RSI-2 mean reversion, backtested

Most people meet RSI as a 14-day indicator and are told over 70 is overbought. The short-term mean-reversion version, popularised by Larry Connors, does almost the opposite: it uses a 2-day RSI, which is jumpy enough to hit single digits several times a year, and buys that washout, but only while the market is still in an uptrend. The trend filter is not a detail. Without it the same rule buys every step of a collapse.

Over September 2021 to September 2026, this setup lost to simply buying and holding. €10,000 became €12,638 (+26.4%), against €18,720 (+87.2%) for buy & hold, over 32 trades, with a worst fall of 8.2%. Change the ticker below and the picture changes. This is one stock over one stretch of history.

The rules, exactly

This is what the run above loads. Every number in it stays editable: change the ticker, widen the stop, and run it again.

Only consider a buy while the price is above its longer moving averages, the trend filter that keeps the rule out of a falling market.
Buy at the close when the 2-day RSI closes under 10.
Sell at the close when the 2-day RSI closes over 70. Holds are short, typically a few days.
0.1% commission and 0.05% slippage per side. This matters more here than on a slow system: a strategy that trades often pays the spread often.

Screens for: $10B+ · 2M+ shares a day · profitable. That filter does nothing on the single-stock run above. It is the pool this strategy is really meant to trade, and it comes into play when step 1 is switched from one ticker to the whole US stock market. The momentum screener is a free way to see a ranked pool like it today.

What to watch for

Short-hold mean reversion produces many small wins and occasional large losses, so a headline win rate flatters it. Check the average loss against the average win in the results below, not just the percentage green. It also assumes the thing you are buying comes back, which is a statement about the business as much as the chart; that is why the screen attached to this template asks for large, liquid, profitable companies. And frequent trading in a taxable account converts a gross edge into a smaller net one.

Before you trust any backtest

A backtest is a measurement of the past under assumptions you chose, and the assumptions are where results go wrong. Three that matter here: the run covers five years of daily closes, so it has seen one bull market and one bad year rather than a full cycle; fees and slippage are included but your broker’s may differ; and a strategy that fired only a handful of times has told you almost nothing, however good the numbers look. The tool reports the trade count for that reason. Read it first.

Want to change the rules rather than the numbers? The full backtester builds a setup out of plain-English conditions, and this template arrives in it already loaded and editable.

The other prebuilt strategies

Buy the dip: Buy any day that falls 2% or more, and sell ten trading days later. No filter, no stop.
Golden Cross: Hold while the 50-day average is above the 200-day; step aside when it crosses back under.
20/50 cross: The same crossover idea as the golden cross, on faster averages: 20-day over 50-day.
RSI 30/70: The textbook RSI trade: buy when the 14-day RSI drops under 30, sell when it clears 70.
MACD crossover: Hold while the MACD line is above its signal line; go flat when it crosses back under.
One-year high: Buy a close above the highest close of the past year; exit on a 50-day closing low.
Trend pullback: In a long uptrend, buy the dip that reclaims the 50-day average, a bounce off the trend.
Impulse pullback: A fresh 6/18 EMA cross marks a brand-new trend; buy its first small dip as it resumes.
Donchian breakout: Buy a 20-day high the way the Turtles did; exit on a 10-day low or the 2R target.
Bollinger reversion: Buy a close under the lower 20-day band; sell the snap back to the middle band.