Crucible research — separate candidate library
Validating profiles. Existing confirmation is required to switch.
Main imported-config backtests: Active, Selected, All Imported, Active + Selected and Each. Preserves the normal global regime gate; additionally blocks longs on the asset’s bearish votes and shorts on its bullish votes, using its previous completed 1-hour candle and Updated S6 profile. Live settings are separate. Research labs retain their own rules.
These values remain active when this tab is closed.
Loading validated profile details…
Original: Bull 4 · Bear 4 · Conflict 1 · Band 0.45% · ADX 18 · Momentum 0.15%. Updated: Bull 3 · Bear 5 · Conflict 1 · Band 0.15% · ADX 18 · Momentum 0%.
These parameter profiles match the canonical Hydra implementation. Performance from the HL study is not an Aster result. Switching updates the Aster live detector and future backtests together. Running backtests retain their launch profile.
| Asset | 4H Last Bar | 30m Last Bar |
|---|
No update run yet.
Import all positive IS results replaces staging from the selected optimizer run, across all patterns and directions, without OOS or row-count filters. Source identity and fixed Custom parameters are preserved. Import does not activate live trading.
Standard Selected, Each and All Imported research runs use each configuration’s requested leverage without an exchange or optimizer maximum. New backtests use one-minute Aster execution (historical Binance fallback). Sealed Custom, SMA adapter and BTC comparison runs currently require a separate minute migration; they cannot silently run coarse execution. Standard RSI and ordinary Score6 remain supported; active-config comparisons and bot activation keep their live limits.
Backtest Active Configs uses the live active config snapshot, current global regime rules and fixed live sizing. The selected date range and optional own-asset S6 veto apply; research-only SL and portfolio overrides do not. Baseline adds no own-asset veto, independent of the live selector. Backtest Active + Selected adds the checked imports to that active book in one portfolio replay, using their stored timeframes and the same active-base risk/global regime rules plus the selected own-asset veto. It does not send or activate configs; a later Send workflow may replace an existing config rather than make the same additive change.
Backtest Each runs each config alone through Backtest Selected, with no baseline. Dates and settings are captured at launch. Current visible tab uses chosen timeframes; All stored imports uses each config’s stored timeframe. Individual net PnL, WR, completed trades and net $/completed trade are separate from optimizer scores and are not a combined portfolio. Keep this page open while the queue runs. Saved results stay in this browser and show their tested settings; changed settings require a new run.
No individual queue running.
Click Show Unique Imports to compare all imported rows with active live configs.
Select unique imports to send their exact stored settings. The server rechecks the comparison before activation. Edit configs in another tab, then refresh this comparison.
| # | Timeframes | Pattern | Direction | Lev | SL | TP | RSI | PT1 Ratio | PT1 Size | PT2 Ratio | PT2 Size | BE PT1 | Optimizer OOS WR | Optimizer OOS Trades | Optimizer IS PnL | Vol Z | Ladder | Evolved Params | Regime | Windows | ★ | Bot | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| No configs imported yet. Click "Import" to fetch from optimizer. | ||||||||||||||||||||||||||||||||
| # | Asset | TF | Pattern | Direction | Lev | SL | TP | RSI | PT1 Ratio | PT1 Size | PT2 Ratio | PT2 Size | BE PT1 | OOS PnL | OOS WR | OOS Trades | Vol Z | Ladder | Regime | Added | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| No candidates yet. Star configs above to add them. | |||||||||||||||||||||
Loading...
Loading...
Run a backtest to record minute execution and source coverage.
Available after a new Backtest Active Configs run. Deactivation disables future entries and does not close existing positions.
| Select | # | Source | Asset | TF | Pattern | Direction | Lev | Pos% | SL | TP | TP:SL | PT1 Ratio | PT1 Size | PT2 Ratio | PT2 Size | BE PT1 | Trades | WR | Total PnL | Avg/Trade | Vol Z | Ladder |
|---|
Closed trades grouped by exit month; full final PnL booked at exit. Earlier reports grouped by entry month. Partial-exit cashflow timing is not modeled.
Target hits: count / all closed trades (%), grouped by final exit month in UTC; open trades excluded. PT1/PT2 use recorded partial-hit flags. Final TP requires a recorded final-target exit at or beyond the target price. TOTAL combines counts across months, not an average of monthly percentages.
| Month | Trades | WR | PT1 hits | PT2 hits | Final TP hits | Net PnL | Start Balance | Lowest Balance | Closed-trade DD | End Balance |
|---|
| Month | Trades | WR | Net PnL | 4H PnL | 30m PnL | Start Balance | Lowest Balance | Max DD | End Balance |
|---|
Drawdown books final trade PnL at exit and nets simultaneous exits. Partial-exit cashflow and unrealized drawdown are excluded. Older reports without exit-time equity show —.
| Rank | SL Reduction | Effective SL | Status | Net PnL | Δ PnL vs Original | Ending Capital | Trades | Wins / Losses | Win Rate | Avg / Trade | Profit Factor | Closed-trade DD | Δ DD vs Original | Return / DD | At 0.50 ATR Floor | Runtime |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| No SL lab run yet. | ||||||||||||||||
| Select | Rank | Exit Policy | Status | Overall PnL | Max DD | PF | Payoff | Pre-split PnL | Retained | Recent PnL | Recent Δ | Recent DD | Recent WR | Recent Payoff | Recent PF | Worst Day | SL Trades | MAX_HOLD PnL | Policy Events | Zone Exits | Basket Closes | Mature / Excluded | Gate Skips | Runtime |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| No Risk / R:R lab run yet. | ||||||||||||||||||||||||
Optimize six detector inputs across 22,295 combinations. Native Aster trades, the captured active Aster configurations and neutral SMA fallback stay fixed.
