Two traders load the same system onto the same $50,000 account on the same morning. One runs it on one contract, the other on three. A year later the first is sitting in the range the backtest described, and the second has blown through two accounts. Same software, same signals, same market. The only thing they did differently was position sizing, and it decided everything.
Position sizing is the quietest decision in automated trading and the one that does the most damage when it is wrong. It has nothing to do with how good a trade looks or how confident you feel. It is arithmetic: how many contracts, tied to how much you can afford to lose, tied to the account you are trading.
Sizing to confidence is discretionary trading's oldest mistake, and it survives into automated trading because it feels responsible. It is not. In a rule-based system there is no "this one looks good," because you do not see the trade before it happens and every trade is sized the same way in advance. The only lever you control is how many contracts ride on each signal, and you set it against your account size and the drawdown you can survive, never against a hunch about the next few trades.
Confidence is not a position size. Your account is.
Size to the drawdown you can sit through, not the profit you want. Contracts scale profit and loss at exactly the same rate, so the right number is the one that keeps the worst historical drawdown survivable on your account and inside your prop firm's rules.
The Algomyth engine places its stop and its target in the market at the same instant as the entry, with the target set at three times the risk. On one micro contract, as an example, that is roughly $250 of risk and $750 of target on a typical trade. The point of that fixed-risk design is that the loss on any single trade is defined before the trade is live, not discovered afterward.
The realized figures over the backtest come in a little tighter, because not every trade runs all the way to its stop or its target. What does not move is the shape: the target is three times the stop on every entry, so breakeven sits at a 25% win rate. That combination, a system that does not need to be right most of the time because it wins bigger than it loses, is the whole engine, and how it holds together is covered in risk, reward and win rate.
How many contracts you can trade is not a free choice. It is capped by the account. A $50,000 account runs one micro contract, which is what the backtest models. A $100,000 account allows up to two, a $250,000 up to five. Those are the sizes the prop firms permit, and the ceiling rises with the account because a larger account can absorb a larger drawdown.
| Sizing level | Annual average, net | Average annual drawdown | Worst drawdown of the period |
|---|---|---|---|
| Low | $9,407 | $2,051 | $2,637 |
| Medium | $18,814 | $4,101 | $5,273 |
| High | $37,629 | $8,202 | $10,546 |
Read the last two columns, not the first. The drawdown is what scales with you, trade for trade, and the right-hand figure is the one that ends accounts. Note that the average annual drawdown and the worst of the period are not the same number, and that it is the worst one you have to survive.
Here is the part the profit math hides. Moving from Medium to High doubles the annual average, from $18,814 to $37,629. It also doubles the worst drawdown, from $5,273 to $10,546, and moving down to Low halves both. The pain grows exactly as fast as the gain, because index futures are linear: there is no contract count at which the reward outruns the risk.
So the question is never "how much could I make." It is "how deep a hole can I sit in without switching the system off or tripping the account's drawdown rule." A drawdown ends more accounts than a bad strategy does. Start with what drawdown is before you settle on a contract count.
Hypothetical performance. The $5,273 drawdown and the other figures above come from a backtest of the Algomyth engine over historical data, not from a live account. Simulated results carry no financial risk and cannot fully reflect real execution or slippage. Scaling drawdown by contract count is an approximation, and real drawdowns can be deeper. Past performance, real or simulated, does not guarantee future results.
Put in your account balance and a contract count and see the backtest scaled to your numbers, the drawdown as well as the profit.
Each published set of figures belongs to one sizing level. The $87,359 net, the $5,273 drawdown, the 1.46 profit factor: all of it describes the Medium level on a mid-size account. The moment you run a different size than the one you read, you are looking at the wrong row. The report gives you all three so you do not have to guess at the arithmetic.
Starting at the tested size is not timidity. It is the only way the numbers you looked at still describe what you are about to experience. Scale later, after the system has behaved as expected through at least one real drawdown, not before. How much you actually need to begin is worked out in how much money do you need, and the funded route in funded capital.
Almost every sizing failure is one of these three, and all three are the same error wearing different clothes.
The system removes the decision to enter. Position sizing is the one real decision it leaves you, which is exactly why it should be the most boring part of your operation.
Take the High level on a mid-size account. The report puts it at $37,629 a year net, with a worst drawdown of $10,546. Now ask the only question that matters before you fund it: if the account fell $10,546 from a high over a few weeks, on your own money, would you leave it running?
If the answer is no, then High is the wrong level for you, whatever the account permits. Drop to Medium, or to Low. The permitted maximum is a ceiling, not a target, and the trader who treats it as a target is the one who does not survive the first ordinary bad stretch. The broader picture is in how much a trading bot actually makes, and the instrument itself in how to trade futures.
Start with the smallest level the report publishes, Low, and move up only when the drawdown column stops worrying you. Your account's contract cap is a ceiling, not a recommendation. The right number is the largest one whose scaled drawdown you could sit through without switching the system off, which for most people is smaller than the ceiling.
No. The per-trade edge stays exactly the same; a larger size simply multiplies both the profit and the loss by the same factor. Going from Medium to High doubles the expected profit, $18,814 to $37,629, and doubles the drawdown with it, $5,273 to $10,546. You are not improving the system, you are amplifying it in both directions at once.
Because a profit figure with no size attached is not checkable. Every published figure, from the net profit to the drawdown to the 1.46 profit factor, belongs to a stated level on a stated account, and the headline set describes Medium on a mid-size account. That is what makes the numbers reproducible: you can load the same strategy at the same level and print the same report, which you cannot do with a figure that has no size attached.
You can, but a good month is not a reason to. The system has no memory and the market does not owe you a repeat, so changing size based on recent results is discretionary sizing wearing an automated system's clothes. Scale on account size and tested behavior over time, not on a streak, or you will reliably be largest right before the average reasserts itself.
About $250, with a target near $750, because the Algomyth engine places the stop and the target in the market at entry and sets the target at three times the risk. The realized averages over the backtest run a little tighter, because not every trade reaches its full stop or its full target. The design risk is the number to size against.
Everything we write comes back to one idea: verify, then decide. The live terminal exists for exactly that.
Unlock Algomyth Live