Position Sizing In An AI Strategy - The Three Numbers To Set

A strategy with a great entry and no sizing rules is a strategy that can bet too much.

On Everstrike, three numbers keep an AI strategy's size under control:

1. Max Trade Quantity - the largest single order.
2. A maximum position - the most it may hold.
3. A risk rule - how much each trade may lose.

The first is a setting. The other two go in your prompt.

1. Max Trade Quantity: the per-order cap

Every strategy has a Max Trade Quantity: the maximum USD value for each order it submits.

It's a hard limit on a single order - the AI can't exceed it. Start low. The FAQ's advice is to begin with something like $10, check that the strategy's events look right, and only then increase it.

2. A maximum position: the total cap

Max Trade Quantity limits each order, but a strategy that buys every hour can still build a large position, one order at a time.

So say how much it may hold in total, in the prompt:

The AI respects position limits you provide. It sees your open positions, orders, equity and available cash on every evaluation, so it knows where it stands against the limit.

You can also ask it to hold a target rather than a maximum. The AI can maintain a specific target position size, reducing or increasing the position as the market moves:

3. A risk rule: how much a trade may lose

The third number is the stop.

That one line does two jobs:

• 20% of the balance per trade sizes each trade relative to the account, so sizes shrink after losses and grow after gains.
• A 3% stop caps the loss on each one.

Together they decide the worst case: a stopped trade costs 20% x 3% = 0.6% of the balance.

Fixed or proportional?

Two ways to size each trade:

• Fixed: "Each trade is $200." Simple, predictable, easy to check in the events.
• Proportional: "Each trade is 20% of the balance." Scales with the account automatically.

Fixed sizing is easier while you're validating a strategy. Proportional sizing is the more natural choice once it's running and you trust it.

Either way, make sure the per-trade size fits inside Max Trade Quantity - or the setting will cap it.

Leverage

Size and leverage interact. A $3,000 position on a $1,000 account is 3x. On perpetual futures, the distance to liquidation at 3x is roughly 33% - 0.5% = about 32.8% (before funding). At 10x it's about 9.5%.

If your stop is 3%, both are fine. If you let the strategy build a $20,000 position on $1,000, the liquidation price comes within about 4.5% - uncomfortably close.

A simple account-level rule prevents that:

Scaling up

When the events look right and the strategy has a track record:

1. Raise Max Trade Quantity first.
2. Then the maximum position in the prompt.
3. Add markets or a faster evaluation frequency last - those change costs, not just size.

Wrapping up

• Max Trade Quantity (a setting) caps each order - start around $10.
• Put a maximum or target position in the prompt; the AI respects it.
• Write a risk rule with a size and a stop; their product is your worst case per trade.
• Fixed sizes are easier to validate; proportional sizes scale with the account.
• Keep total exposure, and so leverage, in check with an account-level rule.