Explainer

How The Drop is chosen and graded

The Drop is the single biggest edge between our projections and the bookmakers' player lines each night, chosen by the model, never by a person, and graded first on whether it beat the closing price.

How it's chosen

• Markets: points, rebounds, assists, threes and the combined markets.
• The line comes from Sportsbet or TAB when they price the market, and from the US market otherwise, labelled as such; the price has to be 1.70 or better, for a player with a settled role (no early estimates).
• Of everything that qualifies, the biggest edge wins. Edges above our strong threshold come first; if none clears it, the biggest one below it is the Drop.
• Opening week (20 to 26 October, US dates): no unders on projected starters. And nothing on a player's first three games back after his team played 20 or more games without him. Both were set before the season, because those are the nights our minutes numbers are least sure.
• If nothing qualifies, there's no Drop that night.

It goes up around 8pm Sydney time the night before the games, on The Drop page and on X, with the lines fetched moments before it's chosen. If the bookmakers haven't posted player lines by then, it tries again at 10pm and midnight.

What's recorded

At the moment it's posted: the player, the market and side, the line and price and the bookmaker, our projection and our probability. Just before tip-off: the closing line and price at the same bookmaker. After the game: the result, COVERED or NOT COVERED. One Drop per night, and it can't be changed or removed once posted.

Closing-line value comes first

Closing-line value (CLV) compares the price we posted with the bookmaker's fair price at tip-off, once its margin is taken out. If we posted an under at 1.95 and at the close the bookmaker's fair chance of the under was 54%, the Drop carried 1.95 × 0.54 − 1 = +5.3% of value. Positive CLV means the market moved our way after we posted.

If the bookmaker moved the line itself (say from 25.5 to 24.5), we carry the closing price back to the posted line using our own numbers, so a line move is counted, not ignored.

Why not just covered or not?

Over a few dozen nights, covered or not is mostly luck. A genuine 5% edge at 1.90 still misses close to half the time, and a run of ten can go either way. Closing-line value settles much faster: beating the close consistently is the clearest sign the numbers are ahead of the market. We show both, CLV first.

Voids and pushes

If the player doesn't take the court, the Drop is void, as it would be with the bookmakers. If the result lands exactly on the line, it's a push. Neither counts as covered or not covered.

The graded props

The Drop is one leg a night. Alongside it, the track record grades every player prop that clears a fixed set of rules, set down before the season and not changed until the first review: our chance beats the bookmaker's fair chance by 5 points or more; the projection sits at least 1.5 points, 2 points for points + rebounds + assists, 1 rebound or assist, or 0.5 threes clear of the line; the price is between 1.40 and 3.00; the player is active 90 minutes before tip, when the line and price are locked; one market per player; one flat unit each; and the same opening-week and long-absence guards as the Drop. The closing line is stored 15 minutes before tip for closing-line value, each prop is graded the morning after, and a player who doesn't take the court voids it. Nothing is removed. Return on stake is published once 100 props are graded; closing-line value is the headline before then.

Preseason

Drops during the preseason are a trial run: they're posted and listed, labelled "preseason trial", and never counted in the record. The record starts on opening night, 20 October (21 October AEDT). Lines come from Australian bookmakers where they price the player, and from the US market otherwise, labelled as such.