forex-scalping.com

Indicators applied to other indicators

Most platforms let you apply an indicator to another indicator's values instead of price. A moving average of RSI, Bollinger bands on RSI, a moving average of MACD that isn't the standard signal line. It's a common way to smooth noisy indicators.

There's some sense to it. RSI on M5 jumps around a lot. A short moving average of RSI gives a smoother line that's easier to read. Crossovers between RSI and its average are used as signals by some traders.

The problem is lag. Every layer of smoothing adds delay. RSI already lags price a little. A moving average of RSI lags more. By the time the smoothed line turns, price has often moved a fair way.

There's also the risk of building something that looks clever but just repeats price action with delay. A moving average of an oscillator of a moving average is still just price, processed and delayed.

If you want to try it, I'd compare the smoothed version with the raw indicator directly. Does the smoothing filter out signals that would have been losers, or does it just delay the good ones? Testing it on your data answers that quickly.

Often, a slightly longer setting on the original indicator does the same job with less complexity.

Fairman wrote: Sun Oct 18, 2026 8:14 am Indicators applied to other indicators

Most platforms let you apply an indicator to another indicator's values instead of price. A moving average of RSI, Bollinger bands on RSI, a moving average of MACD that isn't the standard signal line. It's a common way to smooth noisy indicators.

There's some sense to it. RSI on M5 jumps around a lot. A short moving average of RSI gives a smoother line that's easier to read. Crossovers between RSI and its average are used as signals by some traders.

The problem is lag. Every layer of smoothing adds delay. RSI already lags price a little. A moving average of RSI lags more. By the time the smoothed line turns, price has often moved a fair way.

There's also the risk of building something that looks clever but just repeats price action with delay. A moving average of an oscillator of a moving average is still just price, processed and delayed.

If you want to try it, I'd compare the smoothed version with the raw indicator directly. Does the smoothing filter out signals that would have been losers, or does it just delay the good ones? Testing it on your data answers that quickly.

Often, a slightly longer setting on the original indicator does the same job with less complexity.

Hello Fairman,

You’ve hit on one of the most tempting technical analysis traps—and occasionally, one of its best-kept secrets. Applying indicators to indicators creates what quants call "derivatives of derivatives." While it is remarkably easy to build redundant, lagging systems this way, there are specific, institutional-grade setups where secondary processing provides a distinct mathematical edge.

The core issue you highlighted—lag—is the ultimate enemy of the retail trader. Every time you run price through a moving average or an oscillator, you introduce a phase shift. If you apply a 10-period moving average to a 14-period RSI, you are looking at data that is effectively reacting to price action from several candles ago. By the time this "smoothed" line crosses a threshold, the optimal entry point has already vanished.

However, there is a fundamental difference between smoothing an indicator and normalizing it.

When "Indicators on Indicators" Actually Work

Instead of using secondary indicators just to smooth out noise (which causes lag), the real value comes from using them to measure the relative extremes of the primary indicator.

Bollinger Bands on RSI: This is a classic volatility-momentum hybrid. A standard 70/30 overbought/oversold level on an RSI is static; it fails in strong trends because the RSI can peg at 85 for weeks. By applying Bollinger Bands directly to the RSI, you create dynamic overbought/oversold levels based on the recent volatility of momentum. A signal only fires when momentum stretches beyond its own recent statistical norm.

Stochastic RSI (StochRSI): This is literally an oscillator applied to an oscillator. Instead of measuring where price is relative to its high/low range, it measures where the RSI is relative to its own high/low range over a set period. It is vastly more sensitive than a standard RSI and is designed specifically to solve the lag problem you mentioned, though it introduces hyper-sensitivity.

Zero-Lag MAs on Oscillators: If you must smooth an indicator, standard SMAs or EMAs are the wrong tools. Applying a Hull Moving Average (HMA) or a Zero-Lag EMA (ZLEMA) to an RSI or MACD strips out the jagged noise without introducing the severe phase delay of a standard average.

The Multicollinearity Trap

The risk you mentioned about building something that "looks clever but just repeats price action" is known in statistics as multicollinearity. Traders often build systems with a MACD (based on moving averages), an RSI smoothed by an MA, and a Stochastic. Because all of these are derived from the exact same closing prices and processed through similar averaging formulas, they will all fire signals at the exact same time. The trader thinks they have "confluence" from three independent sources, but they just have the same delayed signal printed in three different colors.

Your advice to overlay the smoothed indicator directly on top of the raw indicator is the best way to test this. If the smoothed version is just shifting the raw signals three candles to the right, it belongs in the trash. The only time an indicator-on-indicator setup is worth keeping is if it actively filters out sideways chop without delaying the entry on a genuine breakout.

Stochastic RSI (StochRSI) was developed by Tushar Chande and Stanley Kroll to solve a specific, frustrating problem: standard RSI often spends weeks oscillating tightly between 40 and 60 without ever reaching the traditional 70 (overbought) or 30 (oversold) trigger levels.

To fix this, Chande and Kroll took the standard min-max normalization formula used in the Stochastic Oscillator and applied it directly to RSI values instead of price.

How It Normalizes Momentum

Standard RSI normalizes the momentum of price. If an asset inches up slowly but consistently every day, the RSI might settle at 55. It feels "neutral," even though the trend is persistently bullish. Because it calculates the average magnitude of gains versus losses, it requires massive, volatile price spikes to push the indicator to the 80 or 90 level.

StochRSI normalizes the RSI itself. It asks: "Where is the current RSI relative to its own high-low range over the lookback period?"

If the standard RSI has been fluctuating tightly between 50 and 55 over the last 14 periods, and today price edges up enough to push the RSI to 55, the standard RSI simply reads 55—nowhere near traditional overbought levels. However, the StochRSI recognizes that 55 is the absolute maximum of the recent range.

In MT4 and MT5, you can apply an indicator to the previous indicator's data or the first indicator's data in the settings. It's easy to apply it to the wrong one by mistake, especially with several indicators in the same window. Checking which data source is selected avoids confusing results. If the smoothed line looks odd, that's usually the first thing to check. In code, the iMAOnArray function in MT4 or buffers in MT5 do the same thing. Labelling each indicator clearly in its settings also helps when you come back to a chart after a few weeks. Without labels, it's easy to forget which line is smoothed and which is raw.

Smoothed RSI is effectively close to a longer RSI in many cases, something to know before getting attached to it. A 14 period RSI smoothed with a 5 period average often looks similar to an RSI with a longer period, just with slightly different turning points. If the results of a strategy are similar with either, I'd pick the simpler one. Fewer components means fewer settings to tune, and less chance of fitting the past by accident. If the smoothed version isn't clearly better in testing, the plain longer setting is easier to explain and maintain. It's also one less thing to break when you move to a different platform.

Agreed on preferring the simpler version. A quick test is to plot both on the same chart and count how often their turning points differ by more than two bars. If it is rare, the smoothing adds settings without adding information. The one case where smoothing earns its place is when a strategy is very sensitive to false signals, and a few bars of lag are cheaper than the extra trades. Comparing the number of signals and the average result per signal for both settings shows which side of that trade off a strategy is on.

FTtrader, Bollinger Bands on RSI is a good example of the useful kind of indicator stacking, since it makes the threshold adapt rather than adding another smoothing layer. A fair test is to compare it directly with the plain 70/30 version on the same signals. If the adaptive bands mainly remove signals in strong trends, that is useful. If they simply shift entries later, the extra complexity may not pay for itself.