Skip to content

Phase B: futures hedge

Date: 2026-04-22 Goal: Building on Phase A (BTC handled by DCA, trend strategy only runs altcoins ETH/BNB/SOL), introduce futures long+short to hedge LUNA-grade bear risk with short positions.

HonestTrendFutures (extends HonestTrendGeneric)

Section titled “HonestTrendFutures (extends HonestTrendGeneric)”
can_short = True
stoploss = -0.08 # Futures must have hard stoploss (spot can use -0.99)
FNG_SHORT_BLOCK = 70 # Block shorts when FnG > 70 (never short euphoria)
leverage = 1.0 # 1x, zero liquidation risk

Short signals (mirror long):

  • crossed_below(ema_fast, ema_slow)
  • minus_di > plus_di
  • adx > threshold
  • volume > volume_sma
  • fng < FNG_SHORT_BLOCK (70) — critical filter, avoids short squeezes in bull markets

Short exit:

  • crossed_above(ema_fast, ema_slow)

Parameter inheritance (from HonestTrendGeneric)

Section titled “Parameter inheritance (from HonestTrendGeneric)”

EMA / ADX / min_hold / pyramid all inherited from the base class — same parameter set runs long+short (proven robust enough via WF validation).

Full history (2020-09-14 → 2026-04-21, 5.6 years)

Section titled “Full history (2020-09-14 → 2026-04-21, 5.6 years)”
MetricA5 spot long-onlyFutures L+SΔ
Trades559901 (508L + 393S)+342
Profit+166%+157%−9 ppt
Long profit+107%
Short profit+50%
Max DD14.77%7.84%−6.93
Calmar9.4518.72doubled
Sortino1.662.56+0.9
WindowA5 SpotFutures L+SΔ
W1 2018 crash+12.31%n/a (no futures)
W2 2019 accum+10.38%n/a
W3 2020 COVID+14.94%+9.91%−5.0
W4 2021 bull+143.30%+91.20%−52.1
W5 2022 LUNA−16.93%+47.62%+64.6
W6 2023 recovery+18.27%+18.71%+0.4
W7 2024 ETF+23.79%+17.15%−6.6
W8 2025 present−1.23%−6.43%−5.2

Key finding: in 5/6 “normal” windows, shorts are a slight drag (funding cost + a few losing shorts); but in the 1/6 crisis window (2022 LUNA/FTX), shorts turned −17% into +48% — a +65 ppt single-window tail-risk hedge.

Treat shorts as insurance with “~4 ppt annualized cost, 60+ ppt payout in bear markets”. The premium looks expensive, but black-swan events repay it 10× over. This is exactly the payoff profile tail-risk hedging should have.

FileDescription
strategies/HonestTrendFutures.pyL+S strategy class
configs/backtest/config_backtest_15m_futures_a5.jsonBacktest config
configs/config_dryrun_honestfutures15m.jsondry-run config, port 8084
scripts/start_honest_trend.shAdded futures and all modes
Terminal window
./scripts/start_honest_trend.sh futures # start alone
./scripts/start_honest_trend.sh all # run alongside dryrun + mtf
  • 8082 — HonestTrend15m (spot long-only, A5 alts)
  • 8083 — HonestTrend1mMTF (spot 1m long-only)
  • 8084 — HonestTrendFutures (futures L+S)
  • 8081 — live bot (1m, BTC+ETH+BNB, old — not yet switched to A5)

Enabled (injected via FREQTRADE__TELEGRAM__TOKEN env var) — will receive notifications from HonestTrendFutures-DRYRUN in the same chat.

  • Funding rate real cost matches backtest (~4 ppt annualized)
  • Short entries trigger in bear markets (FnG filter works correctly)
  • Stoploss -8% is not frequently pierced by wicks/gaps
  • 1x leverage truly has no liquidation risk
  • Pyramid rules behave correctly on the short side
  • Drawdown stays in the backtest 7.84% ballpark
  • DD 7.84% leaves 10+ ppt headroom to 20% kill-switch
  • Theoretically 2x doubles profit
  • But liquidation risk enters: a 20% wick with 2x leverage = liquidation
  • Recommendation: after dry-run stable, backtest 2x first, walk-forward confirm W5 LUNA doesn’t liquidate
  • Currently shorts allowed when FnG < 70, no distinction between 30 vs 60
  • Could try: stake × 1.5 when FnG < 30 (panic-bottom short size-up)
  • Risk: getting crushed on V-shaped reversals
  • Currently doesn’t check funding rate
  • Extremely high negative funding (shorts get extra penalized) should avoid opening shorts
  • Could query last 8h funding in confirm_trade_entry and reject order if > threshold
  • No XRP/DOGE in altcoin futures pool (backtests already showed these amplify DD in spot)
  • No BTC in futures pool (BTC uses DCA, keep them separate)
  • No leverage above 2x (1x→2x already doubles risk, 3x+ is gambling)

If dry-run 2 weeks performs worse than backtest expectations (DD > 12% or short profit negative):

  1. Stop futures bot: pkill -f HonestTrendFutures (emergency_stop.sh also takes it down)
  2. Keep running A5 spot long-only (no short insurance)
  3. Record failure reason, update the “failed experiments” section of this doc
  1. Shorts don’t earn in normal times, they save lives in crises: drag in 5/6 windows, lifesaver in 1/6
  2. 1x leverage is the “safest” futures starting point: no liquidation, pure directional alpha
  3. FnG filter is critical: allowing shorts at bull-top will get you squeezed beyond belief
  4. Funding is a hidden tax: 4-10% annualized, most painful in ranging markets