EchoTrader by Aurion Labs

Aurion Labs

Real quant infrastructure for the firms building their edge.

Systematic trading and financial AI, scoped to your firm.

Request institutional access See what it does

EchoTrader Desktop

Discovery

Market Regime

Risk-On

VIX: 14.2 | Contango

GEX Signal

SPY +$2.1B

QQQ +$890M

Regime composite +1.42σ stable

Risk-on · cyclical leadership · volatility compressing

Transition risk low · Updated moments ago

Live mechanic pipeline

11 active

SPY 0DTE mean-reversion
Scaled

Information ratio

1.82 90d
NDX volatility carry
Medium

Information ratio

1.31 90d
IV dispersion v1
In validation

Gate progress

3 of 4

Factor performance · 21d

Live

Top factors · 21d t-stat

Flow imbalance

3.42

Vol surface

2.87

Cross-asset

2.54

Regime inst.

2.21

Microstructure

1.98

Vol Surface - SPY IV Rank: 24

OTM Put ATM OTM Call

7d → 90d DTE

calibration.drift · 24h within tol.

KL 0.04 p-val 0.34 last 14:32 ET

Aurion Labs

Production methodology.
Ready to engage.

A quant team in a box for funds, family offices, and emerging managers that need real rigor without spending two years building it.

Illustrative

dht_*

Dealer flow: forced-vs-voluntary attribution

rd_*

Regime: multi-factor stability metrics

vrs_*

Vol surface: dynamics, curvature, asymmetry

msxd_*

Microstructure: tick-precision flow primitives

Atlas Factor Engine Live

Cross-asset coverage with named methodology choices: dealer-hedging tape, physics-based regime transitions, per-venue order flow, charm-flow 0DTE.

Engineered factor columns: 1,083 across 41 categories

Regime: risk-on

Dealer imbalance: 2.3σ

Atlas Factor Library

Cross-asset, tick-cadence. Built around the questions most teams stop at: who is forced and who is voluntary, when a regime is becoming unstable, what an alpha looks like before it shows in price. Not a list of features. A body of research.

runtime_risk_gate

Live

KS-1 Macro event proximity

CLEAR

KS-2 Earnings window

CLEAR

KS-3 Exchange halt

MONITORING

KS-4 Witching morning

CLEAR

KS-5 Vol regime

CLEAR

Pre-trade risk, wired in.

A layered pre-trade risk stack sits between the model and the broker. Event-aware gates, volatility-regime conditioning, and append-only audit logging are part of the engine, not a separate compliance overlay added later.

Aurion X

In development

Private deployment

Native tool access to Atlas

Proprietary training data

Open-weight foundation model

Aurion X

A financial-domain LLM trained on the trade history Claude has never seen. Wired natively into Atlas. Deployed privately on your infrastructure so research data never leaves. In active development.

Mechanic discovery

CPCV validation

PBO + DSR gates

Atlas Research Workbench

Where hypotheses become mechanics and mechanics earn the right to trade. Every candidate runs the full validation gauntlet before reaching paper. Whatever survives gets staged capital. Whatever doesn't gets killed before it costs you money.

Stage 5/5

Promotion gate: PASS

mechanic_id: NVDA_gx_sqz_v2

CPCV

PASS

Combinatorial purged cross-validation

Folds with embargo 12 / 12

PBO

PASS

Probability of backtest overfitting

Threshold < 0.5 0.18

DSR

PASS

Deflated Sharpe ratio

Threshold > 0 0.42

Walk-forward

PASS

Regime-conditioned out-of-sample

Windows stable 24 / 24

Where strategies prove themselves.

Most quant teams ship overfit models because they skip the unglamorous part of validation. We don't. Every candidate passes a multi-stage statistical gauntlet plus daily live-calibration monitoring before it touches real capital. The mechanics that survive earn the right to trade.

lifecycle stage: SMALL_LIVE

Paper size: 0

Warm-live size: $100

Small-live size: 0.10×

Medium size: 0.50×

Scaled size: 1.00×

Capital staged, not granted.

No new strategy receives full capital on day one. Each promotion is gated by live performance, drawdown discipline, and operator review. Flawed mechanics are caught at the smallest possible size, before they can do real damage.

mechanic_spec_builder v3 / draft

Strategy thesis

Mean reversion on SPY 0DTE after a regime composite above 1.5σ with compressing vol skew. Tight time stop before charm acceleration.

Factor refs detected

rd_composite vrs_skew_velocity cf_charm_0dte gx_dealer_imbalance

translating to mechanic_spec

Atlas mechanic spec SPY_mr_0dte_v3

// entry

when rd_composite > 1.5 and vrs_skew_velocity < -0.5

// sizing

kelly_qtr × lifecycle_mult

// exit

tp +50% · stop -25% · time 3:45 ET

// gates

block_if(KS-1..5, blackout, decay)

CPCV PASS · 12/12 folds

PBO PASS · 0.18

DSR RUNNING…

From thesis to runnable mechanic.

