The Brooklyn Nets finished their regular season nearly six points below where the underlying numbers said they should have landed. A Luck Index of -69, the most extreme reading across the NBA and NHL this cycle. The standings told one story. The data told another. Now, with both leagues in the playoffs, the question for anyone watching the market is simple: how much of that gap carries forward?
Due a Correction
The Brooklyn Nets posted a Luck Index of -69, driven by an xPoints gap of -5.80. That means Brooklyn's expected performance, derived from shot quality, pace-adjusted efficiency, and four-factor differentials, pointed to a team roughly six points better in the standings than the one that showed up. Close-game variance, an ugly stretch of sub-40% shooting in clutch minutes, and some historically poor free throw luck in the fourth quarter all played a role. Teams that finish a full season this far below their expected output almost never repeat that level of misfortune the following year. For bettors tracking futures markets, Brooklyn's line likely still reflects the record, not the underlying production.
The Sacramento Kings clocked in at -53, with an xPoints gap of -4.30. Sacramento was competitive by the metrics for most of the season. Their defensive efficiency numbers improved markedly after the All-Star break, and their offensive rebound rate remained top-ten league-wide. Yet the wins didn't follow. Variance like this, particularly concentrated in games decided by five points or fewer, typically regresses. The market tends to be slow in pricing the correction.
The Washington Wizards finished at -52 with the widest raw xPoints gap of the three at -6.10. Washington's underlying numbers were not good, to be clear. But they were meaningfully less bad than the final record suggested. The Wizards lost 14 games this season by three points or fewer, a clip that strains any reasonable model of sustainable outcomes.
Six points below expectation for an entire 82-game season is not a rounding error.
Living on Borrowed Luck
The Buffalo Sabres posted a Luck Index of 64 despite an xPoints gap of just 0.40. That near-zero expected points gap paired with a sky-high luck reading means Buffalo's good fortune showed up in the margins: shootout results, save percentage spikes at opportune moments, and a PDO that ran hot through December and January. The underlying territorial numbers were essentially neutral. Bookmakers building playoff series lines tend to trust shot-share data more than standings points, and the Sabres' profile invites skepticism.
The Montréal Canadiens carried a Luck Index of 60 and an xPoints gap of 4.40, the largest overperformance in this group. Montréal's expected goals data didn't support their record at nearly any sustained stretch. Their goaltending ran well above career norms, and their power play converted at a rate that ranked among the highest single-season marks in franchise history. Those percentages tend to come back to earth.
The Boston Bruins landed at a Luck Index of 56 with an xPoints gap of -0.10. That's a fascinating case: a team whose expected points were basically right on the number, but whose underlying process, game state dynamics, and score effects suggest the path to those points was far luckier than the destination. Boston's 5-on-5 expected goals percentage sat in the bottom third of playoff teams.
A Luck Index of 56 on a team the standings say is fine. The data disagrees politely.
The Regression Window
In a seven-game playoff series, regression doesn't need months to arrive. One game of corrected shooting luck, one goaltender returning to career norms, one fourth-quarter stretch where the bounces even out, and a series flips. The NBA and NHL postseasons are where inflated records and deflated records meet real leverage. The market adjusts quickly once a series begins, but the opening lines for Round 1 often still lean on regular-season records. That's where the gap between luck and skill tends to show its hand. MLB, meanwhile, is in the phase where early-season BABIP and strand rate variance is creating its own set of false signals, a story worth tracking as sample sizes grow. We describe the numbers. What you do with them is your call.