Fahali Research

Here is our research, not just our product.

Fahali is an autonomous risk-intelligence layer. This is the open record of how it reads markets — the engines, the method, and the daily live read — written so a human or an AI can understand it in one pass.

Read-only by design. Observation and situational awareness, not financial advice.

Definition

What is autonomous risk intelligence?

A category, not a feature.

Not a dashboard

Dashboards show data and wait for a human to interpret an alert. Fahali investigates the anomaly itself, then explains what it means in plain language.

Not a trading bot

There is no order routing and no path to capital. Fahali observes, infers, and notifies. Your desk decides and executes.

It grades itself

Every call is recorded and later resolved against what the market actually did — and the misses are kept, not deleted.

In one sentence: autonomous risk intelligence continuously detects, explains, and self-scores emerging market risk across asset classes — surfacing stress, capital migration, and contagion early enough to act, calmly enough to explain.

The Engine Library

Eighteen engines, watching at once

Each instrument is read every cycle by 18 independent detection engines. What sits behind the screen:

Flow & liquidity
dark_poolLarge, quiet absorption of supply that doesn't move price the way normal volume would.
order_flowInstitutional buying vs selling pressure, often hours ahead of price.
market_depthThe order book thinning out — a setup for sharp moves.
volume_anomalySudden volume spikes or drains that precede moves.
Risk & crash
risk_intelligenceFlash-crash precursors and outsized whale activity.
crash_predictorDedicated market-crash probability engine.
early_warningA 72-hour forward-looking crash / pump warning system.
leverageLiquidation cascades and margin-call waterfalls building up.
funding_stressStress in funding markets and crypto funding rates.
Structure & regime
market_regimeWhich regime the market is in — calm, trending, high-volatility — and when it flips.
pattern_recognitionAccumulation / distribution and institutional chart structures.
momentumMomentum shifts across multiple timeframes.
volatilityImplied-vs-realized divergence and term-structure stress.
Cross-asset & contagion
tail_dependenceWhen normally-unrelated assets begin to crash together.
correlationEstablished cross-asset correlations breaking.
stablecoinStablecoin de-pegs and unusual stablecoin flows.
Machine learning
ml_ensembleAn ensemble that combines the other engines' signals.
walk_forwardRolling-window out-of-sample validation.
Methodology

How a call becomes a record

The defining trait: Fahali grades its own homework. Here's the loop.

Emit.

A detection fires and is written down with a timestamp, the engines in agreement, and the market context.

Wait.

The system observes realized price over fixed horizons after the signal — 1h, 4h, 24h, up to 48h+.

Resolve.

The call is marked against what actually happened and stored in the signal-to-outcome ledger.

Keep the misses.

Results are retained — wrong calls included. The ledger is the asset; it compounds with every market session and can't be bought or back-filled.

On accuracy, we're deliberately quiet. The outcome ledger is young, and we will not publish a headline accuracy number until it has resolved across a mixed market regime. Until then we publish the method and the record itself — including the misses — not a manufactured percentage. Read the full methodology →

Field notes

The market, read out loud — every day

A plain-language read of what's actually moving across ~600 live markets, published multiple times a day. Free, no signup wall.

For machines, too

Built for humans · designed for agents

Fahali is natively callable. AI agents reach the same intelligence through the Model Context Protocol or a REST API.

MCP

Official registry: com.fahaliai/fahali. Remote SSE at mcp.fahaliai.com — four tools: market verdict, portfolio risk, contagion map, capital flow.

REST + OpenAPI

A documented API with a free developer key (verdict tool, 50/day). Full access on Professional and above.

Get a free key →

Machine-readable

llms.txt and a full reference corpus on both domains, so answer engines can read and cite Fahali accurately.

View llms.txt →