Still testing Parked

Grid winner 20/15 beats hold — still trails fast SMA on units

btc_hold_protect with dd_pct=20, reentry_pct=15 rescues year-split to 5/7 beat B&H (defaults were 1/7) and grows ending units to 0.210 > hold 0.132. Walk-forward beat-hold 43%. Fast SMA 5/20 still wins the pile (0.612 units, 6/7 years). Park further dd/reentry research on this family for BTC.

BTC-USD dd 20 · re 15 fast 5/20 $1,000 main 5ad3750+

vs hold (yrs)

5 / 7

WF beat hold

10 / 23

HP end units

0.210

Fast end units

0.612

Full sample · $ and BTC-units lens

One continuous run: 2020-01-01 → 2026-09-05 · start $1,000.

Strategy	End $	Max DD	Trades	In mkt	BTC units
Hold-protect 20/15	$16,783	−46%	15	55%	0.210
Fast 5/20	$48,827	−40%	70	54%	0.612
Buy & hold	$10,558	−76%	1	100%	0.132

Units lens: 20/15 ends at 0.210 vs hold 0.132 (beats pile) and fast SMA 0.612 (~2.9× gap). Softer than hold on DD; calmer than SMA on trades (15 vs 70).

Year-split · grid winner

Each row restarts with $1,000 on Jan 1 (2026 = YTD).

Year	HP end	PnL	Max DD	Hold end	vs hold
2020	$4,363	+336%	−17%	$3,868	BEAT
2021	$1,736	+74%	−32%	$1,563	BEAT
2022	$913	−9%	−19%	$388	BEAT
2023	$2,363	+136%	−19%	$2,469	LOST
2024	$2,317	+132%	−19%	$2,143	BEAT
2025	$1,109	+11%	−25%	$937	BEAT
2026*	$901	−10%	−18%	$914	LOST

5/7 beat hold · avg PnL +95.8% · worst DD −32.1%. Compare: defaults 1/7 · SMA 5/20 6/7.

Walk-forward (fixed params)

Train 365 → test 90 → step 90 · 23 folds · $1,000 each.

Mean / fold

+14.3%

Mean end $

$1,143

Worst DD

−28.8%

Beat hold

10 / 23

Mean return tops defaults (+12.1%) and SMA (+11.6%). Beat-hold 43% still trails fast SMA ~57%.

Keep · Tweak · Drop

Still testing

BTC core rule (20/15) — beats hold on $ & units & 5/7 years; still ≪ SMA units; not obvious enough to sell core BTC on this alone.

Parked

Hold-protect family (BTC) — dd/reentry grid + stress/WF done. Stop knob spiral. Do not DROP the idea (documented soft-hold recipe).

Dropped

Defaults (25/10) as core — confirmed worse than 20/15 and worse than hold on units.

Lab pick

Fast SMA 5/20 paper baseline — still best ending_units (0.612).

Lab pick

Buy&hold core reference — 100% time-in-market; deepest DD (−76%).

Dropped

Go live — paper only; no explicit OK.

Five insights

Year-split 1/7 → 5/7 with 20/15 — knobs mattered; defaults were wrong, not the family.
Ending units 0.210 > B&H 0.132 — first hold-protect config that grows the BTC pile vs hold.
Still ≪ SMA 5/20 units 0.612 — soft-hold recipe, not the paper alpha winner.
WF beat-hold 43% (SMA 57%, defaults 26%); mean return actually highest (+14.3%).
Time-in-market 54.7% ≈ SMA; trades 15 vs 70 — calmer, closer to core style.

Next build step

Park hold-protect on BTC. Point research at a different family (or alts sleeve). Optional later branch: one-shot Fear & Greed fng_floor — treat as a new experiment if revisited, not more dd/reentry grids. Keep SMA 5/20 as the paper bar on ending_units.

Traceability

experiments/runs/stress_20260906T165910Z — year-split 20/15
experiments/runs/walkforward_20260906T165917Z — WF 365/90/90
experiments/runs/20260906T165938Z/ — full-sample HP 20/15 + SMA 5/20
experiments/runs/20260906T165939Z/ — full-sample B&H
briefs/2026-09-06-btc-hold-protect-dd20-re15.md · experiments/results.tsv