Trading bot risk: drawdown, capital and leverage

Historical return does not mean absence of risk

Drawdown measures the decline from an earlier peak to a later trough. A curve can finish in profit while suffering a large temporary loss along the way. That decline affects the ability to keep trading, available margin and the recovery subsequently required.

Losses and recoveries are asymmetric: after a 30% decline, approximately 42.9% is needed to return to the starting point; after a 50% decline, the required recovery is 100%. This is why ranking bots by return alone is not enough.

How LionLabs estimates combined risk

The Portfolio risk gauge uses active bots to calculate a score from 0 to 100. For each bot, it takes simulated maximum drawdown, adjusts it when configured leverage differs from backtest leverage, and weights it by the share of the account allocated to that bot.

The visual scale labels the result low, moderate, elevated, high or extreme. When the score exceeds 36, a red triangle appears beside the historical-period selector. It prompts a review of capital and leverage; it does not automatically block activation.

Nominal exposure: capital multiplied by leverage

Nominal exposure helps describe the economic size controlled by the bots. Assigning 1,000 USDC at 3× leverage creates approximately 3,000 USDC of nominal exposure. This does not mean the loss will be exactly 3,000 USDC, but it does amplify the effect of market moves on margin.

LionLabs shows total nominal exposure as a percentage of account value. A 100% nominal exposure does not mean a guaranteed 100% loss; it describes size, not outcome. Even so, higher exposure increases sensitivity to adverse moves and liquidation risk.

Account percentage or a fixed USDC amount

Fixed USDC can make the assigned amount easier to control. If the account grows while that amount remains unchanged, its relative weight falls. It does not make a trade safe or impose an absolute ceiling on every possible loss.

Why Vault mode requires particular care

In Vault mode, each entry is sized using 100% of that bot's internal operating capital. If a Vault contains 1,000 USDC and uses 3× leverage, its approximate nominal exposure is 3,000 USDC. The 100% refers to the Vault sizing base, not a statement that all capital will necessarily be lost.

A Vault is an accounting separation inside LionLabs, not a physical subaccount or separate wallet. Funds remain on Hyperliquid. Read the Standard and Vault guide to understand reinvestment and protected profits.

Compounding also magnifies future drawdowns in USDC

Reinvesting profits increases operating capital. If a similar percentage drawdown occurs later, its USDC amount will be larger because it applies to a larger base. Compounding is not inherently negative, but it makes periodic review of capital, leverage and combined risk essential.

This is easy to overlook: a historically profitable system can seriously damage an account when it uses all available capital, excessive leverage or several bots exposed to similar market moves.

What the gauge cannot know

The score does not predict the next drawdown, calculate liquidation probability or guarantee that loss will remain within historical figures. It adds contributions as an orientation, but does not fully model asset correlation or every extreme scenario.

• Bots can lose at the same time.
• Live conditions can differ from the backtest period.
• Slippage, fees, funding, liquidity and execution failures can alter results.
• An asset can exceed its worst historical decline.
• The displayed Vault allocation starts from initial capital; reinvestment can change later economic risk.

Use the gauge to compare configurations and detect concentration, not as a substitute for personal limits, supervision and adequate margin.

Frequently asked questions