[warmly] Hey, it’s Duke. I’m glad you’re here.
We are still in the deep preseason quiet — no games on the board until October 3rd, Miami at Toronto — and the futures market hasn’t budged since yesterday. No power movers. No new odds swings. Just… stillness.
And in that stillness, one number keeps humming: 41.5.
That’s the Elo surge the New York Knicks carried through the final stretch of last season. They finished 70–32 with a league-best 1808 Elo. Not just “hot.” Not just “playing well.” That’s championship-level territory in our system.
Before we dig into what that means — and what it doesn’t — let’s do our quick model check-in.
Over the full season sample, we tracked 1,246 games. Moneyline accuracy: 66.8%. Favorites were implied at 68.3% on average, and they won 66.8% of the time. That’s tight calibration. Our Brier score was 0.211 — which, in plain language, means our probabilities weren’t just picking winners, they were reasonably well-scaled.
Zoom in to the most recent 25 picks, though, and we’re at 60% accuracy with an average margin error of 14.8 points.
[thoughtful] And the interesting part is what’s driving that.
The latest self-audit showed the five biggest misses were all blowouts again. But the injury table wasn’t the culprit. Most of the listed absences had near-zero net rating impact — Jalen Brunson day-to-day at 0.0, Donte DiVincenzo at 0.0, Tristan Enaruna at 0.1. Only Aday Mara registered as high as 0.9.
And yet the actual margins were 24 to nearly 38 points.
In three of the four biggest errors, we actually picked the right team — just priced it like a competitive game. The one real directional miss was Detroit hosting Cleveland. We gave Detroit a 75.2% win probability… and Cleveland won by 31.
[sighs] That’s not an injury-weight problem. That’s a confidence problem.
The model is anchoring to “this should be tight” spreads in matchups where one side clearly had 30-point-outcome volatility baked in. So the adjustment going forward isn’t “bump the injury slider.” It’s being more skeptical of that 60–75% favorite band when the matchup profile says chaos is on the table.
Okay. Back to New York.
Over their final four tracked games, the Knicks posted a 128.8 offensive rating, 102.6 defensive rating, +26.2 net rating — at a 96.7 pace.
That pace number matters. They weren’t running teams out of the gym with chaos. They were controlling tempo and still crushing people. That’s surgical dominance.
On June 14, they closed with a 94–90 road win in San Antonio. That bumped them from 1801 to 1808 Elo. Across the full closing stretch, they climbed 41.5 points.
So here’s the part people sometimes misunderstand about Elo.
Elo is not a power ranking in the “vibes” sense. It’s a rating that updates based on expectation versus result. If you’re rated 1800 and you beat another elite team on the road, you don’t gain much — because you were already expected to compete. If you overperform expectation, you gain more. That late Knicks surge tells us they consistently exceeded what a very high baseline already predicted.
But — and this is the key — preseason Elo resets toward the mean.
They don’t start this year at 1808. They enter preseason at 1701.1. The market’s version of their rating is 1702. Almost identical.
[curious] So what happened to those 107 Elo points?
Nothing dramatic. It’s regression. Offseason uncertainty. Roster fluidity. The model asking, “How much of that peak is sticky?”
Right now, that translates to a 7.71% title probability. +950 odds. Third on the board behind Oklahoma City at 24.55% and San Antonio at 21.89%.
And there’s one big variable hovering over all of it: Jalen Brunson.
He’s listed day-to-day, but he’s set to undergo surgery on his right wrist and forearm area and is expected to need at least two months of rehab.
That doesn’t crater their rating — the injury module currently assigns him a negligible short-term net rating impact in the dataset we have — but contextually? Two months of rehab in August means you’re managing ramp-up into October.
[pause]
That’s not nothing.
Actually, hold on — let me pull up Sonar real quick.
[chuckles] Completely unrelated, but I love this one.
The Philadelphia 76ers’ name comes from a fan contest and refers to 1776 — the signing of the Declaration of Independence. The franchise used to be the Syracuse Nationals before moving to Philly in 1963.
I just enjoy how literal that is. “We are the 1776ers.” Clean. Direct. No ambiguity.
And Philly, by the way, is sitting as our top hype riser in the broader tracking context — even though futures haven’t moved day over day.
One more Sonar nugget because it’s preseason and we can afford it.
The LA Clippers weren’t born in Los Angeles. They started as the Buffalo Braves in 1970, moved to San Diego in 1978 as the Clippers, and didn’t land in L.A. until 1984.
[genuinely surprised] That franchise has traveled.
Okay. Big picture.
No power movers today. The title board is stable. Oklahoma City and San Antonio remain the clear top tier in probability. New York sits in that next band with Boston and Detroit trailing behind them.
And with no games until October 3rd, the last true on-court statement still belongs to the Knicks.
A 41.5-point Elo surge. A controlled-tempo +26 net rating finish. A peak at 1808.
The question now isn’t whether that happened.
It’s how much of it survives contact with a new season — and whether that late-season dominance was the beginning of something sustainable or the perfect storm of form, health, and matchup timing.
[thoughtful] That’s what preseason numbers are really about. Not prediction. Calibration of belief.
We’ll keep watching for any market tremors. We’ll keep tightening that 60–75% volatility band. And when the games start again, we’ll have a much clearer read on whether 1701 was conservative… or generous.
Thanks for hanging out with me. I’m Duke. I’ll talk to you tomorrow.