Why Gap Base Rates Matter
A TSLA earnings-day gap greater than 3% is not a rare event — it has happened 34 times since 2016. That's enough to compute base rates on rather than argue from a single prior print. The question isn't 'did TSLA gap' (it did), but 'when TSLA has historically gapped like this, what has the following week done?'
Gap Hold vs Fade Base Rates
Across the 34 historical TSLA earnings-day gaps greater than 3% (both directions), the 5-day follow-through split roughly evenly: 18 of 34 held the gap direction, 16 of 34 faded. That's essentially a coin flip. But the split is NOT random — it correlates with two specific features:
- Pre-earnings drift: when TSLA rallied into the print, the gap tended to fade (supply from momentum exits). When TSLA drifted into the print, the gap tended to hold.
- VIX level on the gap day: gaps at VIX < 20 held more often (11 of 14) than gaps at VIX 20-30 (5 of 12) or VIX > 30 (2 of 8).
- Volume relative to 20-day average: gaps on 3x+ average volume held more often than gaps on 1.5-2x volume (stronger commitment).
Using the Base Rates
Base rates are a prior, not a signal. The useful thing to do with a 53% historical hold rate is to check whether today's setup matches the 'hold' cluster (pre-earnings drift + low VIX + high relative volume) or the 'fade' cluster. Chart Library's pattern search does that matching automatically — it doesn't just return TSLA's own history, it returns analog setups across 19,000+ symbols that shared the pre-event conditions.
The result is a conditional base rate rather than an unconditional one. Conditional base rates tend to be more useful for agents because they reflect the specific regime the stock is in, not just a pooled long-run average.