Every strategy version, in the open.
Strategies are pinned to semantic versions. Every change is reviewed, backtested, and listed here before it goes live.
v7LIVE2026-04-29ETH performance pass
Targeted improvement of how the strategy treats ETH, refining the parameters that govern entries and exits on that asset. SOL and BNB inherit unchanged from the previous version. Running in production in parallel with v6.4.
v6.4LIVE2026-04-19Uniform long-stop policy
Long-side stop-loss policy unified across all assets. Analysis on winning trades showed they rarely revisit the previous per-asset floor, so a single tighter level limits worst-case losses without sacrificing winners. Running in production.
v6.3RETIRED2026-03-15Per-asset tuning + fee-model correction
Scoring shifted to a recency-weighted geometric mean — recent market regimes matter more than 2020-21 outliers. Backtest fee assumption corrected to match the actual dYdX maker rate (the previous baseline was inherited from a fee-less spot exchange).
v6.2RETIRED2026-02-10Regime exit + early-loss management
Two new exit conditions: a regime-aware exit for LONG positions and an early-loss exit for trades that go against us right out of the gate. Together they meaningfully reduced out-of-sample drawdown.
Want the full detail?
Each version has a corresponding spec document in the bot's source repository. Per-parameter tables and backtest stats are tracked there.