The four kinds of liquidation, and how to budget for each.

Most traders treat liquidation as a single failure mode — the big red bar at the bottom of the account page. It is not. There are four distinct ways a perpetual futures position can reach zero, and each one has a different early-warning signature, a different speed, and a different correct response. We built our L1–L4 protection stack around these four modes. Here is what they are, what causes them, and how to size for each.

Mode 1: The margin call

This is the textbook one. Price moves against you, unrealized PnL eats your available margin, and when your maintenance margin ratio hits the exchange's threshold, the matching engine closes your position — usually at the worst possible price, against a thin liquidity book, after the insurance fund has taken its cut.

Early signature

A linear decline in your margin health ratio. On our health dashboard this appears as a green-to-yellow transition. You have time — typically minutes to hours — to react.

Correct response

L1 notification fires at 50% margin health. L2 closes 25–50% of the position at 30% health — which is often enough to push the ratio back into safety. You decide the thresholds; we execute them.

Mode 2: The cascade

A large seller hits the order book during a period of thin depth. Their market order walks through three or four price levels, triggering stop-losses and partial liquidations that also become market sells, which trigger more stops, which trigger more liquidations. What was a 0.4% move against you becomes a 4% move in 45 seconds.

Early signature

Not visible in your margin health. Visible in order book imbalance, in whale flow into the top-of-book, and in the rate of change of funding. By the time your margin ratio moves, the cascade is already past its first leg.

Correct response

You cannot react fast enough manually. You need pre-configured rules. On TradeFloor these are the L3 cross-exchange hedge (open an offsetting position on a venue that has not yet cascaded) and, failing that, L4 emergency close on the original venue. Both fire within the 60-second loop.

Mode 3: The funding bleed

Your position is fine on price. You are even modestly in profit on the entry. But you are on the wrong side of funding: the perpetual-contract funding rate has been paying the other side 0.1% every 8 hours, and you have been holding for three weeks. That is roughly 6% of your notional, gone, paid to the counterparty, with no corresponding price move required.

Early signature

Entirely visible, entirely ignored. Your account page shows "funding paid" as a line item that most people never look at. Until they do the reconciliation at end of month and find that half their drawdown had nothing to do with price.

Correct response

Cap maximum funding tolerance per position. If accumulated funding exceeds your threshold, close automatically regardless of price action. This is not a liquidation event the exchange will flag — it is a slow bleed you configure yourself as a close condition.

Mode 4: The one nobody talks about

Your own exit logic is broken. Your stop-loss was set below the liquidation price (so the exchange gets there first). Your take-profit ladder closes 80% of the position at 1R, leaving 20% exposed to a reversal that eats the remaining 80% of your expected win. Your close condition was "RSI < 40 on 15m" but you never noticed that RSI was recomputed on the forming candle, so it "crossed 40" for 3 seconds and triggered the close at the worst point of the dip.

This is the mode that ends accounts most often, and the one least discussed. It is not a market event. It is a configuration event.

Correct response

Validate your stops and take-profits against your leverage. Insurance sub-positions. Partial-close ladders that leave a meaningful runner. Exit signals evaluated on closed candles only. We check these at trade creation time and warn you before you commit.

Our L1–L4 stack, mapped to the four modes

How to budget

Pick a number you can afford to lose on this position. Call it your risk budget, R. Then:

• Size so that a 1R loss corresponds to hitting your L2 threshold, not your exchange's liquidation threshold. The gap between L2 and liquidation is your safety buffer.
• Set funding tolerance at roughly 0.3R. If the bleed exceeds that, you are in the wrong trade regardless of price.
• Insurance margin equal to 0.5R, available in USDT on the same venue, earmarked not for trading but for L4 top-ups.
• Take-profit ladder that realises at least 1R at the first level, so you are in a risk-free state before any reversal logic has to fire.

These four numbers — size, funding cap, insurance buffer, first-TP level — are the four knobs that matter. Everything else is detail.

// questions or corrections · [email protected] · more essays · /blog