Buying After Three Down Days, Backtested
Three down days is a dip-buying rule: after three lower closes in a row above the 200-day average, buy the next open and sell after the first higher close. On seven big winners from 2015 to 2026, 61.47% of 532 trades closed higher, and the rule cleared all three test bars on four of the seven.
Buying after three down days is one of the simplest pullback rules there is. It counts closes, not distance: three lower closes in a row count the same whether the stock fell 1% or 9%.
How it works
The version tested here has four parts, all on daily bars:
- Trend filter: the close is above its 200-day simple average.
- Signal: today’s close is below yesterday’s, yesterday’s is below the day before, and that one is below the day before it. Three lower closes.
- Entry: the next day’s open, since the signal is only known after the close.
- Exit: the first close that is higher than the previous close, filled at the following open.
There is no stop and no profit target. One position at a time, the whole account per trade, 0.05% commission each way and one tick of slippage. The runs were done in TradingView’s Strategy Tester in August 2026 on the same code as every other rule in the set.
The exit is what makes it short. It sells on the very first up close, so most trades last a few days. Across all seven stocks the average hold was 3 bars, tied with the gap-down buy for the shortest of any rule tested alongside it.
A streak rule, not an oversold reading
It is a streak rule, not a size rule. That separates it from an oversold oscillator. A two-period RSI usually ends up low after three lower closes too, but it also weighs how far price fell. Connors RSI builds a whole component out of streak length for the same reason: counting days in one direction is a different measurement from counting points.
A worked example
A hypothetical first, to show the mechanics. A stock above its 200-day average closes at $50.00 on Monday, $49.60 on Tuesday, $49.10 on Wednesday and $48.70 on Thursday. Tuesday, Wednesday and Thursday are three lower closes, so the signal fires at Thursday’s close. The buy fills at Friday’s open.
Suppose Friday closes at $48.50, lower again. The position holds. Monday closes at $48.90, higher than Friday. That is the exit signal, and the sale fills at Tuesday’s open. A weekend and one more night passed with no stop behind the position.
Now a real row from the results file. On GOOGL from January 2015 to August 2026, the rule took 92 trades. 61 closed higher, which is 66.30%. The profit factor was 1.955: winners made $1.96 for every $1.00 losers lost. The average trade returned 0.44%.
Put the costs next to that. The round trip pays about 0.10% in commission before slippage. Against an average trade of 0.44%, that is 0.10 / 0.44, or close to a quarter of the result. On MSFT the average trade was 0.29%, so the same costs took about a third.
The original data
Seven stocks, January 2015 to August 2026: AAPL, MSFT, GOOGL, AMZN, NVDA, META and TSLA. Pooled together, 327 of 532 trades closed higher, which is 61.47%. Per stock, the share ran from 54.17% on MSFT to 66.30% on GOOGL.
The test called a result good only if it cleared three bars: at least 60% of trades closing higher, a profit factor of at least 1.3, and at least 30 trades. Three down days cleared all three on GOOGL, AMZN, NVDA and META. It missed on AAPL at 59.76%, short by 0.24 of a point, and on TSLA at 59.26%, short by 0.74. MSFT missed clearly at 54.17%.
Every profit factor was above 1.3, from 1.545 on META to 3.340 on NVDA. Net results ran from +21.59% on MSFT to +135.19% on NVDA. The worst drawdown was 29.59% on TSLA, and the mildest 9.60% on MSFT.
Why seven winners flatter the result
Now the part that matters most. Those seven are the biggest winners of the decade, and every one of them went up enormously over the window. This rule was not re-run on the five stocks that lagged the S&P 500. Three other dip-buying rules were: RSI(2), the five-day low and the gap-down buy. On the seven winners they cleared the bars 19 times out of 21. On INTC, PYPL, DIS, NKE and WBA they cleared them once out of 15.
There is no result here to say three down days would do better on those five. It buys the same kind of dip, in the same trend filter, with the same lack of a stop. Treat the seven-stock table as a description of seven winners, not as evidence of an edge. The full numbers are in the per-ticker results table, and the five-stock retest of the other rules is in the survivorship table.
These are past, simulated results on a hand-picked list of tickers. Simulated fills and costs are not real fills and costs, and no figure on this page is a promise of what the rule will do.
When it fails
It fails when three down days is the start of a decline rather than a pause in a rise. The rule cannot tell the difference. It buys either way, and without a stop it holds until the first up close, wherever that comes. The trend filter helps only until the close drops under the 200-day average, and a falling stock can take weeks to get there.
It fails on the gap after the buy. Every trade holds at least one night. A stock bought on Friday’s open after three down closes can open sharply lower on Monday, and nothing in the rule limits that loss. TSLA’s 29.59% worst drawdown on this rule came from the same code that drew down 9.60% on MSFT.
It fails through costs. A 3-bar average hold and an average trade under half a percent on most of the seven means the commission and slippage bill is large relative to the edge. Double the costs, which is realistic for a small account with wider spreads, and the thin results shrink toward nothing.
It fails when a near miss is treated as a pass. Apple at 59.76% and Tesla at 59.26% are close to the 60% line, and it is tempting to call them good enough. The line is arbitrary, but moving it after seeing the results is how a test turns into a story.
And it fails as a headline number. A share of winning trades above 60% says nothing on its own about money made, which the win rate page explains. MSFT’s 54.17% came with a 1.576 profit factor, while other rules in the same test won more often and lost money.
Related
Pullback covers what a dip inside an uptrend is and how it differs from a reversal, which is the one judgment this rule cannot make. Mean reversion is the bet underneath it, and explains why timeframe decides whether it pays. And drawdown explains the worst-case figures quoted here, which are the numbers to read before any share of winners.
Before trading a streak rule, count how often three lower closes in a row actually happen on the stock in question, and what came next. The count takes minutes, and it tells you whether a sample exists at all.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.