Early Line Value
Early line value is how far the market moves toward our number after we post it. We take the market's price when our model publishes a pick and compare it with the last price before the game starts. If the price moved toward our side, the number is positive. If it moved away, it's negative, and we publish it just the same.
A worked example
Say our NFL model takes Over 44.5 on Kalshi, and when it posts, the Over is trading around 48 cents. That price is the market saying there's about a 48% chance the game goes over.
By kickoff the Over is at 51 cents. The market moved 3 points toward our side, so that pick's early line value is +3 probability points. If it had slid to 46 cents instead, the pick would read −2.
The game result doesn't enter into it. The Over can lose and still have moved our way before kickoff, and the reverse happens all the time. Early line value and the win-loss record are two separate numbers, and we show both.
How we measure it
On an exchange contract (Kalshi), it's our side's midpoint price at the last capture before the start minus our side's midpoint when the pick was made, in probability points. A midpoint is halfway between the best bid and the best ask, so it doesn't depend on which side of the spread you trade.
On a sportsbook line, we take the market's probability with the vig removed, so the two sides add to 100%. College basketball uses the median across books, from the first snapshot after our forecast freezes to the last one before tip, signed toward our number. The WNBA game-line record, and tennis on spreads and totals, measure in line points instead (a total moving from 44.5 to 45.5 is one point), and say so.
Every pick counts, including the ones that move against us. The figure on each record page is the average over all of that sport's settled picks (every frozen forecast, for college basketball), next to how many it covers.
Why every record leads with it
A win-loss record takes a long time to mean anything. Over a few hundred picks, a model with no skill at all can run hot or cold by enough to look brilliant or broken. Line movement shows up on every pick, a day or a few hours after we post, so it builds evidence much faster.
It also measures the thing we're trying to do. We publish early, before the market has settled. When the market keeps moving toward our number after that, it's later information agreeing with us. When it moves away, the market knew something we didn't.
It has limits. Positive early line value over a small sample can be luck too, and it doesn't pay anything by itself. Results are booked separately, to win one unit at the vig-free price of the side taken, and they sit beside the early line value on every record page.
Where to find it
- NFL record: game and prop contracts, in probability points
- College football record: picks and the withheld ledger
- College basketball record: every frozen forecast, by freeze
- WNBA record: game lines and props, by market
- Tennis record: by tour, market and book
In tight table columns it's shortened to ELV. Related terms: walk-forward validation (how a model is tested before it's trusted) and the EV methodology.