dates: 335 trials logged K=116
head-to-head
H0 forecast level: IC +0.0105 (t=+1.12, p=0.1242, n=335, CI[-0.015,+0.037]) H1 forecast revision: IC -0.0035 (t=-0.39, p=0.6592, n=334, CI[-0.020,+0.012]) H2 forecast resid vs mom: IC +0.0161 (t=+1.78, p=0.0364, n=335, CI[-0.007,+0.041]) H3 acceleration: IC +0.0014 (t=+0.16, p=0.4322, n=333, CI[-0.012,+0.014]) H6 revision resid vs mom: IC -0.0055 (t=-0.65, p=0.7468, n=334, CI[-0.021,+0.009]) ctrl momentum126: IC +0.0089 (t=+0.59, p=0.2768, n=335, CI[-0.040,+0.059])
H4 dispersion regime
forecast-top-decile fwd, dispersion low: +1.436% (n=112) forecast-top-decile fwd, dispersion mid: +1.986% (n=111) forecast-top-decile fwd, dispersion high: +3.642% (n=112) high-minus-low means +2.206% (bootstrap p=0.0212) spearman(dispersion, top-decile fwd) = +0.129
H5 width sizing (forecast top-8/quartile)
equal-weight mean +1.953% Sharpe1.16 | inverse-width mean +3.936% Sharpe2.57
paired D +10.742% CI[+3.975%,+18.962%] p=0.0000 (n=11)
H7 market timing (mean forecast vs SPY fwd 21d)
spearman(mean forecast, SPY fwd) = +0.139 (n=335, boot CI [+0.018,+0.252]) timing rule (long if mean fc>0) minus always: D -0.340% CI[-0.892%,+0.305%] Sharpe 0.99 vs 0.96 (n=335)