RSI 2 Strategy: Why It Works on Indices but Not FX
Short-term mean reversion rules have a following among retail traders, and the RSI(2) pullback is one of the most shared versions. The idea is simple: buy a short dip in an uptrend and sell when momentum recovers. The question is whether that idea survives costs and a fixed risk rule, and whether it works the same way across markets.
The headline finding is that the same rule produced a clear positive result on the two US index futures tested, ES and NQ, and a weak or losing result on gold and the two currency pairs. The gap between markets was larger than the gap between years. Read the per-market numbers below before drawing any conclusion about the rule itself.
The rules we tested
Buy when RSI(2) crosses below 10 and price is above the 200 SMA. Exit when RSI(2) crosses above 70. Stop at 3x ATR.
Results
| Market | Trades | Win rate | Profit factor | Net | Max DD | Period |
|---|---|---|---|---|---|---|
| ES 1d | 86 | 81.4% | 1.89 | +12.0% | -3.2% | 10y |
| NQ 1d | 84 | 73.8% | 1.42 | +6.0% | -3.4% | 10y |
| XAU/USD 1d | 59 | 66.1% | 1.07 | +0.7% | -3.7% | 10y |
| GBP/USD 1d | 58 | 65.5% | 0.78 | -2.0% | -3.4% | 10y |
| EUR/USD 1d | 44 | 59.1% | 0.62 | -2.9% | -4.3% | 10y |
1% risk per trade, spread and slippage included, entries at the next candle open, stop checked before target. Periods: ES 2016-11-13 to 2026-11-13 · NQ 2016-11-13 to 2026-11-13 · XAU/USD 2016-11-13 to 2026-11-13 · GBP/USD 2016-11-13 to 2026-11-13 · EUR/USD 2016-11-13 to 2026-11-13.
Equity (% at 1% risk per trade) on ES 1d, 86 trades.
What was tested and why
The rule was tested on daily bars from 2016-11-13 to 2026-11-13, across ES, NQ, XAU/USD, GBP/USD and EUR/USD. It only takes long trades. A long is opened when RSI(2) crosses below 10 while the close is above the 200-day simple moving average. The engine entered at the next candle open.
The stop was placed at 3 times the 14-period ATR. The exit was when RSI(2) crosses above 70. Spread and slippage are included, and each trade risks 1% of the account. If the stop and the target were hit in the same candle, the stop was counted first.
The engine read the rules exactly as follows: Long RSI(2) crosses below 10; Long Close above SMA(200); Stop at 3 times ATR(14); Exit when RSI(2) crosses above 70. No short side was tested, and no other filters were applied.
Results by market
ES produced 86 trades with a win rate of 81.4%, a profit factor of 1.89 and an average R of 0.13. At 1% risk per trade the net return was 12%, with a maximum drawdown of 3.2%. NQ produced 84 trades, a 73.8% win rate, a profit factor of 1.42, an average R of 0.07, a net return of 6% and a maximum drawdown of 3.4%.
Gold, shown as XAU/USD, produced 59 trades with a 66.1% win rate and a profit factor of 1.07. The average R was 0.01 and the net return was 0.7% at 1% risk. That is close to break even after costs, and the sample is modest.
GBP/USD produced 58 trades, a 65.5% win rate and a profit factor of 0.78. The average R was -0.03 and the net return was -2% at 1% risk. EUR/USD produced 44 trades, a 59.1% win rate and a profit factor of 0.62. The average R was -0.07 and the net return was -2.9%.
To be direct about the currency pairs: a profit factor under 1 means the rule lost money after costs. Both GBP/USD and EUR/USD fell into that group. The index results are the only ones with a clear positive edge in this test.
What stands out: years, and long versus short
The index results were not smooth. ES had a losing 2018 (6 wins from 9 trades, total R of -1.5), a losing 2019 and a losing 2022 with only 2 trades. Most of its total gain came from a strong 2017 and 2021. NQ showed a similar pattern, with 2018 and 2019 both negative in total R.
