I stopped using RSI for a month. Here's what broke.

For one month earlier this year I tried trading my full system without RSI. No divergence as a filter, no overbought-oversold gate, no oscillator at all. I wanted to know what the indicator was actually doing for me. The answer turned out to be: more than I had given it credit for, in one specific category of trade I had not been thinking about clearly.

The setup of the experiment was simple. Same pairs, same setups, same risk model, same session windows. The only change was that any trade that would normally have required RSI confirmation was taken without that requirement. If the chart structure said pullback to external level with trend health, I took it, with no further filter.

The first ten days, the equity curve looked indistinguishable from any other ten-day stretch. Win rate slightly down, average winner about the same, average loser about the same. I started to think the indicator had been ornamental. Confirmation bias in fancy clothes.

Days eleven through twenty-four told a different story. There was a cluster of losing trades that had a single common feature, and I had not noticed because I had stopped looking at the screen the indicator usually lived on. Every losing trade in that stretch had been taken into a chart where price was making a higher high or lower low that, on the indicator, was not. RSI would have been showing divergence.

Divergence is not magic. It is a measurable signal that price momentum is fading even as price itself continues to push. Most of the time, divergence resolves with a continuation in the direction of price, because momentum can return. But in a specific category of trade, divergence resolves with a reversal that takes out the structural high or low and stops out anyone who took the continuation.

The trades I lost during the experiment were almost all in that category. Daily impulses that had stretched for two or three sessions, where the pullback into the next continuation level looked clean on the chart, but where the underlying momentum had been fading for the last leg. The signal was there. I had stopped looking.

So I added RSI back, and the next month closed up. Not because I started trading more, but because I started passing on a category of setup that, on average, was a coin flip dressed up as a continuation. The trades I take now are roughly the same chart conditions as before, with one extra gate: if the higher timeframe RSI is in divergence, the continuation is paused. The setup either resolves with the divergence resolving, in which case I take it, or it resolves with a structural break, in which case I was right to skip.

The general lesson is that no filter in a trading system is doing nothing. If you are using something, even something you cannot articulate the value of, before you remove it from your process you should run a controlled test for at least four weeks. Anything shorter than that and you are looking at noise. Anything longer than that and you are running a real experiment with real losses and you should be sized accordingly.

The other lesson is that the indicator is not the signal. Price is the signal. Price is what pays. Indicators are summaries of price designed to surface specific properties of price action faster than your eye can read them. RSI is a summary of momentum. MACD is a summary of trend. Bollinger Bands are a summary of volatility. They are not separate inputs to be added together. They are the same input, looked at from different angles, and the only reason to use any of them is that the angle they show you is something you have proven helps you make better decisions on the actual chart.

I trade with RSI. I do not trade with MACD or Bollinger or anything else. The other indicators do not help me see anything I could not see by reading price directly. RSI helps me see momentum exhaustion in advance, which I cannot reliably read from candles alone. That is the test. If the indicator is pulling weight, keep it. If it is wallpaper, remove it.

If you are reading this and you have ten indicators on your chart because someone on a course told you to, run the experiment. Strip them all off, trade for two weeks, then add them back one at a time and see which ones cost you trades when they are absent. Most will not. The ones that do are the ones earning their pixels.