If you take one habit from this site, take this one: before you read a trading figure, read its label. There are only three, and they are not interchangeable. This page shows the desk's own record with the labels on, so you know what to look for elsewhere.
The three labels
Backtest. How a set of rules would have behaved over past prices. Hypothetical by construction: the rules were known in advance, every fill was clean, and the past does not repeat. Useful for the shape of a strategy, never for what anyone earned.
Paper. Rules executed in real time on a simulated account. No hindsight, real timing, no real capital. The first evidence that a strategy can be run as written.
Live. Real capital, real fills, real behaviour. The only record that is a result rather than a hypothesis or a rehearsal. Always the shortest of the three at any given moment, and always judged by its length before its slope.
A record is labeled when every figure on it carries one of these words, and when the three are never added together, averaged, or drawn on one curve.
The desk's record, as published
The six-sleeve backtest, over its full window including the down year, reads: 9.5% compound annual growth, a maximum drawdown of −6.3%, and −2.2% in 2022, the worst calendar year in the window. All three are labeled backtest, hypothetical, and they sit on the record page next to each other, because any one of them without the other two is a half-truth.
The paper record is published by period, labeled paper: how many signals, the outcome count, the result in R, the worst week. R is the result in multiples of the risk taken on the trade, so a +2.8R trade made 2.8 times what it risked, and the record can be read without knowing anyone's account size.
The live record is published the same way, labeled live, with the day count of the live record stated on the page. When it is seven days old, it says so, and asks to be judged by that.
What blending hides
The most common blended record draws a single equity curve where the early years are backtest and the last stretch is live. The curve is long and smooth, which is what the backtest contributes, and real, which is what the live stretch contributes, and the reader is invited to believe both at once. Pulled apart, the same data says something humbler: a long hypothesis, and a short result that may or may not match it yet.
The second most common blend is a return without a drawdown, or a drawdown over a shorter window than the return. Keeping labels also means keeping windows: the return, the drawdown and the worst year are quoted over the same full window, every time.
How to use a labeled record
- Read the backtest for shape: drawdowns, behaviour by regime, which sleeves carry it. Treat its return as a ceiling.
- Read the paper record for execution: did the entries land where the rules said?
- Read the live record for reality, and read its length first.
- Check that the three are kept apart, and that none is quoted over a shorter window than the others.
If a record passes all four, you can learn from it, whether or not you end up following it. If it fails any, you are reading marketing.
Why the desk does it this way
Not out of modesty. Out of the observation that a labeled record is the one thing a reader cannot produce for themselves and almost nobody publishes, which makes it worth paying for. Every figure on the site, in the letter and in the members room carries its label; a number without one does not get published, by rule, enforced by a linter before a human ever approves it.
Backtested results are hypothetical, do not represent actual trading, and have inherent limitations. Paper results are from a simulated account. Live results are from real capital and are shorter and separate. Past performance of any kind does not predict future results. Nothing here is personalised investment advice.