How the daily Tournament actually grades your prediction

2026-10-01

"Predict your own play" sounds like it should be on the honor system — type a number, get a reward, nobody checking. It isn't. Every entry gets graded against the same real activity data that powers the rest of the site, which is really the whole point: this is a prediction game about a thing that's already being measured anyway.

What you're actually predicting

Once a day, you pick a number for either how many games you'll play or how many minutes you'll spend playing, before that day (UTC) ends. One entry per day — no re-rolling once you've seen how the day is going.

Graded against telemetry, not your word

At resolution, the actual count gets pulled from real session data: games played comes from genuine session-start events, one per game actually opened; minutes played comes from heartbeat events that only fire during confirmed, active play — sitting on a paused game doesn't quietly rack up minutes. Land close to your prediction and you get a reward; land very close and you get more; miss by a wide margin and you get nothing for that entry. Among the people who landed close or better, the closest predictions across the whole player base also earn an extra bonus on top — so being close matters, but being closest matters more.

Resolution doesn't depend on perfect timing

There's a scheduled job that resolves each day right after it ends, but the system doesn't actually depend on that job firing on time. Every time the tournament page loads, it re-checks the last several days and resolves anything that's still pending — so a delayed cron run or a missed midnight doesn't mean a lost or wrong result, it just means resolution happens the next time anyone looks.

A concrete run-through

Say you predict 8 games played today. You end up playing 7 — within the tight band around your guess — so that counts as a genuine hit and pays out accordingly. Someone else predicts 8 and plays exactly 8: an even tighter match, so if the ranking among everyone who landed close or better comes down to it, they're more likely to land in the small group that gets the extra top-predictor bonus. Someone who predicted 8 and played 2 gets nothing for that entry — not because the system is harsh, but because nothing about that prediction was actually close.