Trading signals · FX · Indices · Commodities

Conditional probability, not predictions.

Our signal pipeline supports our own self-trading desk. It blends classical quantitative features with AI-trained overlays, calibrated so we can size by edge instead of conviction. Every signal is traded by us before it touches live capital.

Defam does not sell signals or operate a subscription service. The signals described here power our own positions.

Trading signals visualisation
Calibrated · v8
What a signal is

And what it is not.

A trading signal, properly defined, is a conditional probability statement: given a specific market state, the expected return over a defined horizon is positive (or negative) by a measurable margin. Everything else — the chart, the colour-coded arrow, the push notification — is presentation.

We treat signals as inputs to a sizing decision, not as instructions. A signal that is right 55% of the time on a defined horizon is genuinely valuable, but only if it is sized in proportion to its edge and its variance. Most retail signal services collapse precisely because they ignore this step.

Coverage

Where our signals operate.

FX

G7 majors and a curated set of EM crosses. Features built on order-flow proxies, term structure, and realised-vs-implied vol.

EUR/USD · GBP/USD · USD/JPY · USD/CHF · AUD/USD · USD/MXN · USD/ZAR
Indices

US, European and Asian benchmark CFDs. Cross-asset momentum and volatility-regime overlays.

S&P 500 · NASDAQ 100 · DAX 40 · FTSE 100 · Nikkei 225
Commodities

Energy and metals, with explicit handling of inventory and seasonality features in research.

WTI · Brent · Natural Gas · Gold · Silver · Copper
The pipeline

From hypothesis to position.

  1. 01
    Data

    Tick and minute-bar history per instrument, normalised across venues, with explicit handling of holidays and corporate events.

  2. 02
    Features

    Classical and learned features: term structure, realised vol, cross-asset correlation, order-flow imbalance proxies.

  3. 03
    Model

    Calibrated probability output — gradient-boosted trees, regularised linear models, and selective neural overlays per instrument.

  4. 04
    Sizing

    Position size as a function of calibrated edge and live variance estimate, capped by the global risk envelope.

  5. 05
    Live

    Deployed on our platform or our internal execution layer, with a kill-switch tied to live performance drift.

Sizing and the live kill-switch are governed by our risk management practice — signals never override the perimeter.

See risk management →

Long-form on signals.

Read our practitioner notes on calibrated signals across FX, indices and commodities.

Read the article →