{"schemaVersion":2,"selection":"Current research tests up to sixteen standalone configurations and fifteen timeframe-compatible native strategies per window. Native slots combine rotating coverage with training-profile matches. Selected native strategies also rotate through source-audited variants of their own entry settings. StepGridScalp uses a bounded matched slot on eligible 5m or 15m windows, with higher timeframes assembled during replay. Standalone variants derive grid spacing, pullback thresholds and profit targets from training-only price movement and fees. Active strategies compete for a later cash-reset holdout and increased-fee stress. Exact duplicate training fills from the same strategy share one shortlist entry. Native overrides are checked against a paired baseline, with consistently inert settings retired. Selecting public examples after observing their historical outcomes introduces selection bias.","windowSelection":"Core markets alternate with liquid Binance USDT, USDC and BTC movers. Mover scans reserve slots for rising markets, falling markets and failed rebounds. An asset signal can select a different liquid quote pair. Replays and benchmarks use that pair’s own candles. Mover selection uses a current ticker, which can include the historical holdout period. This retrospective selection favors visible moves. Fifteen-minute candles receive the largest share. Training requires 2,000 candles. The 15m mover lane uses 1,056 holdout candles. Other lanes use at least 2,000 holdout candles. Window lengths vary with candle interval. Returns and benchmarks are measured in each pair’s quote currency. BTC cases start with the configured BTC balance. Older archived examples retain their original methodology.","tradingBehaviors":["range-harvesting","trend-participation","breakout","rebound","capital-protection","profit-and-outperformance"],"recentWeek":"Last seven complete days within the holdout, carried inventory marked to market. Recent features end within 48 hours of selection, gain at least 0.5%, remain positive at doubled fees, and close exits on at least two UTC days (three sell candles, five for range harvesting). Historical standouts have four slots. One exceptional sparse case is allowed.","catalogGates":{"minProfitReturnPct":1.5,"minExcessPct":1.25,"minStressReturnPct":0.25,"maxDrawdownPct":18,"minTrades":4,"minCompletedSells":2,"protection":{"minHoldingLossPct":4,"maxStrategyLossPct":2,"minAdvantagePctPoints":5,"maxStressLossPct":3}},"showcaseDiversity":"One featured case per base-asset episode when holdouts overlap by at least half of the shorter window. The catalog can retain cases in different quote currencies.","showcaseGates":{"minProfitReturnPct":3,"minProfitExcessPct":2.5,"minimumStressedReturnPct":1,"maxDrawdownPct":12,"requiresNearbyGainStability":true,"minSimulatedTrades":5,"minCompletedSells":2,"capitalProtection":{"minHoldingLossPct":6,"maxStrategyLossPct":1,"minimumAdvantagePctPoints":7,"maxStressedStrategyLossPct":1.5}},"poolDisclosure":"Research totals are cumulative from the local qualified library and may be incomplete after VM replacement. Overlapping windows and similar presets are not independent trials.","holdoutAccounting":"Each training and holdout test begins with initial cash. No positions are carried between these windows. This is not a continuously running simulated bot.","sensitivity":"Trend participants may instead retain at least 55% of the rise and at most 75% of market drawdown at both neighboring settings. Other featured simulations must stay nonnegative in return and advantage when their gain setting is moved by plus or minus ten percent on the same holdout. This is not independent validation. Catalog-only cases may lack sensitivity evidence.","comparison":"Buy-and-hold is calculated from the first to last candle close, before buy/sell fees. Simulated strategies use configurable fees but still omit spread, slippage, order-book liquidity and latency.","engine":"Vendored snapshot of the user-supplied Gunbot frontend simulator. Live Gunbot parity has not been verified.","binance":"Source: Binance Spot public REST klines where exchange=binance. No silent exchange substitution.","publicCandles":"Full OHLCV candles are NOT published; compact account equity and simulated trades are provided. Price samples remain disabled until data rights have been reviewed.","fields":["id","market","strategy","training","validation","settings","chart","trades","candleSource","assumptions"]}
