← all research
VALIDATED

"Edge-hunt 46 — options-market clienteles (SKEW + CBOE put/call) vs SPY"

2026-09-28

·

SPY adj, signal = 252d z, lag-1 PIT, IS<2016<=OOS, 10bps/switch, Bonferroni bar p<0.00625 over 8 candidates.

Forward-return IC (block-permutation p, block 5)

signaltargetncorrperm plast date
SKEW1d5719-0.0280.0242026-09-25
SKEW5d5719-0.0170.4442026-09-25
P/C_total1d3131+0.0190.2662019-10-04
P/C_total5d3131+0.0680.0112019-10-04
P/C_index1d3131+0.0220.1992019-10-04
P/C_index5d3131+0.0460.0412019-10-04
P/C_equity1d3131+0.0050.7242019-10-04
P/C_equity5d3131+0.0510.0362019-10-04

Incrementality vs VIX level + momentum

signalcorr(VIX z)IC rawIC partial
SKEW-0.22-0.028-0.083
P/C_total+0.43+0.019+0.136
P/C_index+0.14+0.022+0.044
P/C_equity+0.44+0.005+0.132

Directional overlays (long/flat, matched-exposure placebo)

Adopt bar: placebo p < 0.00625.

signaldirscopeSharpebuy-holdexposureplacebo psurvives
SKEWlong z>0IS-0.240.4651%0.0460no
SKEWlong z>0OOS0.430.8851%0.0000no
SKEWlong z<0IS0.280.4649%0.0000no
SKEWlong z<0OOS0.430.8849%0.0000no
P/C_totallong z>0IS0.060.4633%0.0050no
P/C_totallong z>0OOS-0.130.8816%0.0390no
P/C_totallong z<0IS-0.560.4639%0.5950no
P/C_totallong z<0OOS-0.210.8819%0.1360no
P/C_indexlong z>0IS-0.200.4632%0.0555no
P/C_indexlong z>0OOS0.080.8817%0.0140no
P/C_indexlong z<0IS-0.430.4640%0.3905no
P/C_indexlong z<0OOS-0.580.8819%0.4315no
P/C_equitylong z>0IS-0.200.4634%0.0565no
P/C_equitylong z>0OOS-0.140.8815%0.0360no
P/C_equitylong z<0IS-0.260.4639%0.1275no
P/C_equitylong z<0OOS-0.190.8820%0.1515no

Verdict

REJECT — no candidate clears the Bonferroni placebo bar AND beats buy&hold. SKEW's weak negative 1d IC is not harvestable long-only.

Reading

  • Adopt only if a directional overlay beats buy-hold AND clears the Bonferroni placebo bar.
  • P/C series end 2019-10 (feed discontinued) so their OOS is short; SKEW is current and is the primary read. Runtime 25s.