The Edge That Wasn't
All 81 experiments › #47

#47 — PantasticSMA indicator screen

2026-05-15 to 2026-05-18 (Sessions 059-060)
Verdict: negative · ⛔ closed negative — visual indicator, not an alpha source
A cTrader momentum-colored SMA signal screened across tick bars, M5, H1, multi-timeframe confluence and P&F level series finds no deployable edge anywhere.

Hypothesis — what it tests

Does the smoothed-SMA momentum-rate signal (PantasticSMA) carry directional edge at any bar type, timeframe, multi-TF confluence, or on P&F-filtered price?

Description

A port of the user's cTrader PantasticSMABot — SMA(7) whose momentum rate ((SMA[t]-SMA[t-lookback])/(lookback x bar_minutes), the timeframe-neutral 'WhiteMethod') colors the line and drives Long/Short/Flat signals — screened for genuine backtest edge in fx-core. Session 059 tested 30-tick bars on EUR_JPY and GBP_JPY (no edge in either direction; the momentum-normalization exit produces tiny winners and large losers) and EUR_JPY M5 over 5.5 years, where the only positive config (thr=0.030 trend-follow, IS PF 1.44 / OOS PF 1.29) fired just 14 OOS trades in years — not deployable, though it showed extreme M5 SMA momentum does weakly predict continuation. Session 060 tested all 15 multi-timeframe confluence combos (S5/S30/M5/H1) — every combo including a fast TF was deeply negative from spread churn, and M5 alone topped out at +1.42 p/d OOS — plus 36 P&F level-series configs (3 box x 6 reversal x trend/counter), of which 0/36 were OOS-positive because the SMA-on-step-function fires in the direction of the just-completed move (9-16% WR chasing entries). Research closed; the only imagined residual use was as an untested filter on other entries.

Key result

Best standalone: M5 +1.42 p/d OOS (below significance); tick bars -570 to -1950 p/d at 5-18% WR; only positive M5 config (thr=0.030) = 14 OOS trades over years; confluence: 0 improvement, S5/S30 combos -50 to -1,756 p/d; P&F: 0/36 configs OOS+ (best -12.51 p/d, 15.3% WR)

Indicators

PantasticSMA (SMA(7) momentum rate)SMAP&Ftick bars

Algorithms

grid searchIS/OOS validationmulti-timeframe confluence votingNumba JIT backtest

Code

Interactive version (search, filters, figures): the experiment explorer.

← #46 FIFO 2-box trail sweep (X3c_2_5)#48 Multiscale shock Phase 10 + joint timing-direction model →
This page is one experiment from the audited record behind the book The Edge That Wasn'tget the book · about the project.