The Washington Wizards finished 6.10 points per game below their expected points output this season. That is, by any reasonable measure, a team the standings lied about. And they were not alone. Across three leagues, the gap between what teams earned and what they got has rarely been this wide heading into summer.
This is the season-end regression ledger. The final accounting.
Due a Correction
Brooklyn Nets (NBA): Brooklyn posted the ugliest Luck Index in the league at -63, with an xPoints gap of -5.80. Their underlying metrics, point differential components, shot quality, and opponent shooting variance, all suggested a team that should have won meaningfully more games than it did. Close-game variance buried them. The Nets lost 14 games decided by four points or fewer, going 4-14 in that split. Teams historically sitting this far below expected points tend to see significant improvement the following season, which means the market may be slow to adjust Brooklyn's win total line for 2026-27.
Washington Wizards (NBA): The Wizards' -6.10 xPoints gap was actually the largest raw number in the dataset, even if their Luck Index of -54 trailed Brooklyn's. Washington's offense generated quality looks at a rate that didn't match their scoring output, and their defensive metrics were less catastrophic than the final standings implied. The rebuilding narrative around this team is real, but the degree of losing was inflated. Bookmakers tend to anchor next season's projections on final records rather than underlying performance models, and that gap is where informed bettors find edges.
New York Mets (MLB): Here's where the conversation shifts from retrospective to actionable present. The Mets carry a Luck Index of -50 and an xPoints gap of -3.00, and unlike the NBA teams above, they're still playing games that matter right now. Their run differential and expected win percentage suggest a team performing roughly three wins better than their current record. Variance like this typically regresses over a full 162-game schedule. The Mets' Pythagorean record is notably stronger than their actual mark, and the lines imply the market hasn't fully caught up.
Living on Borrowed Luck
Montréal Canadiens (NHL): Montréal's Luck Index of 71 was the highest in this week's dataset, paired with a 4.40 xPoints gap above expected output. The Canadiens rode an unsustainable save percentage and power-play conversion rate to a record that flattered them considerably. Expected goals luck was overwhelmingly in their favor during a 20-game stretch from January through February. Now in the playoffs, that kind of overperformance can evaporate in a single series.
Buffalo Sabres (NHL): Buffalo's Luck Index of 66 is fascinating because their xPoints gap was only 0.40, meaning the luck showed up in game outcomes rather than raw point accumulation. They won an improbable number of overtime and shootout contests, converting close games at an elite rate without elite underlying numbers. The margin was razor-thin and largely manufactured by timing.
Oklahoma City Thunder (NBA): OKC's 5.90 xPoints gap above expected output and a Luck Index of 58 suggest the Thunder's historic regular season was, in part, variance-assisted. Their net rating supported a great team. Not a historically great team. The market priced them as near-locks for the title, but the data says that gap between very good and generational was filled by fortune rather than structure.
Sometimes the house wins. Sometimes the house just got lucky.
The Regression Window
In the NBA and NHL playoffs, regression doesn't need 20 games to arrive. A seven-game series is often enough for underlying quality to surface, which is why teams riding inflated records tend to face sharper corrections in the postseason than they ever would across a long regular season. For the Mets and MLB at large, the window is wider, typically 30 to 50 games before expected-win models and actual records converge meaningfully. The data doesn't predict when. It just keeps score.