For institutional investors
Modern risk management,
run on a digital twin of your fund.
We build digital twins of your portfolio managers and funds, then simulate every trade and outcome. See how the TWINagentic compares with the PMhuman before capital moves.
Our team, partners and advisers come from
A digital twin of every manager and every fund
The twin is an agent that trades the way its human does. That gives risk a second copy of the book to test, question and run ahead of the real one.
Digital twins
A twin learns how a portfolio manager or a whole fund actually trades, from positions, trade history and mandate.
Trade simulation
Every proposed trade is played forward through many market paths, so the range of outcomes is known before the order goes in.
Human against agent
The agentic twin runs the same book under the same limits. Where it and the human part ways, risk sees it first.
Every trade, simulated before it is placed
Watch it work. A manager proposes a trade, we test it against thousands of possible futures, and the twin takes it, takes less of it, or skips it. Then we compare how the human and the twin each did.
1 A manager wants to make a trade
2 We play it forward 2,000 times
3 The twin decides
Human and twin, side by side
Total return after each of the last 0 trades. Both are offered the same trades. The human takes every one at full size. The twin follows the risk call.
| Human | Twin | |
|---|---|---|
| Total return | 0.0% | 0.0% |
| Biggest loss on one trade | 0.0% | 0.0% |
| Trades taken at full size | 0 | 0 |
| Trades made smaller | 0 | 0 |
| Trades skipped | 0 | 0 |
| By holding back, the twin avoided 0.0% of losses and gave up 0.0% of gains. | ||
A demo running in your browser with made-up trades. Not real positions or results.
A custom engine, built for finance
The twins do not run on a general-purpose AI tool. Underneath is a simulation engine we built from the ground up for institutional portfolios, so it thinks in positions, limits and market moves.
Across five simulated books, the agentic twin finishes 5.3 points ahead of the human
Five example books over 24 months. Each twin takes the same trades as its human manager but sizes down the losing ones, the way a risk overlay would. These figures are generated to show how the comparison reads. They are not real results.
Long/short equity: cumulative return
Simulated book, 24 months. Agentic twin +35.3%, human manager +30.7%.
| Return | Worst drawdown | Sharpe ratio | ||||
|---|---|---|---|---|---|---|
| Book | Human | Twin | Human | Twin | Human | Twin |
| Long/short equity | +30.7% | +35.3% | -3.8% | -2.4% | 1.64 | 2.12 |
| Global macro | +29.6% | +37.6% | -5.4% | -3.3% | 1.37 | 2.04 |
| Credit | +4.8% | +9.4% | -5.8% | -3.6% | 0.48 | 1.10 |
| Event driven | +13.0% | +21.3% | -6.2% | -3.8% | 0.71 | 1.36 |
| Multi-strategy fund | +25.1% | +26.2% | -3.2% | -2.0% | 2.55 | 3.09 |
Generated for this page. Simulated and hypothetical results have limits and do not represent actual trading. Past or simulated performance does not predict future results.
How it works
- Build the twin. We model each portfolio manager and fund from positions, trade history, mandate and limits, so the twin trades the way they do.
- Simulate. Proposed trades and whole books are played forward through many market paths to show the range of outcomes and where limits break.
- Compare. The agentic twin runs alongside the human. Risk sees where they differ, what it cost or saved, and why.
This page describes a risk tool. It is not investment advice.