Chart patterns and candlestick setups
Do three shrinking red candles signal a bottom?
Rejected Pre-registered and tested, July 2026
The only chart pattern that beat all its own control groups — and it still could not clear the error bars.
What the pattern claims
In an established uptrend, three red bars appear within five sessions. Each red body is smaller than the last, volume falls away with each one, and price finishes close to the 50-day moving average. The claim is that selling pressure is visibly exhausting itself against support.
How we tested it
3,025 occurrences on S&P 500 companies — comfortably enough to detect a real effect if one existed.
Every study here was pre-registered: the exact rule, the pass and fail thresholds and the data window were written down and cryptographically fingerprinted before the test was run. That makes it impossible to move the goalposts after seeing the answer.
Results assume you buy at the next morning's open, not at the closing price that triggered the signal, and they include companies that were later delisted or went bust. Returns are measured against what the rest of the market did on the same days, so a rule that made money only because it fired on days everything rose scores zero here.
What happened
Over the full period, +0.19% at 60 days. In the second era we tested, −0.68%, with a median of −2.25%.
This one deserves more credit than the other thirteen. We test every pattern against a battery of placebos — versions with one condition removed at a time — and this is the only retail chart pattern that beat all four of its placebos, in every era, and the unconditional benchmark too. The shape of the result was right. Every condition pointed the correct way.
But no reading reaches statistical significance. Not the full period, not either era, not the market-adjusted version.
What it means
This is what an honest near-miss looks like, and it is worth understanding why it still fails.
The occurrences cluster: these setups fire on the same handful of calendar days across many companies at once. Three thousand occurrences on four hundred distinct days is not three thousand independent pieces of evidence, and once the error bars reflect that, a +0.19% average is comfortably inside the range you would get from nothing at all.
Beating your own placebos is necessary but not sufficient. If you take one thing from this family of studies, take this: the number of trades is not the sample size. The number of independent days is.
A rejection is not "this never works"
It means: tested on this universe, over this period, against a bar written down in advance, it did not clear the bar. Where a test lacked the power to decide either way we record it as undecided rather than rejected. The full research ledger has all 192 experiments and what the failures have in common.
Other patterns we tested
- Does the bull flag pattern work? The pole is the only live part of the pattern. Every flag condition made the result worse.
- Does MACD divergence work? The loose version changed the outcome by less than one hundredth of a percent. The strict version widened both tails.
- Does breakout and retest work? The setups the rule throws away outperformed the ones it keeps.
- Does buying the dip after a spike work? The first pattern in this family to lose money significantly rather than merely do nothing.
- Does a big RSI drop mark a buying opportunity? The family's only statistically significant result, and it vanished completely once we accounted for when it fired.
- Does a tight base after a sharp drop work? Removing the base conditions improved the result. The base was the problem.