Why grade on CLV instead of just wins and losses?

Short-term W/L is dominated by variance. CLV is the single strongest leading indicator of long-run ROI. A bettor who consistently beats the close will land in the green over a large enough sample, even if the next 50 picks run cold.

What closing line do you use?

Pinnacle by default. Pinnacle accepts sharp money and rarely limits, so their close is the cleanest market price available. If Pinnacle is off a market, the engine falls back to LowVig or BetOnline, both of which mirror sharp action with low vig.

When exactly do you snapshot the close?

Inside the final 15 minutes before lock for game-line markets. For PGA outrights and player props the snapshot fires at first tee-time or first pitch of the relevant slate, whichever applies.

Can old picks be edited after the fact?

No. The ledger at /history is append-only. Price, line, sportsbook, and lock time are written at the moment a pick goes live. After grading, win/loss/push and CLV land alongside. Nothing further is mutable.

Methodology

How we grade picks. CLV first, W/L second.

Every pick we lock is captured against the closing line. Closing-line value is the leading metric. Wins, losses, and units are the lagging ones. Public, append-only.

What CLV is

Closing-line value. The difference between the price we locked a pick at and the de-vigged price the sharp market settled on just before the game started.

If we lock the Lakers at +145 and the closing market implies the Lakers are a +120 underdog after the vig is stripped, we caught a number sharper than the close. That gap is CLV. It is the single strongest leading indicator of long-run ROI in sports betting. Beating the close consistently is what separates sharp bettors from lucky ones.

How the close gets captured

Inside the final 15 minutes before lock, the engine pulls Pinnacle's two-sided market on the exact selection. We remove the vig from the pair, which gives the market's true implied probability. That number gets stored on the pick record as the closing probability.

Pinnacle is the default because they accept sharp money and almost never limit. When Pinnacle is off a market, we fall back to LowVig or BetOnline. Both run thin-margin books that track sharp action.

The CLV math

CLV equals the closing probability minus the implied probability of the price we locked at. A positive number means we got a better number than the close. A negative number means the market moved against us before tip-off.

We average CLV across every graded pick and surface that number first on the public ledger, before win rate and net units. Both numbers matter. CLV tells you whether the process is working. W/L and units tell you whether the variance has caught up yet.

The methodology in one paragraph

Lock the price publicly. Snapshot the de-vigged close. Compute CLV. Grade against the ESPN final. Show every row, every grade, every closing-line comparison at /history. Append-only.

Grading

After the game, the engine pulls the ESPN box score and grades the pick. Wins, losses, and pushes land alongside CLV on /history. Filter by sport, by result, by confidence. Click any row to verify against the box score.

Nothing is editable after the fact. The price, line, sportsbook, and lock time are written when the pick goes live. CLV and W/L land at grading. No back-edits, no cherry picks.

What good CLV looks like

A long-run average above +1.5% on US sports is professional territory. Above +2.5% is rare. We publish the running average on every page that touches the track record. If it ever compresses to zero, the engine is no longer beating the close and the receipts will say so.

Open the ledger See the plans