The One White Soldier is a two-candle pattern taught as a bullish reversal: sellers push a stock down, then a single strong green candle takes the whole of that day back. It is easy to spot and it fires often, which is rare enough among candlestick patterns to make it worth testing properly.
So we did. Across 22 US stocks and sector ETFs, over the window the free backtester covers, it fired 595 times. Buying the close of the green candle and holding ten sessions, 313 of those 595 trades finished green: a 52.6% win rate. The median name lost 11.4% over the period, with commission and slippage on, while sitting in the market about 28% of the time.
Why 52.6% is not the good news it looks like
A bullish pattern that wins more than half the time sounds like it works. It only sounds that way because 50% is the wrong thing to compare it to.
The window was mostly a bull market. Most ten-day holds finished green whatever day you picked, so the real bar is the rate at which a randomly timed hold on these same names over these same dates finished green. We measured it:
| Ten-day holds starting from | Windows | Finished green |
|---|---|---|
| any bar | 24,948 | 55.4% |
| any green bar after a red one | 6,042 | 56.0% |
| a One White Soldier | 595 | 52.6% |
The soldier is below both. The pattern's own conditions, the long body, the missing wicks, the close above where the decline began, took a 56.0% starting point down to 52.6%.
Two honest qualifications on that. The gap is 3.4 points and on 595 trades it is not clearly distinguishable from chance (p is about 0.10), so this is not evidence the pattern is actively harmful. And it is one window on one universe. What the test does support is the plainer claim: there is no bullish edge here. Buying a soldier was, at best, the same as buying on any day, and the middle row shows the crude green-after-red flip already gave you everything the full pattern did.
Its bearish mirror, and what the pair cannot settle
We ran the identical test on One Black Crow, the bearish opposite, on the same 22 names over the same window:
| Signals | Win rate buying it | Its matched baseline | |
|---|---|---|---|
| One White Soldier (bullish) | 595 | 52.6% | 56.0% |
| One Black Crow (bearish) | 619 | 50.6% | 56.0% |
An earlier version of this page leaned hard on that 2.0-point gap, arguing that two patterns pointing in opposite directions producing nearly the same number proved neither carried direction. The gap is real but the argument was overstated: on samples this size its 95% interval runs from about 3.6 points the wrong way to 7.7 points the right way. A test that would have missed a genuine five-point edge cannot be used to declare there is none.
The column that does the work is the one on the right. Both patterns came in under their own baselines, and the crow came in 5.5 points under, which on 619 trades is unlikely to be luck. So the pair does not show two directionless signals. It shows one signal with no measurable bullish content, beside its opposite, which does have measurable bearish content. That is a more useful result than the symmetry we first published, and it is the reverse of it.
What is a One White Soldier
Two candles:
- The first is long and bearish, continuing a downtrend.
- The second has a long bullish body, with little or no wick at either end.
- The second opens inside the first candle's body and closes above the first candle's open.
The story is a session in which sellers set the opening price and buyers took the entire day off them, closing above where the decline began.
There is a genuine disagreement about what family it belongs to, and it is worth knowing rather than settling. Most candlestick literature, and this article as originally written, calls it a reversal. Our own signals surface classifies it as a continuation, a strong directional bar that tends to extend the move it is in. Our test does not settle it either way, because both readings expect the following ten days to be strong, and they were not.
Where the eleven percent actually went
The median name lost 11.4% running this rule, and most of that is not the signal.
Re-run with commission and slippage set to zero and the same 595 trades lose 3.9%. The rule trades about 27 times per name, and at roughly 0.3% a round trip that friction accounts for around eight of the eleven points. The rest is the pattern earning nothing across 28% exposure, in a stretch where the same 22 names gained a median 57.3% just by being held.
That is the practical finding, and it is more damning than a bad win rate. The problem is not that the soldier loses. It is that it does not win enough to pay for the act of trading it.
Where the pattern might still earn attention
Nothing below is in the backtest. The test takes every signal the detector finds, deliberately, because that is the pattern without a human choosing the good ones. If you use it at all, these are the conditions that separate the ones worth a second look:
- At a level that has already held. A strong green day off a price buyers defended before is a different proposition from one in open air.
- After capitulation rather than drift. The pattern describes a decisive handover. Three quiet red days followed by a green one is not that.
- On volume above the days around it. A wide body on thin volume is a gap in the order book, not a change of mind.
- With the market not falling apart. Check the market meter first, because in a broad risk-off week most single-stock reversals fail together.
Whether that filtering actually works is testable, and testing it is more useful than trusting it.
Test it yourself
The backtester carries this pattern as a rule. In the free backtester, open Chart setup, add Candlestick pattern, and set it to One white soldier. Pick a ticker, choose a hold period, and run it.
The experiment worth running is the filtered one: stack one of the conditions above on top and see whether the win rate improves or whether you have only reduced the number of trades. Compare it against holding the same ticker over the same span, which the tool draws beside your curve, rather than against 50%. With 595 raw signals there is enough here to find out, which is not true of the rarer patterns like the morning star.
Scanning for them live
Testing what a pattern did and being told when one forms are different jobs. TC2000 has candlestick formations in its scan library and can alert on them, so no formula writing is involved. Finviz carries candlestick patterns in its free screener, which is enough to see today's list without paying for anything.
Neither will tell you what this page just did, which is whether the list is worth acting on.
Frequently asked questions
Is the One White Soldier a reliable bullish signal? Not on our test. Buying 595 of them across 22 US stocks and holding ten days won 52.6% of the time, against 56.0% for any green day after a red day on the same names and dates. It came in below the rate you would have got by picking days at random, which is the opposite of an edge.
Does a 52.6% win rate not beat a coin flip? It beats a fair coin, and that is not the test. The market's own coin was weighted during this window: 55.4% of all ten-day holds finished green. Any win rate has to clear the base rate for its period, and this one does not. Comparing to 50% is how a useless signal gets promoted.
What is the difference between One White Soldier and Three White Soldiers? Count. One White Soldier is a single strong green candle after a red one. Three White Soldiers is three consecutive long green candles, each closing higher, which is a much rarer and more demanding formation.
Is One White Soldier a reversal or a continuation pattern? The sources disagree. Most candlestick literature files it as a bullish reversal; our own signals surface files it as a continuation. Our data does not settle it, because both readings predict strength over the next ten days and neither got it.
What is the opposite of a One White Soldier? One Black Crow: a long red candle after a green one, closing below where the advance started. We tested it identically, and it did rather better at its job than this one does at its own: buying into a crow won 50.6%, a full 5.5 points below its baseline.
How often does the pattern appear? Often, which is unusual for a candlestick pattern. 595 signals across 22 names in four and a half years is roughly 27 per stock, against about three for an evening star. That frequency is what makes it testable at all, and it is also what makes it expensive to trade.
Before you trade this pattern
The figures here come from one run over 22 large-cap US stocks and sector ETFs, in a window that was mostly a bull market, with the rule invested about 28% of the time. Win rates are measured before costs; the total-return figures include commission and slippage, and are stated both ways above. The script that produced them, including the --baseline flag behind the 55.4% and 56.0% figures, is in the repository (backend/scripts/research/candle_study.py) so the run can be repeated rather than trusted. The tool will show you today's numbers on any ticker you choose, and they will not match these exactly. Past prices are not a forecast, and nothing here is investment advice.