This post is a development preview of Sqetch '26 Finance, the fintech platform we are building on top of the open-sourced omega-functions runtime and the closed-source Quantum toolkit. It is not a release. A design-partner window is targeted for late 2026, with broader availability to follow. If the date matters to you, what you need from this post is what we can and cannot promise — and the short answer is that we will not commit to features or timelines that are not already buildable today.
The scope
Sqetch '26 Finance is aimed at high-frequency trading and quantitative-finance workloads. The feature set we are building toward, each grounded in code that already exists somewhere in the Anzaetek stack:
- Worst-case risk management on synthetic data. Conditional TimeGAN + Quantum Circuit Born Machine regime-aware synthetic generators; we wrote about the research prototype recently.
- Market regime detection with Temporal Similarity Search. HMM-labelled regimes as conditioning variables; TSS search over the time-series data lake for historical-analogue retrieval. TSS engine lives in
ts-tssin time_series_base. - Quantum optimal execution. QAOA and VQE loops from the open runtime, wrapped with execution-specific ansatz and constraint layers.
- Quantum portfolio optimization. QUBO / MILP problem formulations dispatched to the runtime's QAOA solver, with brute-force cross-check at small sizes.
- Hardware-accelerated backtests with script strategies. Strategies written in a DSL, backtested against tick and bar data with impact-aware fill simulation.
- Q/ML-augmented repricing and XVA. Amplitude-estimation-style speedups for Monte Carlo where the circuit size is tractable; classical-hybrid fallback when it is not.
- Fast meta-parameter search. Optuna-style hyperparameter search over backtest configurations, run as async batch jobs.
- API, CLI, agent, and GUI interfaces. Same runtime behind each; the agent surface is delivered via MCP, reusing the
ts mcpserver already in time_series_base so LLM agents can drive the platform without bespoke integration code.
The development-preview status
We are committing to "design-partner window, late 2026." We are not committing to:
- General availability by end of 2026 — this is a preview, not a release.
- Specific latency or throughput targets on specific hardware.
- Customer-specific integrations without scoping.
Each feature in the list above maps to code that already runs somewhere, but the integrated platform — one interface, one deployment, one auth model across all of the above — is the work in front of us.
How it sits on top of omega-functions
The runtime does the execution: circuit dispatch, gradient computation, PQC-authenticated serving, WASM-sandboxed optimizers. Sqetch does the domain work: market-data ingestion, regime labelling, backtest engine, portfolio-construction logic, reporting. The same lambda-function model that the runtime exposes for quantum circuits is what Sqetch uses internally for strategy scripts — one deployment model end-to-end.
What a design-partner engagement would look like
No marketing funnel to click through. If you are an HFT shop, a bank, or a fund running quant workloads and you want to give this feedback during the preview, the right path is to write to contact@anzaetek.com and describe the workload you would bring. We will be selective about scope early on — we would rather ship a platform that works for five workloads than one that kind-of-works for fifty.