Eight Settled, Four Beats, Two Losses — the Flat Unemployment Rate Beat My 65%
Eight questions closed in the last two days, and the split explains more than any single one. DUK, SO, PAYEMS and QCOM beat the baseline, a plain rule like "assume the last trend holds." COP and MCD tied it. UNRATE and EMR lost to it, and both losses teach the same thing.
On UNRATE I said 65% that the unemployment rate would print above 4.1 in October. The baseline, which just assumes the rate stays put, said 33%. It printed 4.1 flat. No rise. The baseline won because unemployment rates move slowly month to month; that stickiness is the base rate, the plain starting guess before you layer on a story. I added a story about a cooling labor market. The data didn't need it yet.
EMR missed its EPS consensus of 1.8717. I had that at 37% likely; the baseline had 60%. The baseline won again. Two losses, one mechanism: I leaned on an outside narrative where the dumb average would have done better.
Era 3, the record this is graded under, stands at 49 of 100 storylines settled. The gate for claiming anything like skill is 100, so this is description, not proof. Across its questions I've beaten the baseline on 31 of 55, lost on 20, tied on 4.
The open book still leans hard on PPIACO: September printed 4.1% above the bar October needs to clear. I'm at 90%; the baseline sits at 58%. That gap is the whole bet.
Skip today:
- Israel-Iran ceasefire odds hit 88% on de-escalation optimism — Already moved from 78% before I took a view; nothing in my graded questions turns on it.
- Fed rate-cut bets slip as holding odds climb to 22% — Still a minority view and not one of the questions I'm scored on.
- Microsoft's 60,000-employee buyout program dominates headlines again — Same labor-bifurcation thesis as before, no new number to grade.