The Five-Day Low Pullback, Backtested
The five-day low pullback buys a close below the prior five-day low while the stock is above its 200-day average, and sells on a close above the prior five-day high. Backtested on 12 stocks from 2015 to 2026, most trades closed higher on almost every stock, but the money made did not follow.
The five-day low pullback is a pullback rule built from recent highs and lows rather than an oscillator. It buys weakness inside an uptrend and waits for price to reach the top of its recent range before selling.
How it works
The version tested here, on daily bars, in full:
- Trend filter: the close is above its 200-day simple average.
- Signal: the close is below the lowest low of the previous five days, not counting today.
- Entry: the next day’s open.
- Exit: the first close above the highest high of the previous five days, filled at the next open.
No stop, no profit target. One position at a time, the whole account per trade, 0.05% commission each way plus one tick of slippage. It ran in TradingView’s Strategy Tester in August 2026, on the same code and costs as every other rule in the set.
Why the exit waits for a five-day high
The exit is the interesting part. RSI(2) and three down days sell on the first sign of a bounce. This rule waits for a close above the top of the recent five-day range, which is a full swing away. That makes trades longer: across the seven winners the average hold was 10 to 14 bars, against 3 to 6 for the faster rules. Longer holds mean bigger winners when the stock recovers, and bigger losers when it keeps falling, because nothing cuts the trade in between.
The five-day window is a short-term swing high and low. The rule treats the low of that window as a line the stock should not close under in a healthy uptrend, and buys the moment it does.
A worked example
A hypothetical first. Over the last five days a stock’s daily lows were $101.20, $100.80, $101.50, $100.90 and $101.00. The lowest is $100.80. Today it closes at $100.50, below that line, while still above its 200-day average. The buy fills at tomorrow’s open.
For the exit, take the five daily highs before any given day. Suppose the highest is $103.40. The position holds until a close above $103.40, and the next day’s open sells it. From a $100.50 signal that is about 2.9% away, which is why these trades take two weeks rather than two days.
Now a real row, and the most useful one in the file: Disney. From January 2015 to August 2026 the rule took 46 trades on DIS. 26 closed higher, which is 56.52%. More winners than losers. The profit factor was 0.586, and the account finished down 32.17%.
How most trades won and Disney still lost
How can most trades win and the account lose a third? Profit factor is total won divided by total lost. With 26 winners and 20 losers, a profit factor of 0.586 means the average winner was 0.586 x 20 / 26, or about 0.45 times the size of the average loser. Winners were less than half the size of losers. That is the whole story of the rule on a stock that did not go up: small recoveries, large continued falls, and no stop to limit them. The expectancy page works through the same arithmetic.
The original data
First run: seven stocks, AAPL, MSFT, GOOGL, AMZN, NVDA, META and TSLA, January 2015 to August 2026. Pooled, 279 of 417 trades closed higher, which is 66.91%. The test set three bars: at least 60% of trades closing higher, a profit factor of at least 1.3, and at least 30 trades. The rule cleared all three on six of seven. It failed only on Apple, where 56.36% closed higher at a profit factor of 1.023 for a net +3.55% in eleven years.
The best rows are the ones that sell rules. NVDA returned +385.79%, the largest result anywhere in the seven-stock test, with 77.05% of 61 trades closing higher. META returned +198.62%, MSFT +161.72%.
Second run: the same code on five large caps that lagged the S&P 500, INTC, PYPL, DIS, NKE and WBA, over the same window. Pooled, 111 of 178 trades still closed higher, which is 62.36%. The share of winners fell by less than five points. The results fell much further: DIS -32.17%, WBA +0.17% on just 5 trades, NKE +3.07%, INTC +13.16%. Walgreens left the market on 27 Aug 2025 after a buyout, so it has fewer years of data than the rest.
One of the five did clear all three bars: PYPL, with 62.50% of 32 trades closing higher, a profit factor of 2.368 and +60.30%. Across the three rules retested on those five stocks, that was the only pass in 15 runs, against 19 passes in 21 runs on the seven winners.
So say it plainly. The six-of-seven result mostly measured seven stocks that went up for a decade. The retest list was chosen because it trailed the market, so it is biased too, and an ordinary stock would likely land somewhere between. Both tables are published: the per-ticker results and the survivorship retest. These are simulated past results on a small, chosen set of tickers, and none of them is a forecast.
When it fails
It fails when the stock is falling slowly. A decline made of lower lows keeps closing under the five-day low, so the rule keeps buying, and the exit, a close above the five-day high, keeps not arriving. That is exactly the Disney pattern: 46 trades, a 44.59% worst drawdown, and a loss.
It fails in size, not in count. The share of winning trades stayed above 56% on every stock with more than a handful of trades. A trader watching only that number would never see the problem. The profit factor, 0.586 on Disney and 1.027 on Nike, is where it shows. The win rate page explains why the first number cannot stand alone.
It fails on volatile stocks even when they rise. TSLA cleared the test with a +57.94% result, but its worst drawdown was 53.98%. A rule that halves the account on the way to a modest gain is hard to hold in real life, whatever the table says.
It fails through the long exit. Ten to fourteen bars in a trade means ten to fourteen nights of gap exposure with no stop. The two weeks that make the winners larger are the same two weeks that make the losers larger.
And it fails as a stock-picking tool after the fact. NVDA’s +385.79% is real, but choosing NVDA after seeing it is hindsight. Apple, just as famous a winner, returned +3.55% on the same code.
Related
Pullback explains the dip this rule is trying to buy, and how a pullback differs from the start of a decline. Expectancy is the arithmetic behind the Disney result, where winning more often still lost money. And swing highs and lows covers the structure the five-day window is a small version of.
Read the profit factor before the share of winners on any dip-buying rule. This one kept a majority of winning trades on every stock with a real sample, including the one where it lost a third of the account.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.