Tuesday, May 5, 2026

Two models agree at 85-90% confidence and the data is overwhelming: home avg 22.0, road avg 25.1, season avg 23.7, vs DET avg 21.4. Every single split is 3+ points above the line. Haiku's under at 72% is indefensible given these numbers. Even against Detroit's strong defense, the margin is so large (5+ points above line in most splits) that Harden would need a historically bad game to miss. In a competitive playoff game, Harden will have the ball in his hands in crunch time and should comfortably clear 18.5.

Hachimura has gone OVER 11.5 in all 10 of his last 10 games, averaging 15.7 PTS over that stretch and 16.2 in L5. The line is set 4+ points below his recent average, creating massive edge. Even against OKC's elite defense, a player on this kind of hot streak with this much cushion should clear comfortably. GPT-4.1-mini's under at 70% is the most puzzling pick on the entire board — perhaps it's weighting an OKC defensive matchup adjustment too heavily, but 10/10 recent games over is near-impossible to fade.

Mitchell's L5 avg of 2.4 AST is almost half the line, and his L10 avg of 3.4 is still well under. This isn't a small sample anomaly — it's 10 games of consistent low assist output. Detroit's defense funnels pressure and limits easy kick-outs. Deepseek's 80% over is likely anchoring on Mitchell's season role as a primary handler, but the data screams that his playmaking has dropped off significantly in recent games. With Harden handling more creation duties, Mitchell's assists have dried up.

All three models agree. Hachimura has hit 2+ threes in 8 of his last 10 games, averaging 2.5 3PM over that stretch and 3.0 in L5. He's been on fire from deep and the 1.5 line is low for his current form. Even if OKC's perimeter defense tightens his looks, the volume and accuracy suggest he'll get at least 2. This is a high-floor prop with strong recent backing.

All three models agree with high confidence (78-85%). Smart's L5 avg of 5.0 and L10 avg of 5.5 are double the 2.5 line. He's been a primary facilitator for LAL and 7/10 recent games are over. This is a massive line discrepancy that should clear easily. Smart is a high-IQ passer who creates for others naturally, and the Lakers need his playmaking especially in a tough road environment where half-court execution matters.

Duren's L5 avg of 11.0 PTS is 4.5 points below the line, and his L10 avg of 13.3 is also well under. Cleveland's interior defense with Mobley and Allen is one of the best rim-protection duos in the league, making it extremely difficult for Duren to score efficiently inside. Deepseek's over at 60% may be banking on volume, but Duren simply hasn't been getting buckets recently and the matchup is terrible for him.

Mobley's L10 avg of 18.3 and home avg of 18.5 are comfortably above the line. With Mitchell struggling (20 PPG L5), Mobley could see even more offensive touches. His versatile scoring ability — both inside and from three (avg 1.4 3PM L5) — makes him matchup-proof. Haiku's under at 72% likely overweights a small L5 dip to 17.0, which is still above the line. The 1.8-point cushion across L10 is significant.

Harden's home avg of 8.4, road avg of 7.5, and vs-opponent avg of 8.6 AST are all 1+ assists above the line. As Mitchell's assists have cratered (2.4 L5), Harden is clearly the primary playmaker and should have ample opportunity to distribute. GPT-4.1-mini's 75% under is puzzling — perhaps weighting Detroit's defensive scheme, but Harden's passing genius finds ways regardless. His floor as a passer is extremely high.

Hartenstein's L5 avg of 10.8 PTS is 3.3 points above the line, a huge margin. With Jalen Williams OUT, OKC needs someone to pick up scoring slack, and Hartenstein should see more post touches and put-back opportunities. The Williams absence creates a usage vacuum that benefits Hartenstein directly. GPT-4.1-mini's under seems to ignore both the recent scoring surge and the injury cascade.

All three models agree. Cunningham's L5 avg of 7.0 and L10 avg of 8.1 are both well under 9.5. Cleveland's disciplined half-court defense disrupts passing lanes and limits easy assist opportunities. Even as Detroit's primary creator, Cunningham has consistently fallen short of this high line recently. The 4/5 under rate in L5 and 6/10 in L10 provide strong backing. This feels like a line set on his peak rather than his current output.