The Blocks Prop Died Right on Schedule

The pick was Ausar Thompson under 1.5 blocks. It rose because the slate was weak and this was the one number that looked like it had a real cushion, at least on paper. The case was simple enough: a single-stat under, a model at 0.82, and a board with too many mediocre baseball props pretending to be safer than they were.

The signal did not fully agree. The review section and Perplexity both landed on Ausar, which is how the placed 3-loss experiment got there too. Claude wanted White Sox +1.5 instead, and ChatGPT took Cody Bellinger under 1.5 total bases. That split turned out to matter more than the headline consensus, because the baseball side was dull in the useful way and the blocks prop was volatile in the expensive way.

The result was a loss. Thompson finished with 2 blocks in Detroit’s 116-109 win over Orlando, so the placed bet missed by the smallest annoying margin available. The 3-loss martingale experiment does not reset here, it doubles again. Cade Cunningham also detonated his under with 54 PRA, which was a polite way for the same game environment to keep making the point.

The learning is not subtle. This was the 1st rebounds-under tracked this week, and it lost, 0 of 1. More important, a defensive counting-stat under is not automatically low variance just because the median projection sits below the line. Blocks are still spike stats. Meanwhile, the less glamorous baseball shapes went 3-1 on the day, with White Sox +1.5 winning outright and both Hoerner and Bellinger staying in the one-base bucket. Structural boredom outperformed model elegance.

The verdict is plain enough. The placed bet followed the final recommendation and Perplexity, and that alignment cost the experiment. Claude’s run-line preference aged better than the board’s favorite stat under.