Liquidation data across five exchanges

I’ve spent the last few weeks looking at liquidation data across five exchanges. Binance, OKX, Bybit, dYdX, Hyperliquid.

Because of course I have, I am trying to show there are much better ways to do this.

Different architectures, different risk engines, different insurance funds, different liquidation mechanisms.

Same structural problem: when a large position moves against a thinly capitalised participant, the exchange bears the loss. Insurance funds deplete (when they exist). ADL kicks in. Solvent participants get their positions reduced to cover someone else’s insolvency. Every hedge is broken.

October 2025: $19B liquidated in 40 minutes. $653M in preventable haircuts. 34,983 ADL executions. Across all five.

Q1 2026 added another $5B in cascading liquidations. January, February, March. Same root cause each time.

Every venue is running its own insurance fund, its own risk engine, its own margin model. Siloed. Bybit restructured its fund in December, increased loss absorption 200% per contract, but did it by fragmenting further into per-contract pools.

The pattern repeats because the architecture demands it.

This is not an execution problem. It’s an architecture problem, it’s intent and commercial decision making.

It’s so obvious when you look across the venues. There ought to be proper clearing to enable market makers to scale, which improves pricing and liquidity which reduced volatility because of liquidation.

It’s not difficult to see it should happen, it is all about getting movement.

This first appeared on LinkedIn on 7 April 2026. If you want to comment or discuss, that’s the place.