Every trading YouTube channel eventually does the “mean reversion vs. trend following” video. The pitch is always the same: mean reversion wins often but gets its face ripped off occasionally; trend following loses constantly but cashes in big when a trend finally runs. Two different P&L distributions, same theoretical expectancy, pick your poison based on personality.
It’s a clean story. It’s also mostly untested — delivered as a hand-drawn curve with “not to scale, no numbers on here” written right into the narration.
So I backtested it. RSI(2) mean reversion against a 50/200 SMA crossover, both long-only, both on SPY, both from January 2000 through July 2026. Here’s where the narrative holds up, and where it doesn’t.
The Setup
""" Mean Reversion vs Trend Following — Backtest with vectorbt ----------------------------------------------------------- Two strategies run on the same instrument/period for apples-to-apples comparison of win rate, payoff ratio, and tail risk.
MEAN REVERSION : RSI(2) oversold/overbought (Connors-style)…
Buy-and-hold leads on CAGR partly because the test pays 0% on idle cash. Mean reversion sits in cash 70% of the time; credit that at the T-bill rate and the gap to 8.27% shrinks.
It would be nice if you combined these two strategies into one two strategy portfolio. Its interesting to see if these two strategies would complement each other.
