Quantum
DeCasteljau
Analyzer
Neural Trading Signals. Mathematically Precise.
The world's first trading signal extension built on De Casteljau recursive subdivision — the same geometric algorithm underpinning aerospace trajectory planning and CAD surface rendering — applied to live TradingView price action. A four-model ML ensemble, Kalman noise filtering, and fractional Brownian Motion path simulation converge into a single directional signal that locks on entry and holds until a definitive WIN or LOSS is recorded.
Inside the Extension
Every analytical layer surfaces in a single side-panel — live signal state, multi-timeframe confluence scores, historical performance, and active trade intelligence — without leaving your TradingView chart.
Three steps to your first signal
Install Extension
Add from Chrome Web Store in one click — no API keys, no broker connections, no configuration. The extension auto-detects active TradingView tabs and begins reading chart data immediately on activation.
Open Any Chart
Navigate to any supported instrument on TradingView — Forex majors, crypto pairs, equity indices, or commodities. The extension reads the live OHLCV feed directly from the page without requiring additional data subscriptions.
Trade the Signal
A directional signal fires with a precise Entry price, ATR-calibrated Stop Loss, and three scaled Take-Profit levels. The signal locks on that instrument and persists — unchanged — until a WIN or LOSS outcome is confirmed and logged.
From price to signal in 4 stages
Each signal traverses four mathematically independent processing layers — curve projection, noise filtering, ensemble voting, and signal arbitration — before it is permitted to fire. Any layer can veto the output.
De Casteljau Curve Projection
Raw price action is converted into Bézier curve projections via the De Casteljau recursive subdivision algorithm — the same geometric primitive used in aerospace trajectory planning and precision CAD/CAM toolpath generation. Applied to price history across multiple lookback windows, it resolves true trend curvature that lag-dependent moving averages systematically fail to capture.
Kalman Filter Noise Reduction
A recursive Bayesian state estimator — the Kalman Filter — strips market microstructure noise from the incoming price stream. The filter models price as a linear dynamic system, separating the underlying momentum signal from tick-level randomness generated by algorithmic market-making and order-book imbalances. The cleaned output is a phase-accurate velocity estimate used to confirm curve slope direction.
4-Model ML Ensemble Vote
Four independent ML models vote simultaneously: an XGBoost gradient-boosted classifier trained on 50+ engineered technical features, a Reinforcement Learning agent conditioned on historical win/loss sequences, a Logistic Regression model for binary direction probability, and a Linear Regression scorer for momentum magnitude. Final confidence is computed as a weighted average — only votes with strong directional agreement produce a signal.
Signal Lock System
Once a signal fires, it is hard-locked to that instrument until the trade reaches a confirmed WIN or LOSS. No mid-trade updates, no conflicting re-entries, no signal-chasing. The lock enforces the single most important rule of systematic trading: commit to the plan. Every locked signal is written to the history log with full market context for post-session review.
Everything built in
Start free. Scale when ready.
Free tier available. Pro from €19/mo. Elite from €49/mo with full signal locking and all 9 instruments.