Buying When RSI Drops Below 30｜Backtest of 3 Exits and Risks
A backtest of 100 US stocks in ENTRIQ shows results after RSI drops below 30 vary widely by exit method. We compare win rate, loss size and losing streaks.
These are the results of backtesting 100 stocks in ENTRIQ (5 to 56 years of data depending on the stock) (Dataset v2.0).
RSI below 30 is an "oversold" gauge, but it isn't enough on its own
RSI is an oscillator that measures how stretched the market is on a scale of 0 to 100. An RSI(14) reading below 30 is generally treated as a sign of being "oversold," and it has long been known as a point to watch for mean reversion trades.
But "oversold" does not mean "about to bounce." In a strong downtrend, price can keep falling while RSI stays below 30. In this article, we look at how price behaved in historical data after buying on the next trading day once RSI(14) crossed below 30, comparing three different exit methods.
For a broader look at mean reversion approaches that use RSI, see also Mean Reversion Strategy｜Fade Extremes and Backtest It.
Test conditions
Even with the same "buy when RSI drops below 30" rule, results change when the exit method changes. Please keep the assumptions below in mind as you read.
- Universe: 100 US stocks selected from the current S&P 500 constituents, taking the largest by market cap with up to 15 stocks per sector
- Period: The full history available for each stock; only stocks with at least 5 years of data (5 to 56 years)
- Timeframe: Daily (split-adjusted)
- RSI setting: Period 14 (Wilder's smoothing, the same calculation as ENTRIQ charts)
- Entry: Buy at the open on the next trading day after RSI(14) crosses below 30
- New signals in the same stock while a position is open are skipped (only one position at a time)
The three exit patterns are as follows.
| Pattern | Exit definition |
|---|---|
| A (RSI back to 50) | Exit at the open on the next trading day after RSI returns to 50 or above (if it doesn't, exit at the close on the 60th trading day). No stop loss |
| B (10 trading days) | Exit at the close on the 10th trading day after entry. No stop loss |
| C (-5% stop) | Stop loss at -5%. If the stop isn't hit, exit at the close on the 10th trading day. If the stock gaps down at the open, the fill is at that open price, so a single trade's loss can exceed -5% |
About survivorship bias: This test uses the "current" S&P 500 constituents. Stocks that were delisted or removed from the index in the past are therefore not included. In reality, many stocks dropped out over time due to deteriorating business performance and other reasons, so the results may look more optimistic than what actually happened.
Overall results for the 3 exit patterns
| Metric | A (RSI back to 50) | B (10 trading days) | C (-5% stop) |
|---|---|---|---|
| Sample size | 5,318 | 7,029 | 7,865 |
| Win rate | 72.2% | 56.9% | 45.9% |
| Average P/L | +2.21% | +1.20% | +0.54% |
| Average gain (winning trades) | +6.39% | +6.25% | +6.32% |
| Average loss (losing trades) | -8.78% | -5.58% | -4.40% |
| Profit factor | 1.92 | 1.51 | 1.23 |
| Largest loss (single trade) | -93.3% | -72.5% | -37.9% |
| Longest losing streak (per stock) | 7 | 12 | 16 |
| Median MAE (maximum adverse excursion) | -4.73% | -3.57% | -3.76% |
| Average holding period | 19.2 trading days | 10.0 trading days | 7.6 trading days |
Note: Past sample (A: N=5,318, B: N=7,029, C: N=7,865); not indicative of future results.
Win rate was highest for A (RSI back to 50), with B and C lower. However, a high win rate does not mean low risk. Because A holds for up to 60 trading days with no stop loss, losses on losing trades tend to be larger, and A also recorded the largest single-trade loss of the three patterns.
C had the lowest win rate, but also the smallest average loss on losing trades. On the other hand, because trades were closed when the stop loss was triggered, C had the shortest average holding period per trade and the longest streak of consecutive losing trades within a single stock. Repeating short holds means more trades.
B sits in the middle, with no stop loss and a fixed 10-trading-day hold. Its balance of win rate, average P/L and largest loss falls between A and C.
Results by trend
Market environment definitions: uptrend = close above the SMA200 and the SMA200 higher than 20 trading days earlier; downtrend = the opposite; everything else = range.
| Market environment | A win rate | B win rate | C win rate |
|---|---|---|---|
| Uptrend | 75.6% (N=1,088) | 59.3% (N=1,121) | 52.2% (N=1,161) |
| Downtrend | 70.5% (N=1,906) | 53.8% (N=3,014) | 42.0% (N=3,464) |
| Range | 72.0% (N=2,171) | 59.2% (N=2,700) | 48.1% (N=3,027) |
Note: Past sample; not indicative of future results.
Across all three patterns, win rates tended to be higher in uptrends and ranges than in downtrends.
What stands out in downtrends is the size of the losses. For A in downtrends (N=1,906), the average loss was -10.32% and the largest loss was -93.3%. This reflects positions held without a stop loss, so losses kept growing when price failed to recover. For C (N=3,464), the longest losing streak was 15 (longest losing streak within a single stock), more than in the other market environments. The numbers suggest that during extended downtrends, the stop-loss level was hit again and again.
Results by volatility
Volatility is classified using ATR(14) ÷ close: the top 20% over each stock's full history is high volatility, and the bottom 20% is low volatility.
| Volatility | A avg P/L / avg loss | B avg P/L / avg loss | C avg P/L / avg loss |
|---|---|---|---|
| High volatility | +3.33% / -12.00% | +1.91% / -7.68% | +0.54% / -4.92% |
| Low volatility | +0.55% / -6.62% | +0.38% / -3.31% | +0.31% / -3.18% |
| Normal | +1.84% / -7.12% | +0.86% / -4.62% | +0.58% / -4.08% |
Note: Past sample; not indicative of future results.
