Contents
Strategy Marketplace
Browse and filter listed strategies, and understand the detail page and ratings.
Browse and filter
The Strategy Marketplace shows strategies that have passed listing review and are running. You can browse it without signing in. Each strategy is shown under a codename, and all use the platform's standard performance and risk measures.
Filter and sort
Filter by type (Trend following, Mean reversion, Arbitrage, Market neutral, Market making, Multi-factor, Event-driven, Options / volatility, Multi-strategy, Other) and asset (BTC, Multi-asset, U.S. equities, Cross-market), and sort by listed date, annualized return, Sharpe, max drawdown and more. With no sort selected, strategies appear in the platform's recommended order.
Read a strategy card
A card shows the codename, U-Grade (once rated), annualized return, Sharpe and max drawdown. The most important thing to check is Hosting, which tells you where the performance comes from:
- Fully managed: a strategy built in the AI Strategy Workbench, with metrics from the platform's backtest engine. The small curve on the card is illustrative only; see the detail page for the real equity curve.
- Fully self-hosted: a self-hosted strategy, with metrics from real account data read through an exchange read-only API. The card curve also comes from the real record.
Read the strategy detail page
Click a card in the Strategy Marketplace to open the strategy detail page. The detail pages for backtested strategies (Fully managed) and self-hosted strategies (Fully self-hosted) are structured differently; the main difference is where the performance comes from.
Detail page for a backtested strategy
The page includes performance metrics, the equity curve, drawdown, a monthly return matrix, U-Grade rating details and team information. On the right (below on mobile) are summary statistics, the strategy fit profile and risk factors.
When you read a backtested strategy, the most important thing is to tell in-sample from out-of-sample. When out-of-sample data exists, the equity curve is split into two segments:
- In-sample: the strategy's parameters were chosen using this data, so it naturally looks favorable and cannot on its own show that the strategy works.
- Out-of-sample: this data was not used to choose parameters, and it is the main basis for judging whether the strategy works.
The Out-of-sample metric card shows the return over the out-of-sample period and the trade count recomputed by the platform.
Detail page for a self-hosted strategy
All of a self-hosted strategy's figures come from an exchange read-only API and are calculated independently by the platform. Compared with a backtested strategy:
- The Verification status panel shows Continuously Verified / Data disconnected / Stopped, plus the current continuous run, number of interruptions and last update time.
- Out-of-sample shows "All of it", because every day is real trading.
- Alpha and Beta show "—": live records are not regressed against a benchmark. When a metric cannot be calculated, the reason is shown.
- When some days were rebuilt by backfilling historical data, the chart legend distinguishes Reconstructed from Watched live.
- Open Methodology to see how the metrics are calculated.
U-Grade rating
U-Grade is a rating that operations assigns to listed strategies, in four grades: A, B, C and D. Rated strategies show their grade on the card and the detail page. Until a strategy is rated, the detail page shows "Unrated" and the card shows nothing.
Why there is a rating
Strategies differ widely in where their performance comes from, how long they have been running and their risk characteristics, so looking at returns alone is easily misleading. The rating provides a common reference point. It is a comparison tool, not a recommendation, and it does not mean the strategy has received funding.
Rating details
On the detail page, U-Grade Rating Details shows the score across eight dimensions, out of 100 points:
| Dimension | Max points | What it measures |
|---|---|---|
| Risk-adjusted return | 20 | Sharpe, Sortino and Calmar ratios |
| Return performance | 15 | Annualized and cumulative return |
| Drawdown and risk control | 15 | Max drawdown, VaR |
| Track record length | 20 | Length of the real running record |
| Stability and consistency | 10 | Alpha/Beta, monthly win rate |
| Capacity and executability | 8 | Liquidity, turnover, capacity |
| Extreme-scenario performance | 7 | Performance in extreme market conditions |
| Data reliability | 5 | Data collection, auditing, look-ahead bias prevention |