Everything in the LORE library.
Every category, every product.

The raw, sequenced record of every trade, level, and book event across the nine major crypto venues. Normalized to a common schema, replayable by timestamp, and aligned across exchanges so your AI sees the market the way the matching engine did.

This is the foundation everything else in the library is built on. If a regime tag, a setup label, or a research dossier exists upstream, it traces back to a specific row in here.

Funding rates, open interest, basis spreads, and forced-close events for the perpetual futures universe. The forces moving the market that don't show up in the candle alone.

An AI agent looking only at price misses why the move happened. This category gives it the funding pressure, leverage state, and liquidation context that explain whether a move was real or mechanically forced.

Where traders are crowded, who's trapped, and how aggressive the flow is. The data layer that separates a real conviction move from a forced one.

An AI agent that only sees candles can't tell when shorts are about to be squeezed or when longs have been silently leveraging up for hours. This category surfaces those positioning dynamics directly.

Multi-venue confirmation, lead/lag dynamics, and dispersion signals — so an AI can tell whether a price move is real coordination across exchanges or one venue acting alone.

Single-venue data is the most common AI failure mode in crypto: a move on Binance gets treated as truth without checking whether spot and other perps confirmed it. This category eliminates that blind spot.

What a trained human trader sees on a chart — encoded so an AI agent can read it directly. Support, resistance, trend channels, volume profile, named setups, and natural-language captions for every window.

This is the largest single category in the library by row count. Without it, an AI working on raw OHLC has to invent its own technical analysis. With it, the chart is already pre-read.

The market state your AI is reasoning in. A pattern that works in a quiet compression regime can fail in a liquidation cascade. This category gives every market row a causal regime context: trend state, volatility state, liquidity state, crowding state, cluster age, recent regime shifts, and where each asset stands relative to its peers.

This is also where the Relative Strength Atlas lives — peer rankings across the universe by momentum, funding, volume, volatility, liquidity, OI, crowding, and regime behavior. Most crypto edges are not "BTC went up 2%" — they're "SOL is leading the universe right now while ETH is lagging." This category makes that explicit.

The strategy-discovery memory of the library. Named market episodes, outcome tables, replays of similar historical setups, base-rate comparisons, and ranked research lead cards your AI can drill into.

This is also where the Outcome Atlas lives — forward returns, MFE/MAE, barrier-touch order. These are labels, kept clearly separated from feature-safe inputs so an agent can study what happened next without accidentally training on the answer.

The reliability scoring, validation gates, and leakage-safe pipeline that turn the rest of the library into something an AI agent can train and test against without accidentally cheating.

If the rest of the catalog is the data, this category is the rules of the game. It's what makes the difference between an AI that learns a real pattern and one that quietly trains on the future.