Market intelligence: the meaning layer over price and volume.
Data feeds give you numbers. LyraMind's market engine reads candles and returns what they mean: what regime a name is in, whether the move is supported, how stretched it is, and what changed since yesterday. Runs on Polygon when keyed, labeled sample data otherwise.
One engine, several lenses
The market service computes a stack of deterministic reads over daily and intraday candles: a Wyckoff-style regime classifier, a trend-participation score, a mean-reversion stretch estimate, a multi-timeframe combiner, and a multi-horizon Time Machine. The Market Brain synthesizes them into a single condition, confidence, risks, and a what-changed diff. Each read is a pure function of the inputs, so the same candles always produce the same answer.
Honest about its data
Every endpoint tries the live provider (Polygon for OHLC, Tradier for chains) and falls back to bundled mock data, always reporting which provider actually answered. Probabilities are historical-tendency estimates, never price predictions. The output is educational framing of public market structure: it describes conditions and risks, it never tells you to buy or sell.
- Regime classification: accumulation, markup, distribution, markdown, squeeze, and volatility regimes
- Trend participation score (0-100) from volume confirmation, relative volume, EMA alignment, momentum, VWAP
- Mean-reversion stretch: distance from mean in ATR units, RSI and Bollinger extremes, move rarity
- Multi-horizon Time Machine: volume-validated read per window (1w to 1y) plus one consolidated opinion
- Multi-timeframe combiner across 5m, 1h, 1d, 1w
- Every response reports its data provider and degrades to labeled sample data offline
Questions
Does this predict where the stock is going?
No. It classifies the current condition and estimates historical tendencies (like mean-reversion probability). Those are descriptions of structure, not forecasts, and never advice.
Is the market data live?
It uses Polygon and Tradier when API keys are configured; otherwise it returns clearly labeled sample data and reports the provider on every response.
How is this different from a charting indicator?
Indicators output numbers. This layer combines several of them into a single plain-English read of regime, participation, and risk, with the reasoning attached.