Dual CVD spot + futures, VPIN, Kalman adaptive channel. Under the hood — a built-in walk-forward validation engine: every idea is tested on out-of-sample data before you trust it.
We process raw exchange data the way professional quant systems do — separate taker flow, academic microstructure metrics, adaptive state-space filters.
Separate buying-pressure tracking on spot vs futures. Divergence between them is an early signal of manipulation — or of real demand.
Academic microstructure metrics from quantitative finance. VPIN measures the probability of informed trading — when market makers step back from the book.
State-space filter estimates slope and σ of the channel in real time. The diagnostic subpanel exposes distance-from-mid in ATR, t-stat trend significance and channel width breakouts.
Fair Value Gap with taker-flow analysis across the three-bar setup. Zones are colored by buyer share — confirmed demand vs suspicious imbalance.
An alternative to time-based charts. Bars print on ATR-based price moves — three sensitivity profiles (s/m/l). Filters out chop, exposes trend structure cleaner.
One glance at bullish/bearish bias across all 8 timeframes plus the range modes. Before entering, you immediately see whether higher TFs agree with your signal.
Unique indicators are half the work. The other half is checking whether they actually produce edge. The terminal is built on a research engine: walk-forward tests, a hypothesis runner, and a regimented derisking procedure for strategies.
Strategies are tested on a rolling window: train on one slice of history → verify on the next, never-seen slice. If a signal only works in-sample, it's discarded. No backtest illusions.
Describe a rule in YAML — the engine runs walk-forward across history and produces a verdict: ROBUST, PARTIAL or FRAGILE. A standardized pipeline instead of hand-rolled backtests.
A full 6-phase pipeline (feature engineering, Kalman, multivariate, game theory) delivered a stable out-of-sample advantage over baseline on 4h. The metric, the methodology, the code — all inside the terminal, not a marketing story.
305K simulated outcomes across 22M bars produced objective risk-management thresholds: where to move stop to break-even, where to take partial. Not rules-of-thumb — thresholds out of data.
From classic oscillators to rare microstructure metrics and adaptive state-space filters.
Direct connection to Binance Futures and Spot. Each candle keeps separate taker flow — who actually moved the price.
On top of candles, we compute metrics from academic research and adaptive state-space filters. The same concepts quant funds use.
An idea is encoded as a formal rule and run through the engine out-of-sample. The output is a verdict — visible before you trust the signal with money.