I Ran Bollinger Bands and RSI for Three Years. Then I Deleted Both.
Tuesday afternoon. Pulled both indicators off my charts. Haven't put them back since.
Not because they're broken. Because after three years of logging every single signal they gave me, I finally figured out what they actually measure — and it was never what I was trading.
Everyone says learn these two first. Add structure later, once you've paid your dues. I did it backwards. Turns out the order was the whole lesson.
Year One: I Wasn't Testing Them, I Was Rooting For Them
Let me be honest about where I started. I wasn't neutral. Not even close.
I wanted the standard toolkit to work. Had zero interest in debunking it. I wanted a system I could explain to anyone in one sentence.
So I built a spreadsheet. Every signal got a row. Instrument, timeframe, session, what RSI said, what the bands said, what D1 structure said, what price actually did after. Gold mostly — that's what I trade. A couple other pairs when gold was dead.
Textbook setups only. RSI divergence at a swing high or low. RSI crossing back out of an extreme. The Bollinger squeeze — bands pinch tight, you wait for expansion. Band touches as mean-reversion entries inside a range.
Year one looked great.
That's exactly why it took me three years to see the problem.
Almost everything works in a trending market. Gold's in a clean D1 uptrend, you buy every RSI dip, you look like a genius for months. The indicator isn't making you money in that stretch. The trend is. Any entry works when the market only moves one direction. That's the part I kept refusing to see.
I was measuring the trend and crediting the indicator. Cost me real money before I understood what I was actually looking at.
Year Two: The Trade That Ended My Faith in Signals
One position from that stretch. I still think about it more than any winner.
Gold, clean D1 uptrend. Price pulls back. On the H4, RSI pushes into overbought during the bounce. Bollinger Band resistance sitting right above the move. Every textbook I'd ever read said the same thing — short the bounce, play the range, take the quick fade.
So I did. Full position. Stop above the recent swing high, target back at the bottom of the range.
Price did not care.
Two days of chopping around my entry. I kept telling myself the divergence was still valid, that I was early and not wrong. Then it pushed through the band and closed above it. I remember refreshing that H4 candle every time it flipped. Waiting for the upper band to reject me. Waiting for RSI to curl over.
It didn't curl. It kept grinding higher, the way trends do. And I kept adding to a losing short because the indicator told me I was early, not late.
Covered on day four. Took a hit I had no business taking. That position was never based on structure. It was based on a reading that a moving average calculation had gotten too high.
That sentence changed everything for me.
RSI overbought doesn't mean price is too expensive. It means price has been rising faster than its own recent average. That's a description of momentum. Not a warning. In a strong trend, overbought can stay overbought for weeks. Every trader who shorted it on that basis got run over by the tape.
Same story with the bands. A squeeze tells you volatility has compressed. Tells you a move is coming. Doesn't tell you the direction. Doesn't tell you whether the move matters. I'd been treating the squeeze as a setup.
It's a condition.
Somewhere in year two I stopped logging and just stared at the chart for a while. The trades that worked all had one thing in common — and it wasn't an indicator reading. Price was entering a zone that mattered on the D1, in the direction of the trend. That was it.
That was the whole edge.
Year Three: What My Own Log Actually Said
By the end of year three, the pattern in my spreadsheet was hard to argue with.
Indicator signals worked fine in trending conditions. Which is another way of saying they worked when nothing needed fixing. In ranges? Close to a coin flip. Worse — they encouraged me to take trades in the middle of nowhere.
Not at a level. Not at a swing. Just wherever the RSI line crossed something.
If an indicator can't tell you where you are in the structure, what are you actually trading? You're trading a formula's relationship to itself.
Bollinger Bands measure volatility. How wide recent swings have been, expressed as a channel around a moving average. RSI measures momentum. How fast price has moved relative to its own recent moves. Both are honest descriptions of what already happened. Neither one knows your support shelf, your trend, or your session.
Here's the part nobody tells you: a tool that summarizes the last twenty candles is not the same as a framework that tells you where your next decision gets made.
What They Can and Cannot Do
I didn't throw them away because they're useless. I threw them away because I couldn't use them alone — and when I used them alone, they cost me.
Where RSI actually earns its keep: confirmation on the higher timeframe, inside a range, after structure has already told you a level matters. If the D1 says a shelf is where buyers show up, and momentum starts turning off that shelf — that's worth something. In a trend? RSI fighting the trend is a great way to burn money. I proved that more than once.
Where Bollinger Bands earn their keep: telling you when not to trade. A squeeze is a great reason to sit on your hands and let the market pick a direction first. It's a terrible reason to pick a side before it does.
Both are context tools. Neither is a standalone edge. Took me a decade of screen time to actually feel that, not just know it.
The Framework I Kept Instead
What I kept is boring. And it works.
Structure first. Always.
Every chart starts with the same three questions. What's the D1 trend? Where is price sitting inside the recent range, measured as a Fibonacci retracement? And where are the levels that have already rejected price on the daily?
Trend up? I want to be a buyer. I only want to buy pullbacks into retracement zones that overlap a level the market has already respected. Pullback stalls in the middle of nowhere? I pass. Reaches the deep retracement and the level holds? Now I have structural advantage.
That's the word I care about. Advantage.
Fibonacci isn't magic. It's a ruler. Gives me a consistent way to describe where price is inside the move instead of guessing. When a retracement zone lines up with an old support level and the D1 trend is still up — I have two independent reasons to be interested. If RSI also happens to be turning there? Fine. Bonus. Not the trade.
I also stopped taking entries before the US session opens. London and Asia set the levels. The US session decides whether they hold. Marking my zones before the open, then waiting for the actual reaction — removed most of my impulse entries.
Confidence matters here too. Trend clear, level already tested and held? I'm around 70 percent confident. First test, messy structure? Closer to 60. I size accordingly. I don't pretend to know more than I know.
What I'd Tell You If You're Starting Now
You don't have to repeat my three years. Short version:
Don't build a system around a single indicator. Not RSI. Not Bollinger Bands. Not anything that only reads price against itself. Ask what it's describing. Then ask whether that description tells you where to place your entry and your stop. If the answer is no — it's a filter, not an edge.
Learn where price is before you learn when momentum shifts. Structure first. Trend, level, retracement. Then, and only then, check whether an oscillator agrees with you. If it disagrees?
Structure still wins.
And keep a log. Not because it's fun. Because it's the only way to catch yourself crediting a trend for an indicator's work. I would've saved a lot of tuition if I'd read my own spreadsheet honestly in year one instead of year three.
So here's my question. When your last losing trade went bad — was it because your indicator was wrong? Or because you never actually knew where price was sitting?
Answer that honestly before you add the next tool to your chart.