Zero. That is our wins on STRONG LEAN totals through three nights of the 2026 WNBA season. Three. That is our losses. All three were UNDERS. The first one missed by 33 points. The second by 6. The third by 27. The first and third were not close. The middle one was a coin flip we still lost.
This is a smaller sample than I want to write a postmortem on. It is also a 0-3 with two blowouts and the same directional miss every time. The model is wrong on totals in a specific way and the way matters.
Here is the pattern. The model is taking 2025 pace numbers as priors. Indiana ran 80 possessions per game last year. Dallas ran 82. The model uses those numbers in its total projection. On Saturday May 9, Indiana and Dallas ran 84 and 86. That is a four-possession increase per team. Each extra possession at WNBA scoring efficiency is roughly 1.1 points. Eight extra possessions is roughly nine extra points. The model said 167 total. The actual game went 211. The pace shift alone explains about half the miss.
The other half is shooting variance. The DAL at IND game saw both teams shoot above their 2025 effective field goal percentage. Combined eFG% on Saturday was 51.4 against a combined 2025 average of 49.0. That is the kind of variance you cannot model on a single game basis. It is also not the same problem as the pace shift.
LVA at LAS on Sunday was the same architecture. Vegas ran 96 possessions. The Aces averaged 95 in 2025, which sounds in line, but Los Angeles was a deliberate slow team in 2025 (78 pace). The model averaged the two and projected a slow game total. The Aces and Sparks combined for 86 possessions, 4 above the projected average. Combine that with Vegas going on a 30-15 third quarter run and the total cleared 176.8 by 6.
The third loss was DAL at IND on May 7. Same teams, similar architecture, similar miss. Three-data-point trend.
What is happening to WNBA pace in 2026. The league average pace through opening week is about 84.5 possessions per game. The 2025 league average was 80.8. That is a 4.5 percent increase, which is enormous in a single offseason for a basketball league. Two contributing factors:
First, the schedule. The 2026 season added four games per team (40 to 44). Coaches are running shorter, faster rotations to manage minutes across the longer schedule. Faster rotations mean faster pace.
Second, the talent. The Liberty added Sabally and play four ball-handlers. Vegas added Loyd and Gray. Indiana plays Caitlin Clark, who plays at a tempo no team in the league matched in 2025. Atlanta added Reese, who is one of the fastest-running bigs in the league. The top of the league is getting younger and faster, and the league pace is following.
Our model does not know any of this yet. It uses 2025 pace as the prior and 2025 efficiency as the multiplier. Both are too low for 2026. UNDER picks are fighting that current.
What we are doing about it. Two changes effective Wednesday.
Pace adjustment. We are applying a +4 percent league-average pace correction to every total projection until the rolling 2026 pace catches up. That moves a 167 projection to about 174, which would have flipped the DAL at IND UNDER to a PASS instead of a STRONG LEAN. We will tighten the correction as 2026 data accumulates.
Total tier downgrade. Until the pace correction is calibrated, no STRONG LEAN totals get full stake. STRONG total picks become MODEST. MODEST total picks become PASS. We are essentially shutting off the totals book for two weeks while we get more data. Spreads are unaffected.
Cumulative through three nights stays at 4 wins, 7 losses on the public scoreboard. The STRONG ledger is 2-4. The STRONG spread book is 2-2 and looks healthy. The STRONG total book is 0-3 and it is the entire reason the cumulative is below break-even. Killing the totals book until the pace catches up should pull the model back to roughly even on what we publish, even if the spreads stay 50/50.
Wednesday brings five games on the slate including the Tempo home opener. Game Plan Cards still publish for every game. Total verdicts will still appear on cards. But every total this week gets at most a MODEST label until further notice. Bet accordingly.
The number that matters this week is whether spreads keep producing edges while totals get held in time-out. If yes, the model has a fixable bug, not a structural problem. If spreads also start losing, we have a different conversation. So far the spreads are working. We are sizing them up. The totals are not. We are sizing them down. That is what discipline looks like in the second week of a season.
[ End Report ]
Share This