GOOGL forward 20-day returns after turbulent vs calm sessions (top/bottom decile by intraday range)
The short answer is no: the biggest intraday swings in GOOGL over the past ~3 years did not produce superior 20-trading-day outcomes. I measured intraday range (high−low)/close and split days into top and bottom deciles (top-decile threshold 20.0053%, bottom 2.6639%), then compared mean forward-20d returns. Turbulent days averaged +2.99% versus an unconditional baseline of +3.34%; calm days averaged +2.53%.
Digging deeper, calm sessions show a higher hit rate (≈67.6% vs 54.1%) but lower volatility (std ≈6.7% vs 9.6%), and none of the mean differences reach significance (turbulent vs baseline p=0.7668; calm vs baseline p=0.3384; turbulent vs calm p=0.7342). The full methodology, charts and statistics are below; the headline: this simple “buy the panic” signal is inconclusive for GOOGL on a 20-day horizon.
For GOOGL over the past ~3 years, does 'be greedy when others are fearful' actually pay in Alphabet — are forward 20-day returns after its most turbulent sessions (top-decile daily high-low range) meaningfully above the everyday baseline, or does a volatility spike just flag more pain ahead? Thesis: the widest-range days mark short-term capitulation, so forward 20-day returns following them come in above baseline while the calmest sessions merely tread water, making panic a better entry than calm.
How this was measured
Minute bars were resampled to daily OHLC over the last ~3 years (bounded by data availability). Computed an intraday range proxy as (high−low)/close per day. Using only days with a valid 20-trading-day forward return available, classified 'turbulent' as the top 10% of range and 'calm' as the bottom 10%. Measured forward-20d returns close[t+20]/close[t]−1 for turbulent, calm, and the unconditional baseline, and compared means via Welch two-sample t-tests (unequal variance).
The key numbers
Reading the numbers
Across 732 trading days, the baseline 20-day forward return is about 3.34%. Turbulent sessions average ~2.99% (slightly below baseline) and calm sessions ~2.53%; turbulent days have a lower hit rate (54.05% positive) than the baseline (65.03%).
The charts
The box plot groups show that turbulent days have a wide spread of outcomes (min −13.09% to max +29.6%) and a mean of 2.9918%, while the baseline mean is a bit higher at 3.3368% and calm days average 2.5287%. Look at the whiskers and outliers: turbulent sessions produce both large wins and large losses, so they’re noisy rather than reliably better. For the question of ‘buy panic,’ the chart shows no clean upward shift after turbulent days — more dispersion, not a higher center.
The mean-return bars make the headline comparison simple: baseline ~3.3368% vs turbulent ~2.9918% and calm ~2.5287%. The turbulent edge vs baseline is about −0.345 percentage points, and calm is −0.808 percentage points, so neither turbulent nor calm beats the baseline on average. That small negative gap (and the small Welch t of −0.297) argues against a dependable ‘buy the panic’ mean edge.
The histogram for turbulent days (n=74) shows a mean of 2.9918% but a wide range from −13.09% to +29.6%, and only about 54.05% of those outcomes are positive. What to watch is the fat tails: yes, there are big winners after some turbulent days, but there are also large losses, so the distribution is noisy and not consistently positive. In short, turbulence flags big moves, not a reliably higher short-term return.
The calm-day histogram (n=74) has a mean of 2.5287% with outcomes from −17.0% to +24.53% and a higher fraction positive (67.57%) than turbulent days. That tells you calm sessions more often lead to modest gains, even if their average gain is slightly below the baseline. For your thesis, calm sessions don’t just tread water — they show a higher hit rate, though smaller average payoff, while panic is more hit-or-miss.
Forward 20-day return summary by bucket
| bucket | N | mean | std | fraction_positive | rv_threshold |
|---|---|---|---|---|---|
| Turbulent (top 10%) | 74 | 0.0299 | 0.096 | 0.5405 | 0.2001 |
| Calm (bottom 10%) | 74 | 0.0253 | 0.067 | 0.6757 | 0.0266 |
| Baseline (all days) | 732 | 0.0334 | 0.0852 | 0.6503 |
The takeaway
No — the backtest does not support the ‘be greedy when others are fearful’ thesis for GOOGL over the next 20 trading days. Days in the top-decile of intraday range produced an average forward-20d return of about +2.99%, which is actually below the unconditional baseline of +3.34%; calm bottom-decile days averaged +2.53%. The hit-rate picture is mixed: calm days were positive more often (about 67.6% vs 54.1% for turbulent), but calm returns show lower volatility (std ~6.7% vs turbulent ~9.6%), which explains the higher win frequency despite a lower mean. None of the comparisons are statistically significant — p-values are large (turbulent vs baseline p=0.7668; calm vs baseline p=0.3384; turbulent vs calm p=0.7342) — so these differences look like noise rather than a reliable edge. Practical takeaway: wide-range days did not produce superior 20-day outcomes in this three-year sample; if you want a repeatable short-term payoff, this simple volatility-signal is inconclusive at best and slightly unfavorable on average.
The fine print
- Each bucket only has 74 cases — thin sample for firm conclusions.
- Forward-20d windows overlap heavily, so t-test p-values are optimistic relative to block-resampling.
- Top/bottom decile cutoffs were defined in-sample on the same 3-year window; thresholds can shift out-of-sample.
- High intraday range can be driven by scheduled events (earnings, macro) and those were not controlled for.