Track record
Model track record
Ledger as of 2026-11-19 18:47 UTC. Selected books, live and paper: 11 books selected. Other books, live and paper, are not shown, so this is not the full record. Live books trade real orders (one of them on simulated capital, marked as such); paper books are simulations with no money. Not advice and not an offer.
Live
Live books
| Book | Window | Ledger days | Return | Max drawdown | Last 30 days | Sharpe | Sortino | Win rate | Profit factor | Best day | Worst day | Curve |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Cobalt (Crypto) live | 2026-11-10 to 2026-11-18 | 9 | −0.19% | −3.59% | – | – | – | – 9 trades | – | +2.92% | −3.59% | |
| Account equity, open positions included. | ||||||||||||
| Flint (US stocks) live | 2026-09-18 to 2026-11-18 | 44 | +7.06% | −0.40% | +0.00% | 3.11† | – | – 3 trades | – | +4.67% | −0.40% | |
| Per-trade return on the capital this book uses, compounded. Not a whole-account return. | ||||||||||||
| Onyx (Bitcoin) live | 2026-06-25 to 2026-11-19 | 148 | +141.94% | −38.16% | +14.04% | 2.86 | 6.01 | 55.4% 56 trades | 1.80 | +23.87% | −13.22% | |
| Whole-account equity, net of deposits and withdrawals, open positions included. | ||||||||||||
| Garnet (Bitcoin) live, simulated capital | 2026-10-16 to 2026-11-19 | 35 | +7.92% | −0.33% | +7.51% | 5.91† | – | – 6 trades | – | +3.58% | −0.33% | |
| Net result per closed trade as % of the account balance before entry, compounded. The account trades live orders on simulated capital: no own money at risk. | ||||||||||||
| Umber (Gold) live until 2026-11-02 | 2026-07-15 to 2026-11-02 | 79 | +11.93% | −13.18% | – | 1.00† | – | 57.6% 33 trades | 1.28 | +11.91% | −9.81% | |
| Per-trade return on the capital each trade uses, compounded. Not a whole-account return. Live until 2026-11-02; paper from the next day (listed with the paper books). Read this figure with care: it compounds each trade's return on the capital that trade used. In money, the account's net result over the same live period was about flat, slightly negative. | ||||||||||||
Paper
Paper books (simulated, no money)
Paper books place no orders: their results are simulated from the model's decisions and can differ from what live trading would get.
| Book | Window | Ledger days | Return | Max drawdown | Last 30 days | Sharpe | Sortino | Win rate | Profit factor | Best day | Worst day | Curve |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Quartz (Crypto) paper | 2026-10-20 to 2026-11-19 | 31 | +2.95% | −2.97% | +3.01% | 2.26† | – | 54.8% 31 trades | 1.35 | +1.66% | −1.68% | |
| Paper: the model's decisions in a simulated ledger, compounded. No orders, no money. | ||||||||||||
| Slate (US stocks) paper | 2026-10-26 to 2026-11-18 | 18 | +6.31% | −2.99% | – | – | – | 64.7% 17 trades | 1.80 | +2.04% | −2.50% | |
| Paper: the model's decisions in a simulated ledger, compounded. No orders, no money. | ||||||||||||
| Jade (US stocks) paper | 2026-10-26 to 2026-11-18 | 18 | +2.42% | −2.04% | – | – | – | 64.7% 17 trades | 1.44 | +1.57% | −2.04% | |
| Paper: the model's decisions in a simulated ledger, compounded. No orders, no money. | ||||||||||||
| Topaz (US stocks) paper | 2026-10-26 to 2026-11-18 | 18 | −0.25% | −8.39% | – | – | – | 58.8% 17 trades | 1.01 | +4.85% | −4.42% | |
| Paper: the model's decisions in a simulated ledger, compounded. No orders, no money. | ||||||||||||
| Basalt (US stocks) paper | 2026-10-26 to 2026-11-18 | 18 | +3.47% | −7.24% | – | – | – | 35.3% 17 trades | 1.28 | +5.95% | −3.19% | |
| Paper: the model's decisions in a simulated ledger, compounded. No orders, no money. | ||||||||||||
| Zircon (Crypto) paper | 2026-11-01 to 2026-11-18 | 18 | +1.91% | −1.00% | – | – | – | 55.6% 18 trades | 1.60 | +1.08% | −1.00% | |
| Paper: the model's decisions in a simulated ledger, compounded. No orders, no money. | ||||||||||||
| Umber (Gold) paper since 2026-11-03 | 2026-11-03 to 2026-11-18 | 12 | +11.26% | −3.28% | – | – | – | – 4 trades | – | +9.79% | −3.28% | |
| Paper: the model's decisions in a simulated ledger, compounded. No orders, no money. Paper since 2026-11-03, after its live period to 2026-11-02 (listed with the live books). | ||||||||||||
Returns are percentages on the basis named under each book; they are not comparable across bases and must not be added up; max drawdown is measured on the same basis as the return, so a per-trade or per-closed-trade drawdown is not comparable with a whole-account one. Max drawdown counts from the start of each row's window. Ledger days = days with a ledger entry (trading days for stock and gold books, every calendar day for crypto). "Last 30 days" is shown only for running books with at least 30 days of history. A book that moved from live to paper is shown as two rows, one per part, each measured from its own start; do not add them. Models are updated from time to time; each curve is the book as it ran. † = short sample (see below).
Method
How these are computed
- Return and max drawdown: on the basis named under each book, from the start of the row's window. There is no annualised return (CAGR) on purpose: stretched to a year, a few good weeks look like a huge yearly number.
- Sharpe: average daily return divided by the standard deviation of daily returns, times the square root of the number of ledger days per year (365 for markets that trade every day, 252 for weekday markets, read from the book's own calendar). No risk-free rate is subtracted. Sortino: the same, but dividing only by the downside (days below zero). Sharpe needs at least 30 daily returns; with 30 to 89 it is marked † (short sample: easily off by a lot either way). Sortino needs at least 90, because with few losing days it swings to very large numbers. A book that is often flat, or that only books results on the days trades close, has few moving days, which can make both ratios look better than they are.
- Calmar (annualised return divided by max drawdown) appears only for books with at least 180 days.
- Win rate: share of closed trades with a positive result. Profit factor: sum of winning trades divided by the sum of losing trades. Both need at least 10 closed trades; the count is shown under the win rate.
- Best day and worst day: the largest daily gain and loss on the row's curve.
- All figures come from the same ledger rows as the curves; record.json carries them (schema 2 since 2026-11-18; older snapshots in the chain are schema 1 without these fields and stay valid).