What the StochRSI is and how it differs from the RSI

The StochRSI is almost always introduced as "a more sensitive RSI", and that description falls short to the point of being misleading. It is not a tuned RSI: it is an indicator computed on another indicator, and understanding that difference explains at once why it fires so often and why so many of those signals fail.

Where it comes from

The RSI is computed on price: it compares the strength of advances against declines and returns a number between 0 and 100. The StochRSI never looks at price again. It takes the RSI series and applies the stochastic oscillator formula to it:

StochRSI = (current RSI − lowest RSI in the period) / (highest RSI in the period − lowest RSI in the period)

In one sentence: it measures where the RSI sits inside its own recent range. If the RSI has spent fourteen candles between 45 and 55 and now prints 55, the StochRSI reads 100 — its absolute maximum — even though an RSI of 55 is as ordinary as it gets.

Who invented it and why

The StochRSI did not come from an attempt to improve the RSI in general, but from one very specific complaint. Tushar Chande and Stanley Kroll introduced it in 1994 in their book The New Technical Trader, and the problem they stated was this: the RSI hardly ever reaches its extremes. It can spend months between 30 and 70, and through those months it gives none of the signals the textbook promises.

Their solution was to rescale the RSI within its own range so that there would always be extremes to read. Put that way, the whole thing looks different: the StochRSI saturates because it was designed to saturate. That is not an implementation flaw or a parameter problem; it is precisely what it was asked to do.

Which makes clear what the right question is when using it. Not "is it at an extreme?", because it almost always is. Rather "is an extreme that shows up every few days worth acting on?", and the answer depends entirely on whether the market is ranging or trending.

Why it saturates so much

The mechanism is arithmetic, not opinion. Normalising a range onto a 0-100 scale discards the width of that range: whether the RSI moved twenty points or two, the result uses the full scale either way.

Since the RSI is already a smoothed average, its range over fourteen candles is usually narrow. Stretching it to 0-100 turns any minor swing into an extreme. In the diagram above, over the same stretch of price, the RSI never once reaches its 70 and 30 thresholds while the StochRSI hits its own thirty-two times.

Two practical consequences follow. First, the StochRSI spends most of its time pinned at 0 or 100, not in the middle. Second, in a strong trend it sticks at the top for weeks, and every one of those days looks like a sell signal.

RSI and StochRSI side by side

Where it does help

Sensitivity is not a flaw in itself; it is a flaw when used where it does not belong. The StochRSI earns its place in two specific situations.

In a sideways range. When price has spent weeks between two levels and the RSI sits dead around 50, the StochRSI still separates the extremes of each swing. That is the situation its design was built for.

As an early warning, never as a trigger. Turning before the RSI, it flags that something may be changing. The sensible way to use it is to require the RSI or the MACD to confirm afterwards; the StochRSI raises its hand and something else decides.

Where not to use it

In a strong trend. Its worst case. It pins at 100 and stays there, and anyone reading that as overbought will be selling into the trend over and over. The same problem a saturated RSI has, multiplied, and the reason trends call for the MACD rather than the RSI and not for an even more tightly bounded oscillator.

For hunting divergences. Divergences need a series with room to travel, and the StochRSI lives against its caps. A divergence on a saturated indicator is not information.

On very low timeframes. At 5 or 15 minutes noise dominates and the StochRSI amplifies it twice over: first the RSI smooths price, then normalisation stretches what little is left.

The parameters and the trap of changing them

The usual setup is 14 for the RSI, 14 for the stochastic window and 3 for the %K and %D averages. When someone finds it fires too often, the typical reaction is to lengthen the periods until it quietens down.

The trouble is that this fixes nothing: it cuts false signals and good ones in the same proportion, because the noise lives in the double smoothing rather than in the parameters. And it takes you off the calculation everyone else is watching, which in technical analysis is much of why levels work at all.

StochRSI on each asset

On Bitcoin, with long cycles and sustained trends, the StochRSI adds little outside accumulation phases; there it does help mark the edges of the range. On Ethereum it is worth crossing with the ETH/BTC ratio, because a high StochRSI on ETH with a flat ratio usually means market drag rather than the asset's own strength. On XRP it behaves worst: its moves arrive as abrupt impulses after flat weeks, exactly the pattern that leaves the StochRSI stuck at one extreme.

Frequently asked questions