Someone just ripped the chase card. Are you opening into a looted machine?
We followed 19,052 big hits to see what the next opener actually inherited.
Watching somebody pull the chase card creates the opposite of a hot hand. The best card is gone, the machine looks looted, and opening next feels like volunteering to clean up what is left. That fear only makes sense if one big hit leaves the pool meaningfully poorer for the people behind it.
We found 19,052 big hits — a top-tier card or one worth at least four times the pack price — and looked for trustworthy pool readings immediately before and after each one. We could bracket 16,999 of them. The other 2,053 did not have acceptable readings on both sides, so they are not smuggled into the result.
The median immediate change in pool value was 0%. Pools did sometimes thin; where the measured value actually dipped, the pool's measured value was back to its pre-hit level in a median 3.56 minutes. That is value recovery, not a verified inventory addition.
Then we checked the part a player feels: the next hour of pulls, adjusted so the after-hit group and a disjoint baseline had the same machine mix. Baseline pulls were not within an hour after any earlier big hit on that machine. Value returned per dollar of pack cost was 1.0818× after a hit versus 1.0694× at baseline. Top-tier rates were 1.127% versus 1.091%. The direct-standardized value contrast was +0.0124× after minus baseline: the point estimate leans a hair richer. Under the conservative dependence-aware accounting, clustered by UTC ISO week, its 95% interval runs from −0.0018× to +0.0265× and crosses zero. The difference is indistinguishable from zero.
This is not a claim that a finite pool never thins. It is the narrower finding the data earns: we found no evidence the next hour is poorer, nothing stronger. The fractional lean is not a reason to chase hits either. The same −EV math is waiting either way.
Method and numbers: analysis/report_stories_202608/NUMBERS.md, Measurement B; machine-readable values in analysis/report_stories_202608/numbers.json, measurement_b.primary. Pool readings had to bracket the hit within five minutes without crossing a data gap; next-hour outcomes are direct-standardized to the disjoint baseline machine mix; the ledger reports both the pull-level interval and the conservative dependence-aware bound, clustered by UTC ISO week because one hit's exposure label spans many subsequent pulls.