What a week of falsification left standing

Editorial. Not a new prereg. Not live advice. One map of the cycles already on this log.

Thesis

Almost nothing directional survived real cost, chronological validation, and the “is this just a costume of the last paper?” test.

What did survive is a two-layer residue:

1. A thin next-bar fade on liquid USDT 1h (close location, same-sign streaks, Donchian, taker-buy mix). Statistically real, incremental to raw sign in places, cross-asset on ETH/SOL, and dead at 5 bps round-trip. Loc + streak carry the joint linear signal; stacking does not help.
2. Volatility clustering / activity markers — |r| AC1 is large and stable; shocks lift subsequent magnitude; US hours, weekends, volume, and |ΔOI| are a vol clock, not a signed book.

No PROMOTED cycle. No live signal.

Three buckets

A. Real but not tradeable

Close near the high fades a little, then fees eat it · Three same-sign bars fade a little, then fees eat it · Donchian breakouts fade a little, then fees eat it · Taker-buy extremes fade a little, then fees eat it · Four thin BTC fades are one costume; fees still win · ETH and SOL copy the thin BTC fade — fees still win

B. Vol / risk markers (WATCH)

BTC vol clusters, but the half-life will not sit still · Volume mostly tags the vol regime, not the next move · BTC open-interest swings mark vol, not the next direction

Related texture, not labeled WATCH-as-tradable: UTC hour effects are vol, not a premium · weekday is a vol calendar · after a vol shock, size persists.

C. Clean kills

Flagship + Hawkes · alt/BTC 2021 monopoly · compression→expansion (wrong way) · BTC→alt same-sign lead · range-location · gaps vanished · wick folklore · funding fade · basis OVERFIT · daily TSMOM · 5d extreme MR · 10-name XSMOM · ETH/BTC residual z-fade.

Links: Hawkes · 4h/1d flagship · alt/BTC · compression · lead-lag · range location · gaps · wicks · funding · basis · TSMOM · 5d MR · XSMOM · ETH/BTC residual

What not to build next

Another next-bar fade dummy on BTC 1h klines. Stacking the fade family. Squeeze-breakout direction. Hour or weekday mean-return books. Donchian continuation. Overnight gap fade on 24/7 klines. Funding/basis crowding under a 2020–23 P90. Alt/BTC ratio momentum. Daily TSMOM / XSMOM / 5d-extreme MR with a new (R,H,K). Another Hawkes governor on the same parent. BTC-shock → alt same-sign lead.

Honest open doors (each needs its own prereg)

• Residualized OI — |ΔOI| marks next |r|, but after prior |r| the increment is weak. New object: OI after path-vol.
• Regime-conditional |r| half-life — persistence is real; 29h vs 5h is not one number you can size on.
• Fee venues honestly ≪ 5 bps — loc fade is the least-wrong candidate to re-cost, not a new costume.
• Non-kline microstructure — trades/L2/queue, true halt gaps, funding print vs next perp path.
• 15m volume residual magnitude after |r| control — the only volume increment that did not immediately die on OOS.

Do not treat this map as a strategy.