Stochastic RSI (StochRSI)

The Stochastic formula applied to RSI values rather than price. Produces a much more sensitive oscillator than either parent — same 0-to-100 scale, but values cycle faster and extremes happen more often.

What it measures

• Position of RSI within its recent range. Where the current RSI reading sits between its highest and lowest values over the lookback.
• Amplified momentum signals. Because RSI's natural range is already compressed (most readings live between 30 and 70), normalizing it again produces dramatic readings.
• Earlier momentum shifts. Often crosses 80/20 thresholds before plain RSI does.

Formula

RSI[t] = standard RSI over rsi_period
StochRSI[t] = (RSI[t] − LowestRSI) / (HighestRSI − LowestRSI) × 100
%K = SMA(StochRSI, k_smoothing_period)
%D = SMA(%K, d_smoothing_period)

Inputs in AlgoLift

Recommended settings

• (14, 14, 3, 3) — default: Standard. Works for most timeframes.
• (5, 5, 3, 3): Very fast — for short-timeframe scalping.
• (14, 14, 1, 1): Unsmoothed — exposes the raw Stochastic-of-RSI value. Very noisy.

Outputs in AlgoLift

Reading the components together

StochRSI behaves like Stochastic but with amplified sensitivity:

• %K moves quickly between extremes — visits 80+ and 20− many more times than plain RSI ever would.
• %D lags %K and smooths the wildest swings — used as the reference for crossover signals.
• Histogram shows acceleration — early warning of %K/%D crossovers about to happen.

The two-line structure means the standard Stochastic strategies all port over directly. The difference is that StochRSI fires those signals more often.

Best in / worst in

Best in range-bound markets with clear cyclical behavior, on instruments where you need earlier signals than RSI provides, and as a second-opinion oscillator alongside RSI itself.

Worst in trending markets (same failure mode as Stochastic, made worse by the extra sensitivity) and as a standalone signal without filters. The extreme readings happen too often for "every overbought = sell" to be profitable.

Three setups

1. Crossover at extreme

The canonical entry — same structure as Stochastic but on a noisier signal, so the filter requirements tighten.

• Long: %K crosses above %D AND %K was < 20 within the last 2 bars AND price > EMA(50).
• Short: %K crosses below %D AND %K was > 80 within the last 2 bars AND price < EMA(50).

The trend filter (EMA(50)) is mandatory here — without it, this strategy bleeds in any persistent direction.

2. Histogram-led early entry

Use the Histogram for the very earliest momentum shift, with the crossover as confirmation.

• Long: Histogram crosses above 0 (signals momentum acceleration) AND %K still < 50 AND KAMA Efficiency Ratio < 0.3 (range regime).
• Short: Symmetric.

The combined filters — momentum-acceleration + below-midline + range regime — eliminate most false signals from StochRSI's raw sensitivity.

3. StochRSI as a pullback timer on top of RSI

Use plain RSI for the regime ("is the market in a buyable zone?") and StochRSI for entry timing ("right now?").

• Long: RSI between 40 and 60 (neutral zone, not stretched) AND StochRSI %K crosses above %D from below 20.
• Short: Symmetric thresholds.

The RSI check prevents StochRSI from firing during RSI extremes (which usually mean trend, not chop).

Advanced patterns in AlgoLift

StochRSI for scale-in timing. Take the initial entry on a slower trigger (EMA crossover, MACD signal). Use a Scaled Entry that fires when StochRSI %K bounces from < 20 — a quick re-entry signal that often catches additional impulse during a continuing trend.

StochRSI exit timer in mean-reversion trades. When entering a mean-reversion long, exit when StochRSI %K first crosses above 80 (target reached based on momentum extreme) — independent of price target. Captures the cyclical-extreme exit timing more precisely than price-based targets.

Cross-timeframe StochRSI agreement. Wire StochRSI on both the execution timeframe (5m) and a higher timeframe (1h) via separate data series. Only allow entries when both timeframes agree on direction (both rising from oversold, or both falling from overbought). The cross-timeframe filter dramatically reduces the false-signal rate.

Common mistakes

• Treating it as plain Stochastic. StochRSI fires extreme signals roughly 2–3× more often than Stochastic. The default 80/20 thresholds still hold but produce many more triggers — most of which need additional filters.
• Using it as a standalone signal. The increased sensitivity means StochRSI on its own is one of the noisiest indicators in the library. Always pair with a trend filter, regime detector, or second oscillator.
• Confusing it with RSI. StochRSI's reading and RSI's reading often disagree — when StochRSI is at 80, RSI might be at 60. The two indicators are related but not interchangeable.
• Over-tuning the four parameters. Four parameters means four dimensions to overfit. Stick with (14, 14, 3, 3) unless you have a specific thesis.