Analysts describe a trade thesis in plain language. The console compiles it into a formal Atlas mechanic spec wired to real factors, risk gates, and exit logic, then routes it straight into the validation pipeline. Nothing reaches live capital until every gate clears.

GEX

RRG

Seasonality

Treemap

Screeners

40+ more

...

Atlas Quant Stack

Options flow, market structure, seasonality, sector rotation, breadth, dispersion. The analytics you'd otherwise license from four different vendors, in one stack, feeding one engine.

Strategy Validation

Risk-Adjusted

1.42

Overfit Score

0.12

Out-of-Sample

+23.4%

Confidence

95% CI

Strategy passed all validation checks

Walk-Forward Backtesting

Regime-conditioned walk-forward with embargo windows. Catches the strategies that look brilliant on contiguous data and break the first time the regime shifts. Required before paper, never optional.

atlas.discovery.scan

Live · 1,083 factors

Top Factors (t-stat) IC | Decay

Gamma Imbalance Z

3.42 0.08 5d

Vol Risk Premium

2.87 0.06 21d

Flow Acceleration

2.54 0.05 10d

Skew-Implied Sent.

2.21 0.04 1d

Factor Correlation

-1.0 +1.0

Confluence Matrix

21D

NVDA

●

MSFT

●

○

AMZN

●

○

●

Market State

Risk-On

Confidence: 87%

Composite Score

+2.41σ

IR: 1.84

Factors

512

Sharpe

2.14

Win%

68.3

Max DD

-12.4

A quant team that doesn't sleep.

PRODUCTION

Atlas continuously hunts for new mechanics across the factor library, scores them by information ratio and decay, validates them against statistical gates, and retires the ones whose edge fades. The discovery loop runs whether your team is at their desk or not. New candidates surface in your console, ready for human review.

Inside the engine

What a serious quant engine has to actually do.

Most quant pitches stop at "we have factors." Edge lives at the next four layers.

01 · Read

Read the tape the way market makers do.

Bar data hides who is forced and who is voluntary. Atlas works at tick cadence across the surfaces that actually move price intraday, attributing flow to its source rather than aggregating it into noise. The dealer's forced hedge stops looking like a discretionary buyer.

Methodology signal

Built across multiple domains of intraday microstructure research. Specific techniques, factor compositions, and venue-attribution logic are part of the firm IP and disclosed only under engagement.

02 · Anticipate

See regime shifts forming, not announcing.

Most regime models confirm the shift after the price has already moved. Atlas reads when a regime is becoming unstable, often well before the chart agrees. The portfolio de-risks before the drawdown, not in response to it.

Methodology signal

A multi-component regime model combined with stability statistics borrowed from applied physics. Specific decomposition, instability detectors, and transition-probability logic are part of the firm IP and disclosed only under engagement.

03 · Validate

Filter out the backtests that lie.

Most strategies look brilliant on contiguous data and break the first time the regime changes. Atlas runs every candidate through a multi-stage validation gauntlet built to expose that lie. If a mechanic clears the gauntlet, its live results will look like its backtest.

Methodology signal

A stack drawn from the published canon: combinatorial cross-validation, overfitting probability bounds, deflated risk metrics, false-discovery control under dependence, distribution-aware calibration, live drift monitoring. The published methodology is the easy part. The choreography, threshold settings, and stage ordering are the firm IP and disclosed only under engagement.

04 · Protect

The discipline that lets a backtest become a track record.

Every signal passes a layered pre-trade risk stack before any order reaches the broker. Drawdown discipline cuts size in real time. New strategies advance through staged capital only after their live results earn it. Every gate decision is audit-logged for compliance reconstruction.

Methodology signal

Layered event-aware gates, vol-regime conditioning, drawdown-aware sizing, position-size lifecycle, and capacity-aware capital allocation. The kill-switch architecture, lifecycle thresholds, and gate ordering are the firm IP and disclosed only under engagement.

The published methodology is the easy part. The choreography is the work.

What you're actually buying

Three ways to put Aurion Labs to work.

Pick the engagement that matches how your firm wants to operate. Move between them as you grow.

Engagement modes activate progressively. Research Access and Co-pilot Console pilot from late 2026 once Atlas has accumulated initial paper-trade track record. Strategy Licensing activates once Atlas has live track record.

For quant shops & prop desks

Research Access

A direct API into Atlas's signal output. You plug it into your own stack and route as you see fit.

→Real-time mechanic feed via REST and WebSocket
→Per-mechanic conviction, factor attribution, decay flags
→Validation reports for every new mechanic Atlas promotes
→Audit log access for compliance reconstruction

Co-pilot Console

A white-label operator console with Atlas research, Aurion X commentary, and risk dashboards. Built for your analysts and PMs.

→EchoTrader workspace branded for your firm
→Aurion X reasoning surface (when shipped) for trade rationale and research
→Live mechanic feed, regime overlays, validation gate status
→Dedicated SLA and named technical contact

Forthcoming

For allocators wanting turnkey

Strategy Licensing

License specific Atlas mechanics to run on your capital. We hold the IP. You hold the P&L.

