Update 01 — Launch
Ostinato Systematic is live. Our first strategy began trading on 21 September 2026 as a DARWIN on the Darwinex Zero platform.
It took 108 days, from our first research signal on 5 June 2026 to launch. The numbers behind it:
- 1,452 commits. About 14 a day, every day, for three and a half months.
- ~355,000 lines of code across 1,250 program files: research, execution, risk controls and monitoring.
- 85 pre-registered studies. Every test was written down, with its pass mark, before a single result existed.
- 130+ research reports and 250+ research workspaces.
- 58 automated checkers guarding every change before it reaches live trading.
- 856 instruments in our own price database, built and verified symbol by symbol.
- Hundreds of ideas tested and rejected. Most of the work was killing ideas, not keeping them.
How every idea was tested
Each candidate faced the same gauntlet. Nothing reached live trading without passing all of it.
- Out-of-sample. Judged on years it was not designed on.
- Beat random. Compared against scrambled versions of itself: same trades, same holding times, random timing. Luck does not pass.
- Robustness. Stable across six separate market phases, not one lucky stretch. Neighbouring settings had to work too: no isolated peaks.
- Real costs. Spreads, commissions and overnight financing at the venue's actual rates.
- Reproduced. Every study replicated byte-identical at every review stage.
- Adversarial review. Every study and assumption critiqued and attacked by an independent review panel before deployment.
What we built
- A multi-strategy portfolio. Independent strategies across equities, currencies and commodities, with near-zero average correlation between them (backtest, 2018–2026).
- Rules-based sizing. Every position size comes from fixed rules: market volatility, the strength of a trend, or a ranking. Never from conviction. Hard caps limit concentration.
- Circuit breakers on every component. Each one stops itself if it falls beyond its own historical limits.
- Event risk rules. Known high-risk events are handled by fixed rules.
- Fully systematic execution. Rules-based actions on confirmed market data, executed deterministically and automatically.
- Monitoring and recovery. Daily health checks, fault alerts, independent watchdogs, a backup server cascade and tested recovery pathways.
- Infrastructure. Execution runs on a dedicated server, separate from research. Live prices come from an isolated read-only connection that cannot place a trade. Market data has three independent sources with automatic switching. Every scheduled job reports its own health and raises an alert on any fault, and external monitors raise an alarm if a trading window goes silent. The full system is backed up regularly, and every backup is proven by a test restore before it counts. Credentials never sit in code or in cloud storage. Every change is version-controlled, reviewed and reversible.
Since launch
The first two weeks ran as designed. Every order the system placed filled as intended, and live results tracked the model within expected tolerance.
What comes next
We will post here monthly: design updates, results, market commentary, and reflections on strategy building and risk management.