Eight Edge Bets, Eight Better Prices Than the Close
All eight graded Edge bets this season got a better price than the market closed at. They started 0-4 and are now 4-4. Here is what closing line value said the whole time, using our own ledger.
Eight Edge bets have graded so far this season. Every one of them got a better price than the market closed at. The first four lost. The last four won.
The record swung from 0-4 to 4-4. The price we got never changed: it beat the close every single time. That isn't a contradiction. It's the whole reason we track closing line value instead of just posting a record and asking you to trust it. If you want the full definition of CLV, we've written that up separately. This post is shorter: real bets from our own ledger, and what their CLV says that the final score doesn't.
The short version
CLV is the gap between the price you got and the price the market settled on right before kickoff. Beat the close and you got a number the market later agreed was too generous, whether or not that particular bet cashed. Lose the close and you got a worse number than the one that was actually available late, whether or not that bet cashed either. A single result mixes in a full game's worth of variance. CLV strips most of that out, because it's measuring the price, not the scoreboard.
That's why a service with a good week can still be a bad bet, and a service with a rough week can still be a good one. You can't tell which from one result. You need CLV, and you need more than eight of them.
The bets, side by side
| Game | Our price | Closed at | CLV | Result |
|---|---|---|---|---|
| DEN @ KC, Wk 1 (Edge pick) | DEN +130 | +113.0 avg | Beat the close | Lost |
| ATL @ GB, Wk 3 (Edge pick) | ATL +210 | +202.0 avg | Beat the close | Won |
| ATL @ NO, Wk 4 (Edge pick) | ATL +114 | about -105 | Beat the close, big | Won |
| NYG @ LA, Wk 2 | LA -350 | -318.25 avg | Lost the close | Won |
| ATL @ PIT, Wk 1 | PIT -178 | -270.75 avg | Beat the close, big | Won |
Beat the close and lost. The Denver game was one of our Edge picks, the small weekly subset the model flags as an actual bet, not just a read on the game. We took Denver at +130. By kickoff the same four books (DraftKings, FanDuel, BetMGM, ESPN Bet) had drifted that price down to an average of +113. That's real information: the market moved toward believing what we believed, after we'd already gotten our number. Denver still lost the game. The price was good. The result wasn't. Those are two separate facts, and only one of them is about our handicapping. The other three Edge picks graded so far this season (Miami +158 at pick time vs. +143.75 at close, Atlanta +124 vs. +122.75, Jacksonville +126 vs. +123.25) show the same pattern: positive CLV, and all three lost. Then Week 3: Atlanta at Green Bay, flagged after Thursday's injury report, at +210. It closed at +202 across the same four books, beat the close again, and this time Atlanta won 35-14. Week 4 went 3-0: the Giants at +130 (closed near +120), Jacksonville at +130 (closed near +122), and Atlanta again at +114 on Monday night, a price that had flipped to about -105 by kickoff. The live Edge is 4-4, +1.84 units. All eight beat their closing number. The results caught up to the prices; on eight games, that is still mostly variance doing what variance does.
Lost the close and won. The Rams were a heavy home favorite over the Giants in Week 2. We had them at -350. The market didn't agree as strongly by kickoff, settling at an average of -318.25, a number that would have been better for us if we'd waited. We got the worse price. The Rams won anyway, comfortably. This is the mirror image of the Denver game: a losing CLV number that still turned into a win. Neither the good price on Denver nor the worse price on the Rams predicted its own game.
Moved hard with us, and won. The clearest example this season is Atlanta at Pittsburgh, Week 1. We had the Steelers at -178. Over the following days the same four-book average moved all the way to -270.75, a nine-point swing in implied probability. That's the market catching up to a read we already had. Pittsburgh won. This is the case CLV is built to describe: an early price, a market that comes around to it, and a result that lines up. It's also just one game.
Why this matters more over 239 bets than over 8
None of these results tells you whether the Edge works. A six-season backtest does more of that work: priced at what four real sportsbooks were actually offering at the close (not a fair, no-vig number nobody could have bet), the Edge went 121-118 on 239 picks from 2020 through 2025, for a +17.9% return. That's a backtest figure, not a live one, and it clears a real statistical bar (its confidence interval doesn't cross zero) without depending on any single game in it. The 2026 live Edge is a different, much smaller sample: 8 decided picks, all eight with positive CLV, 4-4. Eight games can't confirm or break a strategy that took 239 to establish. It's a real, current data point, and it's exactly why we're not resting on the backtest number until the live sample catches up.
We'd rather say that plainly than let one number talk you into a story the sample size doesn't support yet. The same honesty applies to the rest of the model: the full slate of picks runs straight up at 66.5% across the 2020-2025 backtest, but the market's own closing number has been a better predictor of the final score than ours in every season we've measured, 2020 through this year, no exceptions. Against the spread, across the same 2020-2025 backtest, we're at 51.5%, under the 52.4% you need to clear the vig. We are not claiming to beat the market. CLV on the Edge picks is the one place we have real, if still early, evidence of a price edge, and we're showing you the losses right alongside it.
What to actually do with this
Don't grade us, or any picks service, on a week. Watch whether the prices we're taking keep beating the close over a real sample, then watch whether that turns into results over an even longer one. The Edge's backtest cleared that bar. Its live record hasn't had enough games to say anything yet. Both of those are true at once, and we'd rather tell you both than round it down to the more flattering half.
Every price and every closing line on this page comes from our own graded ledger. Check the methodology page for the season-by-season detail, or model performance for exactly how the full model compares with the market's own numbers on accuracy, against the spread, and on plain error.