What RSI means (and why misunderstandings happen)
Relative Strength Index (RSI) is a momentum oscillator that maps recent price changes into a bounded number, typically from 0 to 100. RSI values are often discussed using reference levels (for example, “overbought” and “oversold” regions). A common mistake is treating those regions as automatic entry/exit commands. In reality, RSI is descriptive: it summarizes recent movement, but it does not guarantee future direction.
Another misunderstanding is mixing two different roles: (1) defining how RSI will be computed and interpreted, and (2) claiming it predicts outcomes. The first part is stable and definable. The second is uncertain and depends on many variable conditions (market regime, spreads, slippage, and how trades are executed).
How RSI strategies commonly fail in practice
A frequent mistake is inconsistent interpretation. For example, people may use RSI threshold crossings one day and mean-reversion language the next day without specifying which rule set applies. When the “strategy” is not clearly defined, later results may be explained away as pattern recognition rather than a repeatable method.
A second error is ignoring the full “from chart to execution” chain. Even if RSI-based conditions are correct, costs and execution timing can change outcomes. This matters because RSI-based approaches are often sensitive to bar timing (the moment a candle closes, whether you act at the close, and how quickly orders can be filled). If execution differs from the assumptions used when reasoning about the method, results can diverge.
A third failure mode is assuming historical relationships transfer. Past RSI behavior can look similar across periods, but that does not establish that the same conditional expectations will hold in the future. Markets can shift volatility and participation, changing how momentum indicators behave.
A fourth mistake is overfitting or “tuning to the past.” Adjusting thresholds, lookback length, or confirmation rules until results look good on a specific sample can make the approach brittle. Without a neutral check for generality, the method can become a description of one period rather than a robust rule set.
Limitations and risks you should explicitly account for
One material limitation is regime dependence: RSI behavior can differ when price swings are trending versus when they are choppy. In strong trends, RSI can spend extended time in “high” or “low” regions, which can break simple interpretations that expect quick reversals.
Another limitation is that RSI only uses a measure of price changes. It does not directly account for liquidity, trading costs, or order-book dynamics. Because those factors vary, two setups with the same RSI reading can have different realized outcomes.
Finally, a key risk is confirmation bias. People often remember “good-looking” instances of RSI thresholds and discount contradictory examples. That can lead to a narrative that RSI “worked,” even if the underlying rules were not applied consistently or across comparable conditions.
Neutral checks and verification steps (without assuming prediction)
To verify an RSI strategy explanation independently, start by stating assumptions precisely:
- Define the RSI settings (lookback length, calculation method assumptions) and the exact rule for interpreting RSI (for example, what event triggers the next action).
- Specify timing assumptions: using bar closes versus intrabar values, and how you handle partial periods.
- Include cost and execution assumptions as part of the evaluation (even if simplified), and check whether results remain plausible when those assumptions change.
- Test for stability: compare performance across different time windows and different market conditions rather than a single sample.
Use a simple “ready-to-falsify” mindset: identify what must be true for the rule to be useful, and what observations would contradict it. If you cannot describe a contradiction scenario, the method may be more of a story than a rule.
If you want, share the RSI interpretation you had in mind (for example, thresholds and timing), and you can pressure-test it using the neutral checks above.