Every trade in every market was a long. The rule has no short side, so this test says nothing about whether a mirror version works. The sample also thins out in some years. GBP/USD, for example, has no 2022 trades listed at all, and EUR/USD has none for 2019.
Sample size matters here. ES and NQ each have roughly 85 trades, which is a reasonable base for a first look. Gold, GBP/USD and EUR/USD have between 44 and 59 trades. None of the five markets reaches 100 trades, and the currency pair results rest on fewer trades than the index results. Treat the smaller samples as indicative only.
Why it might behave this way
One plausible reading is that the rule was built for markets with a persistent upward drift. Stock indices have spent long stretches rising, and short pullbacks in an uptrend have tended to recover. The 200-day filter keeps the rule on the long side of that drift, so the rule mostly buys dips when the broader trend is still up.
Currency pairs have less of a long-run drift and tend to range for long periods. A pullback below 10 on RSI(2) in a range is not the same as a pullback in a trend, and the 200-day filter does less to separate the two. This is a hypothesis, not something the test measured. The data here cannot show why the markets differ.
Gold sits between the two groups in this test. Its result was close to flat, which fits a market that trends at times and ranges at others.
What to try next
The most useful next step is to check the rule on more history and on more instruments of the same type, so the sample grows past the current 44 to 86 trades. A test on other index futures or on a longer daily history would show whether the ES and NQ results repeat.
It would also be sensible to test the rule with and without the 200-day filter, to see how much of the result depends on it. Testing the stop distance and the RSI exit level separately would show which parameters carry the edge and which are noise. Anyone can run the same test themselves on WickCheck with the rules above, and the results will show whether the pattern holds on their chosen market and period.
Takeaway. On this test, the RSI(2) pullback rule with a 200-day trend filter had a clear positive result on ES and NQ, with profit factors of 1.89 and 1.42 after costs and 1% risk per trade. It was roughly flat on gold and lost money on GBP/USD and EUR/USD, where the profit factors were 0.78 and 0.62. The index results come from about 85 trades each, which is a reasonable start but not proof. The currency pair results are weaker and the sample is smaller. Whether any of this holds on other periods or instruments is still an open question.
Same rules on ES. Change any word, a market or the dates and see what moves.
Terms used here
Questions
- Does the RSI 2 strategy work on the S&P 500?
- In this test, the ES future (the E-mini S&P 500) produced 86 trades from 2016-11-13 to 2026-11-13, with a profit factor of 1.89 and a 12% net return at 1% risk per trade. That is a positive result after costs. It is one test on one instrument and one period, so it does not show the rule will work going forward.
- Why did the RSI 2 rule lose money on EUR/USD?
- The test produced a profit factor of 0.62 on EUR/USD across 44 trades, which means it lost money after spread, slippage and the 1% risk rule. The data does not explain why. One hypothesis is that currency pairs range more than indices, so short dips are less likely to recover, but this was not measured here.
- Is the RSI(2) rule a good strategy to trade?
- This article does not make that recommendation. The test covered one set of rules, one daily timeframe and one ten-year period, and it included only long trades. Several markets had fewer than 60 trades, and none reached 100. Readers can run the same test themselves on WickCheck and check how the results change across markets and periods before drawing their own conclusions.
More tests
We tested the 9/21 EMA crossover with a 200 EMA filter on gold, EUR/USD and the Nasdaq, with costs included. Gold lost money after costs over 10 years.
Previous Day High Low Strategy: A Backtest Across 4 MarketsWe backtested a previous day high low strategy with an FVG entry on ES, XAU/USD, EUR/USD and NQ. Two markets lost money after costs. See the numbers.
Asian Range Strategy: Does the London Sweep Hold Up?We backtested an Asian range sweep and fair value gap setup on EUR/USD, USD/JPY, GBP/USD and XAU/USD. Results were mixed, with one market profitable.
Past results, real or simulated, do not guarantee future results. Not investment advice.