SCOUTING Watching M5 for impulse candles → M1 wick rejection.
Lot
SL
TP
TimeSideLot EntryCurrent SLTP % TP Float P&LStrategy
No open positions.
Total Trades
—
Win Rate
—
Net P&L
—
Profit Factor
—
TP Hits
—
SL Hits
—
Early Exits
—
Avg Win / Avg Loss
—
Best Trade
—
Worst Trade
—
Max Win Streak
—
Max Loss Streak
—
⚡ Early Exit Effectiveness
Losses prevented
—
early exit cut losing trades
Winners cut early
—
profit left on the table
—
📊 Best & Worst Hours
—
—
📅 Daily P&L
DateTradesW/LTPSLEEP&L
No trade data yet.
🕐 Hourly Performance (UTC)
Profitable hour Losing hour No trades
🧠 Strategy Advisor
Click "Analyse & Advise" to get expert recommendations based on your trade history.
IC Markets cTrader 📋 PAPER
Not connected · — · —
📋 Paper mode — orders are simulated locally. No real money touched.
⚡ Live mode — real orders sent to IC Markets. Enable only when intentional.
Chart Mode
Select the charting engine for the live chart
Change takes effect after clicking Apply (page reloads).
Session
Signed in as account —
Logging out clears your session and returns you to the sign-in screen.
Risk Kernel
—
Active Strategies
—
Open Positions
—
Daily P&L
—
Emergency Halt
Strategy Lifecycle Registry V2 — genome state machine
Loading strategy genomes…
Meta Allocator Real-time strategy gate decisions
No allocation decisions yet — click Refresh.
Drift Events Feature / prediction / performance drift
No drift events recorded.
Autonomous Mode System trades, learns, and researches on its own
Stopped
TRADE ON:
Interval:
Auto Trades
0
Research Cycles
0
RL Updates
0
Next Research
—
Next Retrain
—
Start Auto Mode to begin autonomous trading...
System Monitor Real-time health · metrics · alerts · ready-to-trade status
Checking system…
—
—
COMPONENT HEALTH
—
System Maturity When these turn green — system is ready for live trading
MATURITY SCORE
—
Loading...
Trading Controls Lot size · SL · TP · Max positions · Trailing stop
AUTO
AUTO MODE — ATR MULTIPLIERS
TRAILING STOP
OFF
📈 Learning Analytics RL reward curve · Foundation training · Knowledge base growth
🕐 TRAINING TIME
—
📅 LEARNING SINCE
—
🧠 RL REWARD
—
📚 KNOWLEDGE
—
LAST TRAINED
—
RL EPOCHS
—
FOUNDATION EPOCHS
—
💾 DATA SIZE
—
RL REWARD CURVE (best per epoch)
— Training epochs → —
WHAT THE SYSTEM LEARNED
  • Loading hypotheses…
RECENT LEARNING EVENTS
Loading…
Deploy karte time Export Learning Data download karke naye server pe daalo — koi bhi learning lost nahi hogi.
Foundation Model Self-supervised market representation · Masked Autoencoder 4L×4H×128D
Status
—
Architecture
MAE 4L×4H×128D
Best Loss
—
Epoch
—
Time
—
Training Progress 0%
RECONSTRUCTION LOSS CURVE
No training history yet
SAVED CHECKPOINTS
No checkpoints yet.
Alpha Intelligence — BTC Signals Real-time funding rate · liquidations · on-chain netflow · sentiment NLP
—
Funding Rate
—
loading…
OI: —
Liquidations — 1H
Long Liqd —
Short Liqd —
Net Pressure
→ —
Sentiment NLP
—
—
confidence: —
source: —
On-Chain Flow
Exchange Netflow
→ — BTC
—
—
—
Funding Rate History (last 500 periods)
—
No history loaded — click Refresh to fetch.
Auto-refresh: 60s
—
Research Intelligence — Autonomous Discovery AI generates hypotheses · statistical testing · causal signal discovery
Validated Facts
—
Hyp Tested
—
Best Win Rate
—
Causal Edges
—
SYMBOL
HYPOTHESES
Idle
VALIDATED KNOWLEDGE
No validated facts yet — run a research cycle.
CODE SANDBOX — TEST YOUR OWN HYPOTHESIS
Available: numpy (np), math, json, statistics · Context: candles[] loaded
RL Decision Engine — PPO Agent Reinforcement learning · learns from outcomes · never bypasses RiskGate
⚠ Paper trading only — RiskGate required before any live execution. This agent is a CANDIDATE signal.
