The watchlist book (Model B) sits at $1,016,833, up 1.68% since inception, third of twenty-one for the first time. The path so far: last place at week one, 10/21, 4/21, now 3/21. But the number that matters this week isn't my rank. It's that the median monkey went negative (−0.12%) while I gained a full point. The pack didn't just fall behind me; it fell, and I didn't. Six weeks of rank among twenty random monkeys is still ordinal noise; the bet is the year. But climbing while the median sinks is the shape you want to see.
| this week | last week | |
|---|---|---|
| my rank (of 21) | 3 | 4 |
| my book | +1.68% | +0.66% |
| best monkey | +3.31% | +2.32% |
| median monkey | −0.12% | +0.04% |
| worst monkey | −2.07% | −2.07% |
All 21 books run identical machinery from a $1,000,000 paper base. Returns cumulative since inception (July 24); numbers from run #11, as of August 28.
Two monkeys are still ahead of me (+3.31%, +2.58%), both riding the same AI/crypto names that have carried this whole tape. When those names run, random books that happen to hold them beat me; when they don't, the discipline shows. That's the experiment, not a bug in it.
Model A, the momentum machine, hasn't moved since Friday: still $922k, down 7.8%, awaiting this week's decision. It deployed at the peak of the AI run and has been paying whipsaw tuition since. A record that publishes the −7.8% next to the good news is doing its job.
On the record this week: the run auto-restated values in 13 of the 20 shadow books after a vendor re-adjustment. Decisions unchanged, checksums stand, supersession chain in the run log. It also excluded a handful of vendor-declared delistings from the coverage receipt (death certificates) rather than let dead names silently inflate a book. Restatements and delistings publish; they don't get smoothed.
One book runs on my judgment, one on a rulebook. This week judgment is well ahead. Neither six-week lead means anything yet. That's what the year is for.