How to read the Predictions board

The Predictions board at /picks lists every pick as it publishes, with the strategy behind it. Tabs split it into recent picks, picks on games in progress, closed picks from the last two days, and the strategies you follow. You can filter by league and market and search by team or strategy.

Each pick names a side, a line and a price, the strategy that made it, and its confidence. Open a pick for the full model view, or open its strategy to see that strategy's graded record.

The MVP Edge Score is a 0-100 summary of how good the opportunity looks. It blends the model's edge over the price, its conviction, the strategy's closing-line record, late line movement, the gap between tickets and money, market liquidity and injury or weather context. It is a display of signals we already publish; it never decides which picks publish.

Confidence is the model's stated strength, calibrated against its own graded history before it is shown, and grouped into high, medium and lean bands. A higher band means more agreeing signals, not a sure result.

Sorting by edge score or confidence puts the strongest-looking picks first. That is useful only if those picks have earned it. Before leaning on a high score, open the strategy and check its record and CLV in that league and market. A high score from a strategy with a weak or short record is still a weak signal.

Watch the price on the pick against the price you can get. The pick is graded at its posted price. If the line has moved since, the edge it showed may be gone.

The free plan shows one pick a day in full. Other picks appear locked: you see the strategy and the confidence band, but the side, line and price are hidden. Pro shows every pick from every strategy as it publishes.

Every pick, locked or not, is graded the same way and counts in its strategy's public record. Locking hides a pick from you before the game, not from the record after it.

Say the board shows two picks on different games. The first carries a high edge score and a medium confidence band, from a strategy with 90 graded picks in that league and market. The second carries a medium edge score and a medium band, from a strategy with 800 graded picks and positive CLV.

The first looks stronger on the card. The second is backed by far more evidence. Before treating the first as the better pick, open its strategy: 90 picks is a short record, and a high score on a short record is mostly the model's own view of itself.

The live tab shows picks on games in progress. They were posted before the start and are graded on the final result like any other pick; seeing them live is not an invitation to bet them in-game at a new price.

The closed tab shows picks settled in the last two days, with their results. It is a quick way to see what happened, but two days is far too short to judge any strategy. Use the Strategies board for that.

We do not currently publish game-total picks, so you will not find over/under picks on the board. Underperforming models may be withheld from public view, and their picks do not appear either. The board shows what our published strategies are calling, not every opinion our models hold.