The 12-0 run overreaction
Books move the live spread four or five points on a run. With 30 minutes left, fair usually moves two. When the crowd number outruns the math, the side the run hurt is often the value.
NBA live betting is a scoring-rate problem. Runs of 12-0 feel enormous, but the clock is long and rates regress. The fair spread is rebuilt from capped scoring rates and time confidence, so a run moves the number only as much as the math says it should.
Basketball scoring is nearly continuous, so the model works on rates rather than possessions: expected points per remaining second, capped so a hot quarter can't rewrite a team's identity. Time confidence does the heavy lifting late — a six-point lead with two minutes left is worth far more than the same lead at half. The book shades toward the team the public is watching; the fair line doesn't watch anything.
Books move the live spread four or five points on a run. With 30 minutes left, fair usually moves two. When the crowd number outruns the math, the side the run hurt is often the value.
The two minutes a star rests move win probability more than the book's live number suggests if the bench unit has been bleeding points all season. The rates already know.
Late fouling adds points the pregame total never budgeted for. Fair totals account for endgame fouling when the margin is inside two possessions, which is why late overs cash more than intuition says.
Scores and clock come from the public ESPN feed; posted spreads, totals and moneylines are scraped live from a public sportsbook feed, including in-play numbers. Everything refreshes every 15 seconds. Fair numbers are the model's own. For entertainment. 21+ where required.