Trading engine

2026 · open source · Python

A paper-trading engine I've been building since July 2026, to find out whether a strategy, or a model reading the market, makes money after costs. Every strategy goes through the same backtest core and the same evidence checks before it gets a paper portfolio. Nothing has beaten its frozen control yet.

Figure 01 · Engine v2 · deploy checks live

Engine v2 and its deploy checks

Change
Checks
Live paper
Published

fig. 1 — Engine v2 with its deploy checks, live since 21 October: a fresh reviewer, replayed provider responses, a check of every scheduled job and a deploy gate. An invariant monitor, still watch-only, checks each published snapshot before reports read it.

Two engines

Engine v1 was a research simulator. Every strategy ran as a paper portfolio inside it, through eight updates from July to October. Engine v2 has been live on paper since 18 October and does a different job: it fills the orders, keeps the books and charges commissions on every fill, and each strategy gets its own paper account. Every change goes to a fresh reviewer that didn't write it. The checks in fig. 1 went live with the 21 October update. A new engine number means the engine's job changed; updates inside an engine move the number after the dot.

v2
Since 18 Oct 2026 · current
Executor and ledger
The engine fills the orders, keeps the books and charges commissions on every fill. Live on paper since 18 October, with deploy checks since 21 October.

Engine v1 updates

Each update has its own page with the diagram as the engine stood then. Update 1.8 added the shared backtest core, which Engine v2 still uses for research.

1.1
16–27 Jul 2026
Data and the first paper portfolios
Yahoo daily bars for about 12,200 listed names (Stooq was blocked on day one), a nightly trend screen, and ten paper portfolios filling at the next open.
1.2
28 Jul – 3 Aug
More portfolios and a backtest farm
Six research-based strategies took the count to 16 paper portfolios, a farm replayed every portfolio over past years, and the rules for splits and dividends were settled.
1.3
4–17 Aug
AI adjusting the portfolios
Five portfolios let a model veto entries or tune parameters inside fixed bounds, each paired with an untouched twin, and a news analyst wrote a morning brief. I retired all of it on 18 August.
1.4
18 Aug – 17 Sep
Evidence first
Ten-fold walk-forward tests, audits of the fill model and the data sources, and frozen forward monitors that can only continue or kill a strategy. An audit on 2 September found seven simulator bugs.
1.5
18–24 Sep
An agent in the loop
The repo went public. A nightly AI agent trades its own paper portfolio through a locked simulator tool, hourly agents watch without placing orders, a ledger scores every agent decision, and Alpaca is not connected yet while SEC EDGAR capture waits for access.
1.6
25 Sep
TradingView for research
TradingView quotes and bars reached the intraday agents as research input. They never touched prices, fills or orders.
1.7
26–29 Sep
Scoring every candidate
The model now scores every nightly candidate next to a fixed rule, three test portfolios trade those scores, and a paired test decides on fixed check dates. A challenger lab is built but switched off, and new strategy research moved to a private repo.
1.8
2 Oct 2026
One shared backtest core
Every strategy is now a short specification over one no-peeking evaluator, with native event and portfolio simulation, one evidence path and much faster full-market runs.