Every pick, graded against the close,
locked the moment it drops.

Yes. Every number on this page is computed from the picks table in Neon at request time with a 60-second cache. No hardcoded figures. Rows are locked at post time via a Postgres trigger, so we cannot retroactively edit or delete a pick.

Hit rate is wins divided by (wins + losses). It ignores odds. ROI is net units won divided by units risked. It treats a dollar risked at -200 the same as a dollar risked at +200 in terms of denominator. Yield is net units won divided by total settled picks, a flat-staking view. A 60% hit rate on -200 favorites is losing money; a 53% hit rate at +100 is profitable. Never judge a handicapper on hit rate alone.

Because at 100 picks, you cannot statistically distinguish a true edge from pure luck. Roughly 1,900 settled picks are needed before a true 55% win rate becomes detectable at p<0.05 — and a thinner 54% edge needs far more, around 4,700. Our 500-pick mark is a display threshold, not statistical proof. The sample note on each window tells you where this record sits on that curve, and CLV proves edge faster than win rate.

Closing Line Value. The difference between the odds you posted and the final market consensus at tip-off. Positive CLV over a big sample is the single most respected metric among professional bettors because it is much less noisy than short-term hit rate. We compute it in decimal odds: (posted / closing) - 1, averaged across picks with a closing snapshot.

CLV-beat % is the share of our graded picks that beat the closing line (no-vig CLV above zero). A high average CLV can be propped up by a few lucky outliers; the beat rate cannot be gamed that way — it asks, simply, how often did we get a better number than the market's final consensus? A genuine model beats the close on a majority of bets (above 50%), win or lose, because closing line value is the leading indicator of long-run profit. We compute it from the persisted no-vig CLV on every graded pick that carried a closing snapshot, and we show the raw count so you can judge the sample.

Raw ROI flatters anyone who only beat the juice. Vig-adjusted ROI subtracts the average sportsbook hold per pick from raw ROI. We compute the per-pick hold from each pick's closing decimal odds (implied probability times two minus one hundred) and average across the window. When closing data is sparse for a window we fall back to the standard -110 hold of 4.55%. If ROI ex-vig is still positive over a meaningful sample, the edge survives the books. Leans.ai charges $299/mo to surface this metric; ours is in the same grid as raw ROI on every window, free.

No. Every picks row is written with locked_at set to NOW() on insert. A database trigger rejects UPDATE or DELETE of locked columns. Only grading fields (result, graded_at, closing_odds) can change, and those are append-only logs, not edits.

No. NuroPicks is entertainment and education. Past ROI or CLV does not predict future results. 21+ only. If sports betting stops being fun, call 1-800-GAMBLER or visit the responsible gambling resources page.