Funding vs next perp is the same null as funding vs spot

Verdict: NO_EDGE · Asset: BTCUSDT USDT-M funding → next 8h mark · Sample: chronological 60/20/20 by funding event

Leftover from the spot-aligned funding kill: maybe the print fades the perp path, not spot. Frozen own discovery cuts (not the parent’s): P10=−1.658e−5, P90=+3.771e−4. Alignment: funding T → log(mark_{T+8h}/mark_T). n=4306/1436/1436.

What happened

Q1 val FAIL. HIGH n=57 mean −10.4 bp, MID +4.2 bp, LOW +20.2 bp. Signs look like fade, but t(H−M)=−0.62, t(L−M)=+0.79, ΔP fails (P(neg|HIGH)=0.474, wrong way). Continue also fails val and OOS. OOS n_high=0 under frozen P90 — same 2020–21 costume as the parent.

Q2: prior |mark r| does not salvage (val β_f t=−0.18). Mark vs spot 8h correlation = 0.9999. The leftover door is empty. Sub-bp mark−spot residual is the funding–basis identity, not a book.

Trading

Q3 not opened.

Final Verdict

NO_EDGE

Do not rebuild crowding-fade on mark instead of spot. You are measuring the same null twice.