$7,500 capital · 10% per position · maximum 5 positions · no compounding. Train January–March 2026; rank the training top 20 in April–May validation. Lock one winner before June–August confirmation. This confirmation period has been examined in earlier research. Every window reserves its final 144 hours for exits.
Grid: bullish/bearish votes 1–6; conflicting votes 0–5 below both thresholds; sideways band 0.15–0.75%; ADX 12–30; momentum 0–0.30%. At least 10 trades required in training and validation. Highest PnL wins, with lower realized maximum drawdown breaking ties.
Loading detector research settings…
| Strategy | Stage and dates | Net PnL | Trades | Realized max DD | Realized max DD % | Six detector inputs |
|---|
Stages: January–March training ranks the full grid; April–May validation chooses and locks one winner; June–August confirmation evaluates that locked winner. Validation ranks 2 and 3 have no confirmation results by design.
Drawdown uses realized native trade cashflows, not an intrabar marked-to-market estimate. Research results do not change the live detector or bot.
Keep original Score6 active regimes unchanged. During neutral periods only, use confirmed BTC 1h Supertrend ATR 10 / factor 1.8 to choose long or short territory. Preserve the native trend/sideways subtype and require matching trade direction, including Custom “all” configurations. Existing exits stay unchanged.
Captured active Aster configurations · $7,500 capital · 10% per position · five slots. January–March supplies prior history; April/May are chronological validation folds; June/July/August are reserved confirmation folds. September is a separate partial-period description. No parameter fitting or tuning occurs.
Each fold starts flat with $7,500 and stops new entries 144 hours before its cutoff. Aggregate PnL is the sum of reset folds; aggregate drawdown is the worst fold, not a continuous-account drawdown. These historical periods have already been viewed in earlier research.
Loading fixed research settings…
| Arm | Fold | Net PnL | Trades | Maximum DD USD | Maximum DD % |
|---|
Research only. This panel does not update the live detector, configurations or trading book.
Search combinations of shared-pattern variants together in the portfolio. The 70 unique configurations stay fixed. Results show the best combination tested within the search budget, not a guaranteed global optimum. Research only; this panel does not change live configurations.
Uses the shared Backtest settings: dates, capital, position size, maximum open positions and regime detector. Uses native Aster data and fixed fees/slippage, with no compounding or progress cap. Auto and Score6 are supported; Force/manual regimes are not. Each saved search records its settings; reopening it leaves the current form unchanged.
Native signal simulations use the selected GPU; exact sequential portfolio combination evaluation runs on CPU. Ranking uses full-period portfolio PnL. Pre-split and recent-period results are descriptive breakdowns, not independent holdout validation. Disabled variants are included only as research candidates when “All 152” is selected.
No search loaded.
| Rank | Combination | Portfolio PnL | Max drawdown | Trades | Enabled variants | Exact variant details |
|---|---|---|---|---|---|---|
| No completed search yet. | ||||||
Baseline: current Score6. Treatment: closed BTC 1h Supertrend. Green: longs only. Red: shorts only. EMA200: reference only. Existing exits remain in use. Permission persists through later same-color hours. Each strategy still needs its own entry signal; a flip never forces an existing position to close.
Minute migration: the sealed BTC comparison is currently unavailable; ordinary Standard Score6 backtests remain supported.
Fixed preset: ATR 10 · factor 1.8 · BTC 1h · diagnostic EMA 200 · ADX off. Both arms use the same captured source and native Aster candles. The ordinary regime dropdown does not change this comparison. Research only; no promotion or Send to Bot action.
TradingView source: Bitcoin SuperFlip by blitz_locked. Hydra adapts its line color to persistent permission; this comparison does not reproduce the source strategy's EMA-filtered orders.
Choose immutable stored rows for this experiment. Standard, qualified Custom and staging Custom retain their identity and stored timeframes. This selection does not activate rows.
Validate the source, closed-candle coverage and fixed settings before running. GPU 0 stays reserved.
| Arm | Net PnL ($) | Trades | Win rate (%) | Profit factor | Realized max DD ($) | Realized max DD (%) | Fees ($) | Long trades | Long PnL ($) | Short trades | Short PnL ($) |
|---|
Deltas are Supertrend minus Score6. Drawdown follows realized exit events, not intratrade adverse excursion. Undefined ratios appear as —. The report is a retrospective comparison, not an independent holdout or a recommendation to activate a strategy.
Green Supertrend: longs only · Red Supertrend: shorts only · EMA200: diagnostic · Shaded areas: unavailable. Hover for the latest closed candle and last color change. Only chart display points may be reduced.
Hover over the chart to inspect an hourly state.
| UTC exit month | Score6 PnL ($) | Supertrend PnL ($) | Delta ($) | Score6 trades | Supertrend trades |
|---|
| Source ref | Asset / pattern / direction | Type / interval | Score6 trades | Score6 PnL ($) | Supertrend trades | Supertrend PnL ($) | Delta ($) |
|---|
Compare current Hydra with calculation-only, filter-only, and combined RSI variants using the same configs, dates and portfolio settings. These are experimental results; they cannot be promoted or sent to the bot. Custom configs are excluded and counted; their detector settings remain fixed.
Saved results use their recorded dates and settings.
Uses the shared dates and portfolio controls in the Backtest tab. GPU 0 stays reserved.
| Variant | Status | Trades | PnL | Δ vs current | Win rate | Max drawdown | Details |
|---|---|---|---|---|---|---|---|
| No RSI comparison yet. | |||||||