Research
Do RSI signals work?
The Relative Strength Index measures how fast price has moved. The classic rule buys below 30 and sells above 70. We sent these signals in our first version, then tested them properly. Here is what happened.
| Signal | Trades | Net per trade | vs random | Verdict |
|---|---|---|---|---|
| Stochastic RSI overbought | 4,293 | -0.4% | +0.4% | Beat random, still lost money |
| RSI overbought (above 70) | 1,883 | -0.5% | +0.3% | No edge over random |
| RSI oversold (below 30) | 742 | +1.0% | -0.3% | No edge over random |
| Stochastic RSI oversold | 4,024 | -0.0% | -1.1% | No edge over random |
24 markets (crypto, US stocks, forex, gold, oil), 2017–2026. Entries at the next open, 1 ATR stop, 2 ATR target, 10-day limit, costs of 0.03–0.15% per side. "vs random" compares with random entries in the same markets and direction using the same exits; a bearish signal can beat random shorts in a rising market and still lose money. Hypothetical results have inherent limitations.
What we use instead
A much faster RSI. A pullback rule built on a 2-period RSI: buy when RSI(2) drops under 10 while price is above its 200-day average, and sell after the first close above the 5-day average. It beat random entries on data it was not chosen on.
See the rules, backtest and live record →Questions
Do RSI signals work?
Not on their own, in our test. Across 24 markets since 2017, using the alert's stop and target and realistic costs, the RSI signals we tested did not make money compared with random entries in the same markets.
How was this tested?
Each signal was replayed day by day using only data available at the time, entered at the next open, exited at a 1 ATR stop, a 2 ATR target or after 10 days, and compared with thousands of random entries using the same exits. Results are hypothetical.
What does Vantum use instead?
A much faster RSI. A pullback rule built on a 2-period RSI: buy when RSI(2) drops under 10 while price is above its 200-day average, and sell after the first close above the 5-day average. It beat random entries on data it was not chosen on.
Get the next trade in time to act on it
Pro sends each entry as the daily close confirms it and an exit alert when it's time to close. $19/month after a 7-day free trial. Free accounts see every trade 24 hours later.