About RiskLoom

Institutional risk infrastructure, built deliberately.

In continuous production since January 2025. DMCC member. Dubai-based. Built by an operator for desks running leverage exposure.

RiskLoom is built by an operator with 12+ years across data engineering, platform architecture, and analytics, with an early career on the trading floor of a London proprietary firm. The problem it solves is one the founder lived: by the time a liquidation cascade is visible in price, it is already too late to act on it.

The premise is narrow and deliberate — instrument the structural pressure that precedes forced selling, with enough lead time for a desk to act. Built and run in continuous production since January 2025, the approach has been validated against 314+ documented cascades and counting, every claim reconcilable to a source-hashed, per-minute event in the archive.

RiskLoom is now built for any desk carrying leverage exposure — risk officers, prop traders, market makers, exchanges, funds, and family offices — who need the same warning.

Methodology

Engineering rigor, quantitatively validated.

The full methodology is disclosed at evaluation, under NDA. What follows is the high-level architecture — four instrumentations, each reproducible and reconcilable to source data.

01 · CRI

Per-symbol calibration

Every asset is calibrated against its own 90-day distribution — not one universal band. What's extreme for a large-cap is ordinary noise for a smaller name. This is most of the work, and the difference between a signal a desk trusts and one it mutes.

02 · Cross-Venue

CEX + DEX confirmation

Each cascade is matched against Hyperliquid within a tight window to confirm it is structural, not a single-venue artifact. Cross-venue confirmation is first-class, per symbol.

03 · Contagion

Permutation-tested clustering

Cross-asset clusters validated against a shuffled-onset null model over 10,000 permutations — 4 validated clusters, ~600× over chance, p<0.0001. Measured, not asserted.

04 · Reconciliation

Atomic, per-minute notional

Liquidation figures are real, atomic, per-minute forced-liquidation value — not rolling aggregates that flatter the headline. The $1.49B across the 2026 cohort is the smaller, defensible number, chosen on purpose.

Production Timeline

Milestones, in production.

Jan 2025

First cascade detection in production

The CRI engine goes live, detecting and documenting its first liquidation cascade under real market conditions.

Mar 2025

First 100 documented cascades

The verified archive crosses 100 reconcilable cascade events, each with source hash and per-minute inputs.

Late 2025

Cross-venue confirmation live

Hyperliquid integration adds CEX/DEX cross-venue validation as a first-class signal.

Early 2026

Cross-asset contagion validated

Permutation-tested cluster detection confirms correlated cascade onset across markets — 4 validated clusters at p<0.0001.

2026 cohort

154 events reconciled to per-minute data

The 2026 cohort is fully reconciled to per-minute liquidation data — $1.49B real notional, 134-minute median advance lead.

Q3 2026

Probability surface v2

Directional sizing layer planned, paired with regime and explicit confidence semantics.

Company

Structure.

Legal Entity
RiskLoom Technologies FZCO
DMCC member · Dubai, UAE
Disciplines
Engineering · Quantitative Research · Trading
Lean by design — operators close to the product, direct founder access during evaluation.
Where We Are

On the institutional circuit.

Dubai-based, engaging institutional counterparts across the global financial calendar.

Zug
Swiss digital-asset and crypto-finance hub — institutional counterparts and custody.
London
Founder's prop-trading network and the European institutional desk community.
Singapore
Asia-Pacific funds, exchanges, and the regional derivatives market.
Approach

Built by operators, not generalists.

RiskLoom is built by a team with hands-on backgrounds in data engineering, quantitative research, and institutional crypto trading. Based in Dubai, DMCC member, operating through 2026 and beyond. The product is held to one standard: every public claim reconciles to a documented, reproducible event.

The Institutional Standard

Every public claim reconciles to a documented event.

Institutional discipline

Every claim is reconcilable to SQL. Numbers shown anywhere can be traced to the data that produced them.

Audit-grade

Source hashes, manifest-backed outputs, reproducible during evaluation. Built for diligence.

No predictions, no signals

Detection and lead time only. RiskLoom does not forecast price or generate trades.

Deterministic outputs

Same inputs always produce the same outputs. No hidden state, no silent drift.

trials@riskloom.ai
terminal.riskloom.ai
DMCC member · Dubai, UAE
Request institutional briefing →