→Selected mechanics validated through the full Atlas gauntlet
→Live signal delivery scoped to your strategy mandate
→Custom mechanic discovery on request
→Compensation aligned to AUM or P&L share

Pricing is scoped during the discover stage. There is no self-serve tier and no published rate card. Engagements begin with a confidential walkthrough.

HOW IT WORKS

A five-stage engagement path

Working with Aurion Labs follows a deliberate enterprise progression. Each stage is gated by mutual fit and clearly-defined deliverables.

Discover and Evaluate stages are available under engagement-only disclosure today. Pilot and Deploy stages activate progressively as Atlas advances through paper trading into live operation in late 2026. Scale activates post-track-record.

Discover

Confidential walkthrough of Atlas's factor library, validation methodology, and Aurion X's planned capabilities. Mutual fit established.

Evaluate

Read-only access to Atlas research on a defined set of tickers or strategies relevant to your portfolio. Two to four week scoped pilot.

Pilot

Live research feed and operator console access in a sandboxed environment. Your team validates output quality against your own conviction.

Deploy

Full production engagement. API access, operator console for your analysts, integration with your OMS or EMS. Defined SLA and dedicated technical contact.

Scale

Expanded engagement: additional asset classes, custom mechanic discovery, strategy licensing, white-label deployment of EchoTrader.

Cross-asset

Options, equities, futures, macro

Multi-stage

Validation gauntlet

Layered

Pre-trade risk stack

Audit log

Every gate decision recorded

Institutional access

Request a confidential conversation.

Aurion Labs operates by direct engagement with institutional partners. There is no self-serve tier and no public rate card. Engagements begin with a confidential walkthrough scoped to your firm.

Name

Firm

Role

Firm type

Use case

Frequently asked

Common questions from institutional prospects.

What is Aurion Labs and what does it build?

Aurion Labs is an AI-native quantitative trading firm. Three products in parallel: Aurion Atlas (the systematic trading engine), Aurion X (a financial-domain LLM in active development), and EchoTrader (the operator console and API delivering both to institutional clients).

How does Aurion Atlas differ from a generic quant backtester or signal vendor?

Three dimensions. First, the factor library is built around questions most teams stop at: who is forced and who is voluntary, when a regime is becoming unstable, what an alpha looks like before it shows in price. Second, the validation pipeline goes beyond textbook gates with multi-stage statistical filtering, live calibration tracking, and adversarial checks. Third, the risk infrastructure is production-wired with a layered pre-trade gate stack, capital-staged lifecycle, and append-only audit logging. The methodology choices behind each layer are part of the firm IP and disclosed only under engagement.

Why Aurion X, not Claude Code or another off-the-shelf LLM?

Off-the-shelf models can be wrapped in orchestration, but they cannot be trained on data they were never given access to and cannot be deployed inside your environment. Aurion X is the underlying model, fine-tuned on the trade-outcome record Atlas generates internally, integrated natively with Atlas's research infrastructure, and designed for on-premise deployment so your research data stays in-house. Three things a client gets that an off-the-shelf API cannot give them: a model exposed to data not in any public corpus, integration that does not depend on third-party tool calls, and deployment that meets compliance requirements around external data egress. Aurion X is in active development.

Can Aurion X be deployed privately so our research data stays in-house?

Yes. Aurion X is designed from the start for on-premise institutional deployment. Research data never leaves your environment and never enters third-party AI training corpora. Every reasoning trace is audit-logged. Versioned model snapshots support deterministic replay for compliance and historical analysis.

How does EchoTrader integrate with our existing trading infrastructure?

EchoTrader is both an operator console and a REST and WebSocket API. Funds can consume the API directly into their own stack, deploy the console as a research surface for analysts, or both. Interactive Brokers Gateway is the active integration today. Compatibility with major OMS and EMS platforms is built in, with prime brokerage execution paths on the roadmap. OMS and EMS integration is part of the deploy stage of our engagement model.

What is the track record and how can we evaluate without committing?

Aurion Labs is preparing for institutional engagement, with first pilots beginning late 2026. We do not publish performance numbers we cannot defend. The evaluate stage of our engagement model is a two-to-four week scoped pilot during which your team validates output quality on tickers or strategies relevant to your portfolio, against your own conviction. No commitment beyond the pilot scope.

What engagement models do you offer?

Three primary engagement modes, scoped during the discover stage. Research access via API for quantitative shops and prop desks that prefer to integrate research outputs into their own stack. Co-pilot console for small funds and family offices that want a white-label operator surface plus an AI analysis layer. Strategy licensing for funds wanting turnkey strategies to run on client capital. Specific terms are scoped to mutual fit.

What is the pricing structure?

Pricing is structured around the engagement model and scoped during the discover stage. Aurion Labs does not publish a standard rate card because institutional engagements vary significantly in scope, asset coverage, and deployment configuration. There is no self-serve tier; all engagements begin with a confidential walkthrough.