What is RSI? A practical guide for traders
The Relative Strength Index measures momentum on a 0–100 scale. Here is what it actually represents, how it is calculated, and how disciplined traders use it without overfitting.
Published · Last reviewed
The Relative Strength Index, introduced by J. Welles Wilder in his 1978 book New Concepts in Technical Trading Systems, is one of the most widely-used momentum oscillators in technical analysis. It compares the average size of recent up moves to the average size of recent down moves over a lookback window — by convention, 14 bars. The output is a number between 0 and 100. Higher values mean recent buying has outweighed recent selling; lower values mean the reverse.
The exact formula matters because there are two variants in the wild and they give different numbers on the same data. Wilder's original RSI (the one Drogo uses, and the one most professional charting platforms use) uses an exponential-style smoothing of average gain and average loss: after the first 14-bar simple average, every subsequent average is (prev_avg × 13 + new_value) / 14. The "cutler's RSI" variant uses a simple moving average throughout. The two diverge meaningfully on volatile stocks. If your screener says one thing and your charting platform says another, this is usually why.
Conventional thresholds put RSI ≤ 30 in oversold territory and ≥ 70 in overbought territory. These numbers are convention, not law — Wilder picked them because they captured roughly the bottom and top deciles of his daily-bar testing universe in the 1970s. They are reasonable defaults on liquid US equities and crypto majors. They are too tight on high-volatility small caps, where extending to 20/80 reduces false signals, and too loose on slow utilities, where 40/60 captures the actionable extremes.
The most common mistake new traders make with RSI is treating "oversold" as "buy now". Oversold means selling has been intense, not that it has stopped. In strong downtrends, RSI can stay below 30 for weeks; selling those readings as if they were reversals is one of the fastest ways to bleed. The classic Wilderian rule — buy oversold readings only in confirmed uptrends, sell overbought only in confirmed downtrends — is more robust because it conditions on a higher-timeframe trend filter that the indicator alone cannot see.
A more sophisticated use is divergence. Bullish divergence is when price makes a lower low but RSI makes a higher low; bearish divergence is the inverse. Divergences flag exhaustion of the dominant move and are most reliable on the 1-hour and daily timeframes for liquid instruments. They are extremely unreliable on 1-minute charts because microstructure noise dominates. Drogo's screener catalogue includes a daily-RSI-divergence preset that filters for this pattern at scale.
Finally, RSI is most informative when paired with structural levels. An oversold reading at a multi-month support that has been defended on multiple touches is a different setup than an oversold reading in the middle of a clean downtrend with no structure beneath it. The indicator does not see the level; you have to. The Drogo terminal's AI commentary will surface both the RSI reading and the nearest support / resistance level so you can decide whether the two agree.
For the exact numerical recipe Drogo uses (including how we handle the avgLoss = 0 edge case and how we annualise across timeframes) see the signal methodology page.
References & further reading
Ask Drogo about this:
Drogo Research — Quant editorial
נבדק לדיוק עובדתי; המתודולוגיה מקושרת למטה.