// trading & markets
Infrastructure for people who trade.
// on this page · 4 sections
We trade our own capital on automated systems — live models on regulated prediction markets, multi-venue data pipelines, execution with tick-level exit logic — and we build the same class of infrastructure for trading firms, funds, and serious independents.
Every position reconciles to the exchange's own records — fills, settlements, actual outcomes — before any P&L number gets trusted. That discipline came from running our own capital, and it ships with every build.
What we build
multi-source ingestion, normalization, storage built for backtests: point-in-time correct, leakage-audited, and honest about survivorship bias, where only the things that survived get counted.
order management against exchange APIs, position tracking, alerting that wakes you for the right reasons.
the unglamorous layer that catches the sign-convention bug before it costs a quarter's edge. Fills-true P&L, settlement verification, drift detection between what your model believed and what the venue recorded.
backtest harnesses with pre-registered thresholds and calibration tracking under regime change, so a promoted strategy earned it.
Kalshi and Polymarket specifically: execution, order state, and fills-true reconciliation. Details on the prediction market integration page.
How we test whether an edge is real
Most strategies that look good in a backtest are measuring the backtest. We run our own research the way we would want yours run: each hypothesis goes on paper with its thresholds and its kill criteria before any data is pulled, so the result cannot be chosen after the fact.
The test covers what a filter rejected as well as what it passed. If the rejects would have done just as well, the filter is decoration. Costs go in before any conclusion does, including the fees and the slippage, the gap between quoted and filled price, that a real order pays at the size you trade. In our own work, that step alone has erased more than one apparent edge.
Every number is reconciled to the venue's own records before it is believed. When a measurement will not reconcile, we suspend it rather than explain it. The same practice is available as a standalone review of your research pipeline — see independent model audit.
What we are not
Not a fund, not a signal seller, not financial advisors — we don't touch your capital and we don't tell you what to trade. We build the plumbing that lets your edge survive contact with production. Engagements are scoped and confidential; the systems we run for ourselves are the reference architecture. studio@anchor163.com
Common questions
Write the thresholds and kill criteria down before looking at the data. Test what the strategy rejected alongside what it took, include realistic costs at your order size, and reconcile every result to the venue's own records. A backtest that survives that is worth paper trading; one whose thresholds moved after the numbers arrived was measuring itself.
No. We do not touch your capital and we do not tell you what to trade. We build the systems your research runs on, and we audit whether the numbers they produce are true.
A single scoped job starts at $5,000, and most builds land between $15,000 and $40,000. A review of an existing backtest or research pipeline is scoped to a defined claim rather than an open-ended audit.
studio@anchor163.com
// copied