trade across the whole distribution — and what the win rate stood in for. A positive average with a low rate works; a high rate with a negative average only feels like it.
Fix risk per trade as a fraction first: the units only mean the same thing if the risk did.
What a win rate is not
It is not expectancy. Expectancy weights the rate by the average outcome; the rate weights nothing.
It is not the probability of your next trade. It describes trades already closed.
It is not evidence of skill. A wide enough stop lifts the rate of a poor method.
It is not comparable between traders. Two people with the same rate can hold opposite results.
When it fails
The market did the work
In a range a mean-reversion rate looks spectacular. Fading the extremes of a trading range wins until the range ends — then one trade returns the lot.
The base rate was most of it
A higher close ten bars later occurred in 54% of 566 observations here, and 52% over one bar across 571. A rate near those is the market, not a method.
The count excluded what was still open
Breakouts above a 20-bar high closed back below within ten bars in 85% of 39 events, and above a 55-bar high in 100% of 11 events. A rule that omits open positions counts something else.
The stop moved after entry
Widening a stop mid-trade turns losses into wins often enough to flatter the count. It also turns a loss you accepted into one you did not — which risk management prevents.
The sample never grew
Twenty trades, a conclusion, a rebuild, then twenty more. Trade review is how a sample grows into something readable, and there is no shortcut.
The original data
Scanning 31,760 video titles in research/broker-coverage.json, 211 discuss win rate — median
11,527 views, 130 channels, a maximum of 4,509,636 — while expectancy returns 0 videos, risk of
ruin 0 and monte carlo 0. The half that decides whether a method earns is absent; the half that
decides how it feels is everywhere.
And research/series-measurements.json, built by site/measure_series.py, sets the bar it must
clear: 54% of ten-bar windows on this 576-bar history closed higher, across 566 observations. A
claimed 55% is one percentage point above doing nothing, and the round trip costs more than that.
So never quote a win rate without the average win and the average loss beside it.
Related
Probability is what a win rate estimates, and why it needs a large sample.
Risk per trade fixes the denominator, so results log in units of risk.
And trade review is where both halves get read together.
I chased a high win rate for years because being right felt like the point, and it quietly made me worse. I took profits early to protect the count and let losses run because closing them would spoil it, which is exactly backwards. What changed it was writing the average win next to the average loss and seeing that the pretty number had been hiding the ugly one. I still like being right; I just stopped paying for it.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.