Flashcrash arbitrage:
DEX dumps before CEX reacts

When a token drops sharply in a DEX pool (large sell order, rug, hack, panic, project event), the on-chain price changes the moment the transaction executes. The CEX book catches up over minutes: market-makers pull their bids only when they see the DEX has actually moved down. In the gap between 'DEX already crashed' and 'CEX realised it crashed' a large spread opens — sometimes from 20% to 99%+ on catastrophic events.

Why DEX falls faster than CEX

A DEX pool works on the formula x · y = k: one large sell immediately moves price along the curve proportional to volume. No order queue, no delay — every transaction changes pool state in the same blockchain transaction. A CEX book, by contrast, is built from market-maker orders that actively adjust to the market. When drama happens (rug, hack, mass exit), CEX market-makers often step back — pull bids and wait. Until they post new low prices, the CEX 'hangs' at the old high.

Where the profit comes from

Trade logic: buy the token cheap on DEX (where price has already fallen) and simultaneously open a short on the CEX perpetual (where price is still high). When the CEX book catches down to DEX — the CEX side gains, and the on-chain token can be sold back at the now-aligned price. Profit comes from the gap closing, not from the down-move itself.

The full trade is two simultaneous operations: buy on-chain (where price is low) and short on the CEX perp (where price is still high). When CEX catches down to DEX, both positions close — profit comes from the gap, not from direction.

Why a perpetual rather than a spot sell

You could just sell the token on CEX spot at the high price. But that only works if you already own the token ON THE EXCHANGE — which for flashcrash arbitrage is almost never true (the DEX buy just happened, the token is in your wallet, not on the exchange). Transferring to the CEX and selling = too slow (5–15 minutes transfer = the window closes).

The perp solves both: (1) no token ownership required to open a short — it's a margin contract; (2) execution is instant, at the moment of click. Close is instant too. The perp settles by an index price reflecting multi-exchange consensus, so convergence to fair pricing is inevitable — that's the profit.

Trade timing

In the first 1–3 minutes after the signal, the spread is at its peak. After 5–10 minutes CEX market-makers usually start to adjust, the spread narrows. By 15–30 minutes on a typical dump the spread is mostly closed. Reaction must be fast — open both positions in the first minutes after the alert.

Unlike other arbitrage types, here the clock runs in minutes — the window lives short. The algorithm is optimised for speed: filter and checks done in advance, execution runs almost in parallel.

Step by step — a realistic flashcrash trade. Specific numbers model a typical signal: 35% dump in 5 minutes on a mid-liquidity alt, CEX catches up 25% over 8 minutes. This isn't the SFUND outlier (that's in the case below) but the typical picture of a modal catch.

Logic: the CEX short delivered +$220 (price moved from $1.00 to $0.78). The DEX side lost ~$150 (bought at $0.65, sold at $0.75, but slippage and fees ate some of the gain). Difference +$70 minus fees = clean ~$65 on the trade.

This is a typical profit on a mid-magnitude flashcrash. On large catastrophes (60%+ dumps) the same math gives $500–2000 on the same $1000 position — hence outlier results like SFUND in the case below. But the average catch is tens of dollars, not thousands.

Flashcrash arbitrage is rare big windows, not a daily flow. Typical rhythm — 3–8 signals per week, with spread depending on market activity. On quiet weeks it might be 1–2 signals; on volatile ones (mass liquidations, rugs, events on major projects) — dozens.

What drives results

Main factor — market volatility. Calm markets = few flashcrashes; turbulent periods (especially memecoin seasons and BTC drawdowns) = frequent large windows. The same signal delivers different results to two traders depending on reaction speed, position size, and how cleanly the exit is set up.

What's math, not luck

Average catch on a typical trade — $50–300 on a $1k position. That's smaller than the SFUND case ($100 → $3,028, ×30 multiplier) — the case is shown because it actually happened, not as an expectation. Most signals deliver moderate profit; rare catastrophic events produce outliers.

Two hard rules follow:

• Without a CEX perp — it's not flashcrash arbitrage. A spot listing alone isn't enough because there's nothing to short instantly.
• Without the anti-fake-dump filter — statistical loss. Half the 'dumps' in the wild are the start of real collapses, not bounces.

How flashcrash arbitrage differs from other types

Unlike regular CEX-DEX, the rhythm here is rare-but-large. Unlike cross-exchange, there's no transfer phase between exchanges — the position is held delta-neutral through the perp, not through physically moving the token. Unlike funding arbitrage, the hold is short (minutes, not days) and profit comes from a price gap, not accumulated funding.