In high-volatility periods, average P/L was larger for all three patterns than in the other buckets, but the average loss on losing trades was larger as well. The data suggests these are periods when the overall range of returns widens.
For C in high volatility (N=3,143), the win rate was 37.9%, the lowest figure across all three patterns and all volatility buckets. With bigger price swings, the -5% stop-loss level appears to have been hit more easily. The median MAE (median maximum adverse excursion) in high volatility was -5.00%, indicating that in roughly half of the cases, the unrealized loss reached a level close to the stop.
Results around earnings
Earnings date data is available only from 2008 onward. We compared the 5 trading days before and after earnings ("near earnings") with all other periods ("away from earnings").
| Period | A win rate / profit factor | B win rate / profit factor | C win rate / profit factor |
|---|---|---|---|
| Away from earnings | 75.4% / 2.31 (N=1,750) | 60.8% / 1.66 (N=2,309) | 50.3% / 1.33 (N=2,597) |
| Near earnings (5 trading days before/after) | 67.3% / 1.65 (N=551) | 49.5% / 1.29 (N=656) | 40.9% / 1.01 (N=726) |
Note: Past sample; not indicative of future results.
For all three patterns, win rates near earnings were lower than in other periods. In particular, C (-5% stop) near earnings had a profit factor of 1.01, roughly break-even. The numbers suggest that sharp drops on earnings announcements often hit the -5% stop-loss level, and that direction was harder to read afterward.
Note that samples without available earnings date data (A: 3,017, B: 4,064, C: 4,542) are not included in the figures above.
Limitations and caveats of using RSI 30 on its own
Here are the limitations that this backtest brings to light.
1. Whether you use a stop loss changes the nature of your losses A and B have no stop loss. When price keeps falling, losses grow because the position is held without a stop. A's largest loss was recorded at -93.3%, and it is worth keeping in mind as a fact that a loss of this size in a single stock did occur in the past.
2. Losing-streak risk in downtrends For C in downtrends, a streak of up to 15 consecutive losses (longest losing streak within a single stock) was observed. Even with a -5% stop loss, when a trend persists, the stop-loss level can be hit repeatedly.
3. Risk of holding through earnings C near earnings had a profit factor of about 1, which, as a historical observation, is roughly break-even. Around events with high price uncertainty such as earnings, the RSI-below-30 condition alone appeared insufficient to assess price direction.
4. Impact of survivorship bias As noted above, this test does not include stocks removed from the S&P 500 or delisted. If you use the same conditions in a real market environment, results may differ from what was observed here.
With these limitations in mind, it is important to check how the numbers change for the stocks, timeframe and environment you actually trade. With ENTRIQ's Chart Replay, you can manually run the same RSI-below-30 condition on your own stocks, one bar at a time. You can record and tally your results by condition with Strategy Tags.
Summary
We compared a backtest of buying on the next trading day after RSI(14) crossed below 30 (100 stocks, Dataset v2.0) across three exit patterns.
- A (until RSI returns to 50): Highest win rate, but because positions were held without a stop loss, losses on losing trades were large and the largest single-trade loss was also the greatest among the three patterns. Average holding period was relatively long at 19.2 trading days
- B (fixed 10 trading days): A middle-ground design with no stop loss and a fixed holding period. Both win rate and losses fall between A and C
- C (-5% stop): Lowest win rate, but the average loss on losing trades was contained. Win rates dropped noticeably in high volatility, downtrends and near earnings
For every pattern, results were relatively better in uptrends and away from earnings, and a common tendency for results to worsen in downtrends and near earnings was observed. RSI below 30 can work as an "oversold gauge" in some situations, but decide whether to build it into your own rules only after understanding, through data, its limitations when used on its own.
For how to run a backtest, see also What Is Backtesting｜Systematic vs Discretionary Approaches and Discretionary Backtesting Software｜Why No-Code Isn't Enough.
Frequently asked questions (FAQ)
Q. Is pattern A, with the highest win rate, the best? A. Win rate alone can't tell you. A is designed to hold without a stop loss until RSI returns to 50, so losses grow when price keeps falling. Its largest loss was recorded at -93.3%, and the maximum adverse excursion while holding (median MAE) was -4.73%. You need to look at win rate, loss size and longest losing streak together and weigh them against your own risk tolerance.
Note: Past sample (N=5,318); not indicative of future results.
Q. What happens with an RSI period other than 14? A. This article reports observations with a period of 14. A shorter period produces more frequent drops below 30, while a longer one makes them less frequent. Since results change too, consider checking again with the period you actually use.
Q. What was observed when buying on RSI below 30 during a downtrend? A. In this backtest, win rates in downtrends were lower than in other environments. In the patterns without a stop loss, losses on losing trades also tended to be especially large. Please check the same conditions with your own stocks and rules before making a decision.
Q. Do results change if you limit trades to periods away from earnings? A. For all three patterns, the win rate and profit factor "away from earnings" were better than near earnings. However, this is an observation based on past samples and does not indicate future results.
Q. Can I use these results directly in my own trading? A. Results change if the stocks, period, timeframe or exit rules differ. The figures in this article are observational data meant to help organize your thinking. Please verify them with ENTRIQ's Chart Replay under the conditions you actually trade.
This article is provided for informational purposes to examine and organize historical data and does not recommend any specific trade. The figures shown are past samples and do not indicate future results. All investment decisions are your own responsibility.
Reproduce this validation yourself
With ENTRIQ's chart replay, you can trade through past charts using the same rules.
Start 14-day free trialValidate your method on past charts
The same validation in this article can be reproduced with your own method. On past charts, do the classics really work?
- ✓ Replay past charts
- ✓ AI feedback on your trades
- ✓ Validate methods with statistics