Bottom Line
I actually like the main correction in this video. It takes RSI away from the lazy “above 70 means short, below 30 means long” routine, which has probably trapped half the internet at some point.
The better idea is to treat RSI as a momentum read. That feels much more useful. But I am not ready to call it a strategy yet. The video still leans on phrases like “cleanly breaks,” “retests around the 50 level,” and “put more emphasis on nearby divergences.” Those sound right on a chart, but they are not complete rules.
So my current verdict is promising but unproven. Interesting enough to test. Not solid enough to trust.
What The Video Said
The main claim is that most traders use RSI backwards. They see 70, assume the market is exhausted, and start hunting shorts. They see 30, assume the market is cheap, and start hunting longs. Sometimes that works, but it is also how you end up standing in front of a trend and wondering why the chart refuses to care.
The video argues that 70 and 30 are better used as context, especially for exits or for judging whether a divergence is meaningful.
The more important line, according to the video, is RSI 50:
- RSI breaking above 50 suggests bullish momentum.
- RSI breaking below 50 suggests bearish momentum.
- Pullbacks toward 50 can behave like momentum support or resistance.
- A moving average on RSI can smooth the signal.
- Divergences matter more when they happen near the outer RSI zones.
- Divergences should not be stretched across too many candles; the video ties that idea to the RSI length setting.
That is a much healthier way to think about the indicator. RSI stops being a reversal button and becomes a momentum state variable. Less dramatic, probably more useful.
How I Translated It Into Rules
For the first reusable test, I kept the rules deliberately simple. This is the “does the basic version have a pulse?” pass, not the final boss version of the strategy.
| Component | Test rule |
|---|---|
| Input data | A CSV supplied by the reader with datetime, open, high, low, close columns |
| Indicator | 14-period RSI |
| Trend filter | RSI moving average, default 9 periods |
| Long setup | RSI is above 50 and the RSI moving average is above 50 |
| Short setup | RSI is below 50 and the RSI moving average is below 50 |
| Entry | Next bar after the signal |
| Exit | Opposite regime or a fixed max holding period |
| Costs | Round-trip basis points, configurable |
| Sizing | One unit of notional for return math |
This is not the full video concept yet. It does not fully encode the “box around 50” retest idea or the divergence logic. I am starting with the plain momentum filter because if that has no edge at all, the fancier parts need to work very hard to save it.
Data Used
The public script does not include this site’s private data files. To reproduce the test, download the script and pass your own OHLC CSV:
python3 run_backtest.py --csv your_ohlc.csv --cost-bps 2 --max-hold 20
The script writes a summary table and a trade list next to your chosen output folder. It is intentionally boring in the best way: no hidden data dependency, no credentials, no private feed. Just your CSV, the rule, and the result.
Cross-Market Test
Because the video talks about stocks, crypto, and Forex broadly, I did a first sweep across the data I have locally: major Forex pairs, gold, US equity indices, and Bitcoin. I am publishing the aggregate results, not the raw data files.
The setup was the same simple version from above:
- 14-period RSI
- 9-period moving average of RSI
- long when both are above 50
- short when both are below 50
- exit on opposite regime or after 20 bars
- 2 bps round-trip cost
- daily bars (
1d)
This is where the idea got more interesting, but also messier. Bitcoin looked great in isolation. Nasdaq looked decent. The broad Forex basket was mostly not good. Gold was basically a shrug. Then the buy-and-hold comparison poured some cold water on the exciting parts.
Crypto 2017-02-16 to 2026-06-04
| Symbol | Trades | Win rate | Avg trade (bps) ? | Net return | Buy & hold net return | Profit factor |
|---|---|---|---|---|---|---|
| BTCUSD | 244 | 35.2% | 240.69 | 3,042.2% | 6,155.7% | 1.70 |
Forex 2011-02-08 to 2026-05-22
| Symbol | Trades | Win rate | Avg trade (bps) ? | Net return | Buy & hold net return | Profit factor |
|---|---|---|---|---|---|---|
| AUDUSD | 357 | 30.0% | -6.12 | -23.8% | -29.7% | 0.90 |
| EURUSD | 378 | 27.2% | -7.64 | -27.9% | -14.9% | 0.86 |
| GBPUSD | 372 | 29.8% | -4.88 | -19.3% | -16.6% | 0.90 |
| NZDUSD | 372 | 27.4% | -8.84 | -32.3% | -24.1% | 0.87 |
| USDCAD | 370 | 26.2% | -3.37 | -14.6% | 39.5% | 0.93 |
| USDCHF | 377 | 24.4% | -12.93 | -42.0% | -17.9% | 0.78 |
| USDJPY | 375 | 27.2% | -2.24 | -12.9% | 93.3% | 0.96 |
Gold 2014-10-17 to 2026-06-04
| Symbol | Trades | Win rate | Avg trade (bps) ? | Net return | Buy & hold net return | Profit factor |
|---|---|---|---|---|---|---|
| XAUUSD | 273 | 30.0% | 3.05 | -2.7% | 262.9% | 1.03 |
Indices 2014-11-05 to 2026-06-04
| Symbol | Trades | Win rate | Avg trade (bps) ? | Net return | Buy & hold net return | Profit factor |
|---|---|---|---|---|---|---|
| NAS100 | 262 | 35.9% | 24.57 | 59.7% | 628.8% | 1.20 |
| US30 | 257 | 32.7% | -0.68 | -12.1% | 196.1% | 0.99 |
| US500 | 268 | 32.8% | 5.15 | 4.9% | 275.7% | 1.06 |
This is exactly why I do not want to stop at one pretty chart. If you tested only Bitcoin, you would get excited. If you added buy-and-hold, you would calm down fast. The RSI version made a lot of money, but it still lagged just owning Bitcoin over the same period.
The same thing happened on Nasdaq. Positive strategy result, but nowhere near buy-and-hold. Forex was even less flattering: the strategy lost money on every major pair in the aligned-period sweep. That makes the RSI 50 momentum concept look less like a universal trading strategy and more like a rough trend filter that still needs a real reason to beat simple exposure.
First Read
The idea is worth testing because it asks a better question than the usual RSI video. Instead of “RSI is high, should I short?”, it asks “what momentum regime am I in, and did momentum actually shift?”
That distinction matters. A strong trend can stay above 70 for a painfully long time. Shorting it just because RSI is high is often a bet against strength. Using RSI 50 as a state filter at least tries to stand on the same side as momentum.
What Could Break It
The obvious danger is lateness. By the time RSI confirms momentum, the easy part of the move may already be gone. That is the eternal tradeoff with confirmation: it feels safer, but it often arrives after the party started.
The other problem is chop. RSI can flip above and below 50 repeatedly, creating lots of little trades that look harmless until costs start quietly eating the account.
The divergence idea might be more selective, and I want to like it. But it is harder to test honestly. You need objective swing points, a maximum bar distance, and a rule for what happens when price makes a marginal new high or low. Without those details, divergence testing turns into chart poetry very quickly.
Verdict
Verdict: Promising but unproven.
The video’s framing is better than the usual overbought/oversold shortcut. That part is a real improvement. The sweep also makes me more interested than I was before, because Bitcoin and Nasdaq responded well to the simple momentum version.
But it is not a blanket win, and the buy-and-hold comparison matters a lot. Forex was weak, gold was not convincing, and the strongest results underperformed simple exposure. The first downloadable script is a starting point for testing the momentum-filter version on your own market data, not proof that RSI beats the market.
Disclaimer
This is personal research and not financial advice. Backtests can be incomplete, overfit, or impossible to execute live.