Warren Buffett's old line is simple: be fearful when others are greedy, and greedy when others are fearful. It sounds obvious after the fact. It is much harder when the screen is red, volatility is exploding, and every headline sounds like the end of the cycle.
We tested the practical version of that idea: when the VIX spikes, does buying SPY actually work? The short answer: yes, over this sample, fear improved forward returns. The longer answer is more useful: it worked best when the signal was treated as a medium-term allocation cue, not a perfect one-day bottom timer.
The data matters here. SPY daily bars came from Alpaca, adjusted for splits and dividends. Alpaca did not return CBOE VIX index bars, so true VIX closes came from Yahoo Finance's chart endpoint. Alpaca also only returned SPY history back to January 4, 2016, so this is a 2016-2026 test, not a full market-history study.
The signal
We tested three fear regimes:
- VIX 20-30: elevated fear, stay alert.
- VIX 30-45: panic, start getting aggressive.
- VIX 45+: crisis-level fear, maximum fear regime.
There are two ways to measure this. First, look at every trading day that falls in each VIX bucket and measure forward SPY returns from that close. Second, use a more tradable version: buy SPY at the next open after a fresh VIX threshold trigger, then apply a 21-trading-day cooldown so the same panic does not get counted over and over.
Every panic day: higher fear, better forward returns
Looking at all bucket days, forward returns improved as VIX moved higher. The effect was clearest beyond one month.
| VIX bucket | 21d avg | 21d win | 63d avg | 252d avg |
|---|---|---|---|---|
| All days | +1.32% | 70.1% | +3.81% | +15.77% |
| VIX 20-30 | +1.48% | 66.8% | +4.71% | +18.48% |
| VIX 30-45 | +4.40% | 85.9% | +8.23% | +25.94% |
| VIX 45+ | +11.53% | 92.3% | +23.39% | +56.33% |
This is the core evidence for the Buffett framing. The market paid investors better forward returns when volatility was already elevated. The best returns came when the fear was obvious, uncomfortable, and widely discussed.
The tradable version: buy the next open
The more realistic test is stricter. It only buys after a fresh threshold trigger and waits 21 trading days before accepting another signal from the same threshold. That reduces repeated counting during prolonged panic clusters.
| Trigger | Events | 21d avg | 21d win | 63d avg | 252d avg |
|---|---|---|---|---|---|
| VIX >=20 | 111 | +1.66% | 66.7% | +4.85% | +15.08% |
| VIX >=30 | 30 | +4.54% | 80.0% | +7.10% | +20.41% |
| VIX >=45 | 3 | +5.82% | 66.7% | +19.76% | +43.48% |
This version still supports the idea. A fresh VIX 30 trigger produced an average 21-trading-day SPY return of 4.54%, with an 80% win rate. Over one year, the same trigger averaged 20.41%.
The VIX 45 results look spectacular, but they come from only three independent threshold events in this data window. That is not enough to call the signal statistically robust. It is enough to say that, in this sample, the most terrifying moments were not moments to panic sell.
Why timing still matters
Buying fear does not mean every entry is immediately profitable. The 5-day and 10-day results were noisier, especially around the most extreme spikes. During a real volatility shock, price can keep falling after the first headline panic.
The cleaner version was buying after the VIX peak inside each panic episode. That is not knowable in real time, but it shows an important point: the edge improves once volatility stops getting worse. For VIX 30+ episodes, buying after the episode's VIX peak produced a 21-day average return of 5.83% and a 63-day average return of 8.41%.
What the test does not prove
This is not a proof that VIX spikes are free money. There are several limitations:
- The backtest only covers January 2016 through May 2026 because that is the SPY history Alpaca returned in this environment.
- SPY came from Alpaca, but VIX came from Yahoo Finance because Alpaca returned no bars for the CBOE VIX index.
- VIX 45+ produced only three independent threshold events, so the sample is too small for strong statistical claims.
- The test uses SPY only. It does not include options, leverage, slippage, taxes, margin, or position sizing rules.
- The 21-trading-day cooldown is a modeling choice. Different cooldowns would change the number of counted episodes.
The practical takeaway
The Buffett idiom held up directionally: fear was useful. VIX 20 was a heads-up. VIX 30 was a stronger buy zone. VIX 45 was rare and historically powerful in this sample, but too rare to size blindly.
The better interpretation is not "buy the exact bottom." It is: when the crowd is forced to de-risk, future returns tend to improve for patient buyers. Fear is not the whole strategy, but it is a better input than chasing calm.