CURRENT DECISION
FLAT
CONFIDENCE
0.00
Episodes
—
Mean Reward
—
Best Eval
—
Trades/Ep
—
SYMBOL
EPISODES
Idle
Ensemble Signal Fusion Foundation · RL · Validated Hypotheses · Alpha — fused into one decision
SIGNAL SOURCES
No data yet — click Get Signal
DIRECTION
—
AGREEMENT
—
POSITION SIZE
—
RISKGATE VERDICT
Not checked
SYMBOL
Execution Bridge — Order Management Paper mode active · RiskGate enforced on every order · Live requires explicit confirmation
🔒 PAPER — Safe
Orders simulate against PaperTrader. No real money at risk.
BALANCE
—
DRAWDOWN
—
OPEN ORDERS
—
FILL RATE
—
MANUAL ORDER (TEST)
SYMBOL
DIRECTION
SIZE (lots)
ORDER HISTORY
No orders yet
Continual Learning Loop System improves after every trade · EWC prevents forgetting · Evidence-triggered
ONLINE RL UPDATES
—
FINETUNINGS
—
WIN RATE (30T)
—
AVG REWARD
—
TRIGGER STATE
TRADES → NEXT FINETUNE—/50
Total trades: — EWC: — λ_EWC: 400
REWARD TREND (RECENT UPDATES)
Autonomous Research Organization 6 specialized agents · hypothesis tournament · genome evolution · self-improving
SPECIALIZED AGENT POOL
Loading agents…
GENOME POPULATION (sorted by fitness)
No genomes yet — run a cycle
LATEST RESEARCH DIGEST
No digest generated yet. Run a research cycle to generate one.
SYMBOL
Idle
Signal Model Training Walk-forward ML training with DSR + Monte Carlo validation
STEP 1 — FETCH LIVE MARKET DATA (XAUUSD + BTCUSD)
Loading data stats…
STEP 2 — TRAIN SIGNAL MODEL
SYMBOL
MODEL
TRAINED MODELS ON DISK
No trained models yet. Fetch data → Train.
Research Graph Lineage: Observation → Hypothesis → Experiment → Strategy
No research graph nodes yet. Start a research cycle.
Autonomous Research Scientist
Recent Hypotheses
No hypotheses yet — run a research cycle.
Recent Experiments
No experiments yet.
Model Evaluation Dashboard
Live metrics for all ML/DL models — updated on demand
—
Monitor not yet run
Click Run Monitor to evaluate all models
—
System Scorecard
Signal Accuracy
—
—
Sharpe Ratio
—
Annualised from trade P&L
—
Max Drawdown
—
Peak to trough cumulative P&L
—
RL Best Reward
—
—
—
Avg R-Multiple
—
Expected return per unit risk
—
Brier Score
—
Confidence calibration (lower = better)
—
Signal Accuracy metrics are computed from live trade outcomes. RL Classification metrics (Precision, Recall, F1) are computed from directional accuracy on held-out market data (20% test split). Foundation metrics reflect reconstruction loss on held-out bars. All metrics update automatically after each training run.
Signal Classification MetricsPer-class precision, win count, and confidence from trade outcomes
Signal ClassTradesWinsPrecisionAvg P&LAvg Confidence
Refresh to load classification metrics
Signal Sources
—
RL Agent (PPO)Reinforcement learning reward curve across training epochs
Unique Epochs
—
Latest Reward
—
Reward Trend
—
Checkpoints
—
Classification Metrics — Directional Accuracy on 20% Held-out Data
Signal ClassPrecisionRecallF1 ScoreTP / Predicted
Train RL model to populate classification metrics
PPO Loss Curves — Actor and Critic per Update Step
Train Reward (mean)
—
Eval Reward (val)
—
Actor Loss
—
Critic Loss
—
Entropy Loss
—
Eval Steps
—
Foundation Model (Market MAE)Self-supervised market encoder — estimated parameter count and training status
Epochs Trained
—
Est. Parameters
—
Checkpoints
—
Days Since Trained
—
Reconstruction Loss per Epoch — Train
Final Train Loss
—
Best Train Loss
—
Test Loss (held-out)
—
Research Knowledge BaseHypothesis engine quality — convergence signal for autonomous discovery
Total Hypotheses
—
Validated
—
Rejected
—
Acceptance Rate
—
Monitor Alert LogAutomated diagnostics — agent flags threshold breaches and triggers corrections
No alerts yet — click Run Monitor
Tenant Accounts
AccountStatusActions
Set Password
cTrader App Credentials (global)

Shared Client ID/Secret every tenant's connection falls back to. Changing this updates .env and applies immediately to all active accounts — no restart needed.