Trading Lab · Experiment brief
Sun Sep 6, 2026 · paper only · PT

Published 2026-09-06 19:02:20 CEST

Still testing Dropped

Sticky SMA vs fast 5/20: more “stick” didn’t buy more Bitcoin

After PR #4 we compared the new sma_sticky defaults (20/50, confirm 2, exit gap 1%) to the fast SMA 5/20 baseline and plain buy&hold on BTC. Sticky was meant to stay invested and grow units. On this sample it roughly matches hold on dollars, stays out ~44% of the time, and loses badly to fast SMA on both $ and BTC-units. Grid the knobs before killing the idea — don’t ship these defaults.

BTC-USD sticky 20/50 · c2 · g1% fast 5/20 $1,000 main 10bb811 · PR #4
Sticky vs hold (yrs)
3 / 7
Fast vs hold (yrs)
6 / 7
Sticky end units
0.127
Fast end units
0.612
Bar charts: ending equity, BTC units, and time in market for sticky, fast SMA, and buy and hold

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
Sticky 20/50$10,169−52%2456%0.127
Fast 5/20$48,827−40%7054%0.612
Buy & hold$10,558−76%1100%0.132
Units lens: Sticky ≈ hold on BTC you could buy at the end (0.127 vs 0.132) despite being flat nearly half the time. Fast prints ~4.6× the ending units — confirmation lag on sticky costs early bull entries.

Year-split · sticky defaults

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

Year Sticky end PnL Max DD Hold end vs hold
2020$2,678+168%−17%$3,868LOST
2021$1,092+9%−22%$1,563LOST
2022$607−39%−47%$388BEAT
2023$1,165+17%−26%$2,469LOST
2024$1,343+34%−37%$2,143LOST
2025$1,153+15%−14%$937BEAT
2026*$1,255+26%−14%$914BEAT

Sticky: 3/7 beat hold · avg PnL +32.8% · worst DD −46.7%. Fast refresh still 6/7 · avg +89.9% · worst DD −32.5%.

Year-split PnL bars and truth-serum summary sticky vs fast

Walk-forward (fixed params)

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

Sticky mean
+2.1%
Fast mean
+11.6%
Sticky beat hold
12 / 23
Fast beat hold
13 / 23

Sticky WF is a coin-flip vs hold and clearly behind the fast baseline. Softest story for a “stickier core” rule.

Keep · Tweak · Drop

Dropped
Sticky defaults as BTC core rule — loses to fast SMA on $ & units; ≈ hold $ while out ~44% of bars.
Still testing
Sticky idea family — confirm + exit-gap still worth a grid before killing (RESEARCH rule).
Lab pick
Fast SMA 5/20 as paper baseline — best of the three on this path; live core still TWEAK (sells often).
Lab pick
Buy&hold as core reference — 100% time-in-market; deepest DD (−76%).
Dropped
Go live — paper only; no explicit OK.

Five insights

Next build step

Grid-search sticky params; optionally try fast entry / sticky exit or shorter windows with a gap. Keep fast 5/20 as the champion to beat until a sticky config clearly wins BTC-units and softens dips.

Traceability