When ROI cannot be computed honestly

Why the denominator breaks

A trader who deposits mid-window changes the capital base without changing the trading. A withdrawal after a good run flatters every subsequent percentage. Transfers between an account and its sub-accounts look like flows in and out of a book that is actually intact.

Any ROI computed over a shifting denominator is an artefact of the flows rather than a measure of skill. Presenting one anyway would make the ranking look more complete and be less true.

• Mid-period deposits and withdrawals distort the base capital.
• Sub-account transfers can appear as flows without any economic change.
• Cross-margin shifts move collateral between markets without a trade.
• Short records make any annualised figure statistically meaningless.

What is shown instead

Where ROI is unreliable, realized closed PnL is used as the secondary metric. It is denominated in currency rather than percent, so it cannot be inflated by an unclear capital base, and it counts only closed trades — no open-position optimism.

The trade-off is that PnL alone says nothing about efficiency: a large PnL earned on enormous capital is a weaker result than the same PnL on a fraction of it. That is why PnL is read alongside profit factor and discipline rather than on its own.

How this affects model performance figures

Model portfolio ROI over 7D and 30D windows is an estimate derived from public trader history under model assumptions. It is not a record of any real account and does not include your own fill, timing and fee differences.

Past performance is not indicative of future results. Perpetual futures are leveraged instruments and carry a substantial risk of loss, including the loss of your entire position.