Guides · Guide 05
Money management for copiers: how much, and what to expect
Sizing an allocation is a question about drawdowns, not returns. Here is what the record says a week, a month and a year look like.
- Weeks that were negative
- 52%
- 3-month windows negative
- 25%
- 12-month windows negative
- 1%
Start from the drawdown, not the return
The return you hope for is irrelevant to sizing. What decides the amount is the loss you can hold through without acting. Every strategy that works over years has stretches that look broken over weeks, and the copiers who lose money in a winning strategy are the ones who leave during those stretches.
Our backtest's worst drawdown was 18.8% and it took 241 days to make a new high afterwards. The next one may be deeper. So the question is: an amount that, down 20% for eight months, you would leave alone. That is your maximum. If the honest answer is zero, the right allocation is zero.
What a window of time looks like
Over 300 weeks of backtest, here is how often a holding period ended in a loss, and the range of outcomes:
| Holding period | Ended negative | Worse than (1 in 10) | Median | Better than (1 in 10) |
|---|---|---|---|---|
| 1 week | 52% | −2.5% | −0.1% | +6.1% |
| 4 weeks | 42% | −5.0% | +1.4% | +14.0% |
| 3 months | 25% | −6.0% | +7.0% | +25.6% |
| 6 months | 11% | −0.2% | +13.1% | +42.4% |
| 12 months | 1% | +8.3% | +31.6% | +65.1% |
Every overlapping window in the backtest, net of costs and funding. Hypothetical; the live record is a few days old and is published here.
Read the first row carefully. In a strategy that compounded at 41.8% a year, more than half of all weeks lost money. A month was negative two times in five. Only at a year does the distribution move decisively above zero. Anyone judging a copy after a week is reading noise, and anyone judging a lead trader by a week-old record is doing the same.
For comparison, Bitcoin held outright over the same windows: 49% of weeks negative, 49% of three-month windows negative, and 46% of twelve-month windows negative, with a one-in-ten twelve-month outcome of −47%.
Sizing rules that survive contact with a drawdown
- The mechanical floor. On Binance, a copy under about $1,000 cannot place many of the strategy's orders and drifts from the lead (Guide 03). Below that, the copy is a different strategy.
- The ceiling. The amount that can fall 20% and stay untouched, and that you will not need for at least a year. Twelve months is the shortest window over which the record has been reliably positive.
- One allocation, not a drip. Adding on the way down "to average in" changes your risk mid-drawdown at the moment you are least objective. Decide once.
- Never a loan, never rent. Obvious, and the most common way a survivable drawdown becomes a forced exit.
What to expect, stated plainly
A target of 20–40% a year over a full cycle, with drawdowns of 15–20% expected along the way, months of flat or negative returns as normal, and the possibility that the next drawdown is worse than any in the record. If that description is unattractive, the strategy is not for you, and the record will not change that.