The Lab

The systems we run with our own money.

Consultancies show you slide decks. We show you production. Every system below is deployed on Google Cloud, trades against live market data through the Alpaca brokerage API, and is operated — monitored, debugged, improved — by the same engineer you'd be working with. This is our research platform and our proof.

Balanced Portfolio Engine

Live · GCP · Alpaca

Our most conservative book: a multi-asset portfolio system designed around one question — how do you keep compounding when the market changes character? The engine classifies the prevailing market regime and adapts position sizing and exposure accordingly, with hard drawdown circuit breakers that de-risk the book automatically rather than relying on a human watching a screen.

Regime-awareSizing adapts to detected market state

Circuit breakersAutomatic de-risking on drawdown thresholds

Cloud-nativeContainerised on GCP with health monitoring

python regime detection portfolio construction cloud run alpaca api

Frontier Optimizer

Live · GCP · Alpaca

Classical portfolio theory meets the modern information environment. The system constructs efficient-frontier portfolios, then tilts allocations using an NLP sentiment pipeline that scores market news as it arrives — so the optimizer isn't blind to the narrative driving prices. Mathematics does the allocation; language models read the room.

Mean–variance coreEfficient-frontier optimization with practical constraints

Sentiment overlayNews-NLP scoring feeds allocation tilts

Continuous deliveryCloud Build pipeline, versioned deployments

optimization nlp / sentiment python cloud build alpaca api

Volume Predictor

Live · GCP · API-ready

Execution quality is decided by when you trade, and timing depends on liquidity. This system forecasts intraday volume with machine learning trained on market microstructure features — the unglamorous signal that makes every other system's execution smarter. It's also the most requested capability we build, which is why it's the first candidate for the FluxMetrics public API.

ML forecastingIntraday volume profiles from microstructure features

Execution signalFeeds order-timing decisions in our other systems

API-first designBuilt to be consumed as a service

machine learning microstructure forecasting rest api

Interactive: Bayesian regime detection, live in your browser

A year of prices is simulated from a hidden 3-state Markov model (bull / chop / crisis). Press run and a Gaussian-emission HMM forward filter — the same recursion inside our Balanced Portfolio Engine — recovers the hidden regime from daily returns alone, one day at a time, no look-ahead.

bull (µ>0, low σ) chop (µ≈0, mid σ) crisis (µ<0, high σ) de-risk zone: P(crisis)>50%

α_t(j) ∝ 𝒩(r_t | µ_j, σ_j) · Σ_i α_t−1(i) P_ij — hover the chart after the sweep to read the posterior on any day.

A note on honesty

These systems currently trade in Alpaca's paper environment while we harden them — that's the same discipline we'd insist on for any client: no method touches real capital until it has survived production conditions end-to-end. We publish what each system does and how it's engineered, not cherry-picked return figures. If you want to see how they work under the hood, book a session and we'll walk you through the live dashboards.

Want this discipline applied to your stack?

We'll build yours the way we build ours.

Strategy validation, data pipelines, execution infrastructure, risk systems — engineered, documented, and handed over. Start with a free 30-minute working session.

Schedule a session See the services