RSI signals misled momentum traders

Momentum traders keep relearning the same lesson: a clean signal is not the same thing as a durable edge. The setup here is simple enough to fit in one line — buy when RSI says an asset is oversold and volume suddenly surges. In calm or trending markets, that can look brilliant. But the 2022 bear market was a brutal reminder that “oversold” can stay oversold for a long time, and a volume spike can mean forced selling, not a rebound. ### What is this setup actually trying to catch? RSI — the Relative Strength Index — is a momentum gauge on a 0 to 100 scale. Traders usually read below 30 as oversold and above 70 as overbought. Add a volume spike and the idea becomes more specific: maybe the selloff has climaxed, weak hands are out, and a bounce is close. That logic is why the combo is so popular in stocks and crypto. (fmz.com) ### Why did it look so smart before? Because mean reversion works beautifully in the right tape. One published RSI backtest on SPY found that after a 14-day RSI breach of 20, the ETF moved 2.67% more than usual over the next 20 days. That kind of result is catnip for traders — it feels simple, repeatable, and grounded in behavior everyone thinks they understand. (chartschool.stockcharts.com) ### So why did 2022 break it? Because 2022 was not a normal pullback regime. In bear markets, volatility jumps, correlations rise, liquidity thins, and liquidation cascades hit fast. Under those conditions, a low RSI is often not a reversal signal. It is just a measurement of strong downside momentum. The volume spike makes the trap worse — what looked like “capitulation” in a bull market can just be another leg of panic in a downtrend. (optionsamurai.com) ### Isn’t RSI supposed to warn you about extremes? Yes — but that is the catch. RSI measures momentum, not value, and not trend direction. Even mainstream RSI guides make the same point: in strong trends, RSI can stay overbought or oversold for weeks or months. That means traders who treat a threshold crossing like a hard reversal trigger are basically fighting the tape with a single indicator. (stratbase.ai) ### Where does curve-fitting enter the story? Right where the backtest starts looking too perfect. If you test RSI 14, threshold 30, volume 1.5x average, and a bullish candle filter on a handpicked period, you can absolutely find a version that sings. But each extra rule can just be another way of memorizing one market regime. The FMZ strategy write-up is unusually candid about this — it flags parameter sensitivity and market-environment dependency as core risks, not footnotes. (fidelity.com) ### What are traders changing now? The smarter shift is away from “does this backtest make money?” and toward “when does it fail?” That means isolating bear-market windows, adding slippage and liquidity stress, comparing against buy-and-hold drawdowns, and testing whether the edge survives different thresholds and different assets. Basically, traders are trying to break the strategy on purpose before the market does it for them. (fmz.com) ### Does that make RSI useless? Not at all. It just makes RSI contextual. In range-bound markets, oversold readings can still help frame good entries. In trends, RSI often works better as a strength gauge or with filters that identify regime first. The mistake is using one oscillator as if it knows the difference between a temporary washout and the start of a real unwind. (stratbase.ai) ### Bottom line? The story is not that RSI “stopped working.” It is that a setup that looked elegant in 2020–2021 turned out to be regime-specific. 2022 exposed the hidden assumption — that sharp selling would mean-revert quickly. Sometimes it does. In a real bear market, turns out, it just keeps going. (stratbase.ai) (interactivebrokers.com)