Strategies

FMZ Single-SuperTrend 45/3 State-Machine Port on BCH OKX 15m

Outcome: Abandoned

BchOkxFmzSingleSuperTrend453

Outcome Summary

This strategy ported the FMZ SuperTrend 45/3 state machine to 15-minute BCHUSDT.OKX with 20%-of-equity sizing and no extra risk exits. The full-history backtest produced 5748 trades with a -0.375% average trade return, PF 0.66, Sharpe -1.94 and negative mean returns in every year, which confirmed QA's earlier sandbox concern (-0.466%, PF 0.556). The backtest-review gate abandoned it after one iteration for negative expectancy, and it never reached optimization, analysis or risk review. The verdict covers only this frozen BCH port, not the original BTC/USD strategy or SuperTrend in general. The reviewer also flagged ledger inconsistencies for the operator: trades continue after the 2021-02 liquidation, avg_holding_bars reads 0, and avg_position_pct is 33% against a 20% target.

Strategy report

Hypothesis

Port 善's explicit SuperTrend trading rules from the FMZ article, using 15-minute BCHUSDT.OKX linear perpetual bars instead of the historical BTC/USD inverse quarterly contract. Preserve period 45, factor 3, both trading directions, and exit-before-reentry behavior. Adapt the source's repeated aggressive two-tick limit orders to ordinary market orders at the signal bar's close; this preserves directional intent but does not reproduce its polling, cancellation or partial-fill timing. Replace the fixed one-contract amount with 20% of current equity in position notional, without leverage amplification. The source has no separate stop-loss, take-profit or time stop; do not add them. All six supplied figures were inspected: the cover and GitHub image add no trading rules; the comparison charts show BTC daily SuperTrend 10/3 overlays around the March 2020 decline; the parameter screenshot confirms pd=45, factor=3, Amount=1, TradeInterval=2000ms, PriceTick=0.5 and Symbol=quarter; the performance screenshot shows an unverified historical BTC/USD quarterly backtest. The trading experiment uses the article's explicit 15m and 45/3 settings, not the illustrative daily-chart settings or GitHub issue's 8/2.1 settings.

This is an explicit source replication with recorded market, execution and sizing adaptations, rather than a claim of a newly discovered mechanism. OKX is under its 5% venue target at 2.9%, and both directions address the portfolio's 78.7% long-only concentration. BCH avoids another BTC-series experiment, although corpus_concentration is empty and BCH's individual historical share cannot be verified from this brief. The single-symbol scope remains over target because adding a hedge or basket would alter the submitted trading rules. Unlike the failed LINK triple-SuperTrend port, this source uses one 45/3 indicator, symmetrical long/short flips, and no ROI ladder or trailing-stop overlay. Unlike the pending BTC ATR-channel proposal, it has no regime filter. Preserve these distinctions without treating them as evidence of edge.

negative_expectancy (fee-floor check: -0.375% vs the 0.15% OKX futures floor; about -0.27% before commission, so fees are not the cause). This rejects the frozen 15m BCHUSDT.OKX market-order, 20%-notional port of 善's FMZ SuperTrend 45/3 state machine. It does not reject the source's BTC/USD inverse-quarterly strategy or SuperTrend generally. Evidence: 5748 closed trades, PF 0.66, Sharpe -1.94, negative mean trade return in every year 2020-2026, executed-signal IC -0.005. QA's sandbox concern (-0.466%, PF 0.556) is borne out on the full history. Optimization cannot help because the contract fixes every parameter and the tunable set is empty. Operator note: the reported liquidation in 2021-02 while trades continue to 2026, avg_holding_bars 0, and avg_position_pct 33% vs the 20% target point to a ledger/impact-folding inconsistency that needs separate review. The per-trade price returns are negative regardless.

Implementation

Single 45/3 SuperTrend on closed BCH OKX 15-minute bars; direction-flip entries and exits, full closure before opposite reentry, 20% equity notional, no independent stop, target or time exit.

Preserves the source ATR seed, band recursion, equality conventions and directional transitions with constant-time scalar state. Implements the frozen research plan without extra filters. Source: https://www.fmz.com/digest-topic/5438. Profitability remains to be measured.

Verification Results

The truncated sandbox reports 471 closed trades, -0.466% average net trade return and profit factor 0.556. Full-history analysis should assess this negative expectancy; it does not establish a correctness defect.

Backtest Review

Sharpe: -1.94
Total return: -100.00%
Max drawdown: 100.00%
Trades: 5748
Win rate: 30.5%
Profit factor: 0.66

The trades match the mechanism: symmetric SuperTrend flips, 2874 long and 2874 short trades, exit before re-entry, and 20% equity notional at entry (peak entry leverage 0.20). The code implements the stated port.

The sample is large and reliable: 5748 closed trades over 2020-01 to 2026-09 with clean input data.

avg_trade_return_pct is -0.375%, against the 0.15% OKX futures floor. Adding back the roughly 0.10% round-trip commission still leaves about -0.27% per trade, so the gross edge is negative and fees are not the cause (commission is 11% of gross).

Profit factor 0.66, win rate 0.30 with payoff 1.50, Sharpe -1.94 (95% CI -4.17 to -0.01).

Average trade return is negative in every year 2020-2026: -0.75, -0.21, -0.50, -0.35, -0.31, -0.22 and -0.27%.

Executed-signal IC is -0.005 (t=-2.2) with negative gross edge. The signal carries no forecasting information.

The equity was liquidated on 2021-02-13 (reported -100%), yet trades continue to 2026. avg_holding_bars is 0.0 while average holding is 10h 11m. avg_position_pct is 33% against the 20% target, and $84k of impact is folded into equity. These ledger and metric inconsistencies need separate operator review. The verdict does not depend on them, because the per-trade price returns alone are negative.

Outcome Summary

Before paying for a backtest, check whether a fixed-parameter port of a source strategy (here moved from BTC/USD inverse quarterly to BCH 15m) still has a positive gross edge on the new instrument and timeframe, since a negative gross edge cannot be tuned away when nothing is tunable.

The backtest-review gate abandoned it for negative expectancy: -0.375% per trade against the 0.15% OKX futures floor, still about -0.27% before commission, so fees were not the cause; optimization could not help because every parameter was fixed and the tunable set was empty.

A faithful port of the FMZ article's single SuperTrend (period 45, factor 3) state machine to 15-minute BCHUSDT.OKX perpetual bars, trading both directions with exit-before-reentry, market orders at bar close, and 20% of equity as position notional with no added stops or take-profits.

Over 2020-01 to 2026-09 it closed 5748 trades (2874 long, 2874 short) with avg_trade_return_pct -0.375%, profit factor 0.66, win rate 0.30, Sharpe -1.94 (95% CI -4.17 to -0.01) and a reported -100% total return with liquidation on 2021-02-13; average trade return was negative in every year 2020-2026 and the executed-signal IC was -0.005 (t=-2.2).

Analysis

JavaScript version SuperTrend strategy

Analysis

SuperTrend · Issue #30 · freqtrade/freqtrade-strategies

SuperTrend策略（教学）

Backtest and paper results are hypothetical. Trading involves risk of loss.