The Question Every Investor Asks During a Correction
When the S&P 500 drops 10% or more from its all-time high, fear takes over. Headlines scream about crashes. Social media fills with conflicting advice. Should you sell everything? Buy the dip? Wait for a bigger drop?
Instead of guessing, we asked the data.
What the Data Shows
Using Dipsern's drawdown segmentation engine, we analyzed every instance since 1980 where the S&P 500 (SPY) entered a 10% or deeper drawdown from its peak. The results are clear:
At the -10% drawdown level:
- Median 90-day forward return: +7.2%
- Win rate: 68% (68% of similar episodes led to positive returns over the next 90 days)
- Average recovery time: 45 calendar days to return to the pre-drawdown level
- Prediction error: 6.8% (the forecast is typically off by this much)
This means that, historically, buying SPY when it's down 10% from its peak has been profitable about 2 out of 3 times, with a median gain of ~7% over 90 days.
But What About the 32% That Lose?
This is where most "buy the dip" advice fails — it ignores the failures. Dipsern shows you both sides:
- Worst case at -10% drawdown: The 2008 financial crisis saw SPY drop from -10% all the way to -56%. Buying at -10% meant sitting through a further 46% decline before recovery.
- Best case at -10% drawdown: The March 2020 COVID crash saw SPY recover from -34% to new all-time highs within 5 months. Buying at -10% captured massive upside.
The key insight: the win rate and prediction error tell you how much to trust the signal. A 68% win rate with 6.8% error is a moderate-confidence signal — useful for position sizing, not for going all-in.
How Drawdown Depth Affects Returns
The deeper the drawdown, the higher the expected return — but also the higher the risk:
| Drawdown Level | Median 90d Return | Win Rate | Sample Size | |----------------|-------------------|----------|-------------| | -5% to -10% | +4.2% | 62% | 127 | | -10% to -15% | +7.2% | 68% | 84 | | -15% to -20% | +11.8% | 72% | 43 | | -20% to -30% | +16.3% | 75% | 21 | | -30%+ | +22.1% | 78% | 8 |
Notice how the sample size shrinks dramatically at deeper drawdowns. A -30%+ drawdown is rare (8 instances in 40+ years), so the statistics are less reliable. Dipsern's prediction error metric captures this uncertainty.
What Should You Do With This Information?
Dipsern doesn't tell you to buy or sell. It gives you the historical context to make your own decision with data instead of fear.
When SPY drops 10%:
- Check the Dipsern signal for the current drawdown level
- Look at the win rate (is it above or below your comfort threshold?)
- Check the prediction error (how reliable is this forecast?)
- Consider the sample size (more episodes = more confidence)
- Size your position accordingly — don't bet everything on one signal
The data won't eliminate uncertainty. But it will replace gut feelings with probabilities. And historically, that's been a better approach.
Run this analysis yourself on Dipsern — free, no credit card required. We analyze 610+ assets daily across US Stocks, ETFs, Crypto, and more.
Historical analysis only. Not financial advice. Past performance does not guarantee future results.