The Lab · Reading the market · updated 2026-07-27
The closing line — the number that grades everything
The last price before a game starts is the sharpest number in betting. Here's why every call should be measured against it, and what we got wrong doing exactly that.
A betting line is not one number. It's a number that moves — opening days before the game, drifting as money arrives, jumping when a lineup card drops. By first pitch it has been argued over by everyone with an opinion and a bankroll.
That final version, the last price posted before the game starts, is called the closing line. It is the single most useful number in sports betting, and almost nobody looks at it.
Why the last price is the best price
Every price before the close is working with less information. The closing line has absorbed the starting pitcher, the injury report, the weather, the sharp money, and the public money. It's the market's final answer with the most evidence it will ever have.
That makes it the natural yardstick. Our own board is the receipt: across 266 graded games, the market-derived favorite won 54.9% of the time while the market's own de-vigged consensus claimed 55.8% — a gap of about one percentage point. The market keeps its word to within a rounding error. When something is that well calibrated, you don't argue with it. You measure against it.
Beating the close
Here's the move that matters. Suppose you bet an underdog at +150 in the morning. By first pitch the market has settled on +130 for that same team.
You didn't get lucky and you don't know yet whether you won. But you bought the same outcome at a better price than the sharpest number of the day. +150 pays $150 on a $100 bet; +130 pays $130. You captured a price the market later decided was too generous.
That gap is closing line value — CLV. Positive CLV means you consistently got better numbers than the close. Negative CLV means the market moved against you, and you were the slow money.
Our one qualified pick so far graded at +16.43% CLV. That's a large gap. It is also one bet, which is the next thing worth understanding.
Why CLV matters more than your win-loss record
This is the part that surprises people, so here is the arithmetic plainly.
Suppose you have a genuine 3% edge — excellent, professional-grade, the kind that compounds into real money. How many bets before your profit-and-loss record could prove it, as opposed to being a lucky or unlucky streak?
Roughly nine thousand bets.
That isn't a typo. Betting outcomes are close to coin flips; each one is worth about ±1 unit of noise while the edge you're hunting is worth 0.03. The signal is tiny and the static is enormous, so it takes thousands of results before one separates from the other. At a hundred bets, a true 3% edge and a true 3% loss look roughly the same. At a few hundred, a losing bettor can be comfortably up and a winning bettor comfortably down.
Which means: for the first several years, your win rate cannot tell you whether you're any good.
CLV can, much sooner. It doesn't wait for the game. It scores each bet directly against a benchmark — the closing number — instead of inferring quality from a binary outcome. You get a clean reading per bet rather than a noisy one, so the picture sharpens in dozens of bets instead of thousands. Sharp bettors track it obsessively for precisely this reason, and books limit accounts over it long before those accounts have proven a profit.
What we got wrong
If the closing line is the yardstick, you have to actually keep it. We didn't, and it's worth saying exactly how.
Our system polls sportsbook odds every thirty minutes — including while games are being played. In-game prices are wild by nature: a team up seven runs might be −2500. Those prices were landing in the same table as the pregame ones.
Three consequences, all found this week:
- Finished game pages displayed in-game odds. A completed game showed its favorite at −2500 next to a pregame call of 55.3%, as though anyone could have bet that.
- Our storage cleanup deleted the wrong row. To keep the database small, we collapse each finished game down to one price per book. That routine kept the last price seen — which for any game polled during play was an in-game number. It was throwing away the closing line and keeping the garbage.
- So our CLV coverage had holes, and the average was computed over whichever games still happened to have a closing price.
All three are now fixed: odds capture stops at first pitch, cleanup keeps the last pregame price, and finished game pages show closing lines, labeled as such. The damage that can't be undone is the closing data already deleted for older games, so our own record page now reports how many games have a closing benchmark rather than quietly averaging the ones that do.
We'd rather publish that than let a number look sturdier than it is.
What this doesn't tell you
CLV is the best early signal available. It is not proof of profit, and we haven't proven it for ourselves — "does closing-line value actually predict long-term profit, or is it just the market grading our homework?" sits unanswered at the top of our open questions, and it will stay there until we have a sample worth citing.
Two honest caveats. Beating the close on a stale line at one slow book is easier than beating it consistently across a market, and a bettor can post fine CLV and still lose money to bad sizing or fees. And our own CLV sample is currently one graded pick — interesting, not evidence.
What you should take from this: when you evaluate anyone's betting record, including ours, ask what price they got relative to the close. A tout showing you wins without prices is showing you the least informative version of their record.
Model output is informational and entertainment content, not betting or financial advice. If you bet, bet what you can afford to lose.