Documentation

Two products and the evidence behind them. Start with the strategy reference if you are configuring a strategy, the audited builds if you want to know how one actually behaved on real data, and the changelog if you want to know what changed and when.

Strategy reference
All 34 documented strategies, each with every configurable setting, its full tooltip, an academic primer and references for further study.
34 strategies
Audited builds
Three years of real hourly data across seven markets, after fees and slippage. Median return, edge against buy-and-hold, drawdown, and every defect found and fixed.
6 audited of 419
Audit methodology
The cost model, the fill model, the five correctness invariants, the shared colour contract, and the engine self-test the whole pipeline aborts on.
53 known-answer tests
Changelog
Every change to the pack, dated and in plain language — the patch notes written when each fix landed, the pack-wide milestones, and one entry per completed audit.
58 entries
AI Desk
How the desk screens candidates, what the fail-closed risk gate checks before an order survives it, which providers it speaks to, and why your exchange keys never enter it.
Port 8030

Where to start

If you are choosing a strategy

Read the audited builds before the strategy reference. The reference tells you what a strategy does and how to configure it; the audit tells you what it actually did on three years of real data, against the far simpler benchmark of just holding the asset. Most of the pack loses to that benchmark over this window, and the reports say so on the first screen.

If you are configuring one

The strategy reference lists every setting the GUI exposes with the tooltip that explains it. Each audited strategy also publishes its measured settings — which knobs the code actually reads at runtime, established by running the strategy with a recording proxy rather than by reading the source. A setting that is exposed but never read does nothing, and the audit names it.

If you want to know what changed

The changelog runs back to the first library-wide sweep and forward to the audit that completed most recently. It includes the entries that found nothing and the ones that report a strategy performing badly, because a changelog that only carries good news is not evidence of anything.

What the documentation does not cover

Nothing here was measured on a live exchange. The measurement engine models spot and long-only with a single price per bar — it carries no order book, so it cannot see spread, queue position or depth, and a strategy whose edge is the spread cannot be measured by it at all. Every audit page states this alongside its numbers.