Crypto Newbie / Simulators / DEX-CEX Arbitrage
DEX-CEX Arbitrage Simulator
The screen shows ETH at $3,510 on Binance and $3,500 on Uniswap — "$10 per ETH free profit!" Reality: gas $20 + Binance withdrawal $5 + DEX slippage on $50k worth ~$15 + 5-minute transfer during which ETH drops $8 = total cost $48. The $10 spread becomes a $38 LOSS. This simulator runs the full all-in math so you can see when an arb actually pays vs when it's a trap. Saved many beginners from buying expensive lessons.
Prices
Trade size
Costs + assumptions
Direction
DEX cheaper → buy DEX, sell CEX
Arbitrage result
Gross spread
0.29%
Gross profit
$100.00
Profit per unit
-$4.52
Cost breakdown
Gas cost
$20.00
CEX withdrawal fee
$5.00
DEX slippage (expected)
$12.25
Price drift during transfer (1-σ)
$107.95
✗ NOT profitable — costs eat the spread
Net profit: -$45.20
Margin (net / gross): -45.20%
Break-even check
Break-even spread per unit: $2.50
The minimum spread per unit needed just to cover fixed costs (gas + CEX fee). If gross spread < break-even, the arb is mathematically impossible regardless of size.
The five costs retail traders forget
(1) DEX gas: $5-50 per swap on Ethereum (often $1-3 on L2s). (2) CEX withdrawal fee: $1-30 fixed depending on asset. (3) DEX slippage: scales with (your_size / pool_depth). On thin altcoin pools, 1% slippage is common at $10k+ sizes. (4) Transfer time: ETH 5-15 minutes, BTC 30-90 minutes, often longer during congestion. (5) Price drift during transfer: based on volatility, BTC moves ±0.3% per 5 minutes on average; on volatile days, ±1%+. Together these typically eat 0.5-2% of the trade value — the gross spread needs to exceed this to profit.
Why pro arb desks make money where retail can't
Pro arb desks: (1) Pre-position inventory on BOTH venues — no transfer needed, sell on the high side, immediately rebuy on the low side. (2) Sub-second execution via dedicated infrastructure — no time for price to drift. (3) Volume discounts on CEX fees — pro Binance accounts pay 0.02% taker vs 0.1% retail. (4) Programmatic monitoring — never miss an opportunity, never act on stale data. Retail competing on the same trades is buying expensive education. The math doesn't work without pro infrastructure.
When retail arb can actually work
Three setups: (1) NEW LISTINGS — when a token first lists on a CEX, retail Telegram alerts often spot persistent 5-15% spreads vs DEX for hours. Requires being subscribed to the right alerts + fast wallet setup. (2) BRIDGE ARBITRAGE — cross-chain spreads (Stargate, Across) can stay wide for hours during liquidity imbalances. Profit accessible to retail with $1-10k capital. (3) FUNDING RATE ARBITRAGE — cash-and-carry on perps (long spot + short perp, capture funding) — this is closer to passive income than arb, but the principle is the same: known structural inefficiency, modest size, patient execution.
Calculating expected drift mathematically
Volatility scales with √time. If BTC's annual volatility is 80%, the standard deviation of price moves over time t (in years) is 80% × √t. For 5 minutes = 5/525600 of a year, std dev = 80% × √(5/525600) = ~0.24%. That's the 1-sigma move; expect adverse movement ~0.24% of trade size during transfer. For a $100k BTC arb attempt with 5-min transfer, expected adverse drift is $240. Add gas + fees + slippage, and the arb needs > $500 gross spread just to break even.
Frequently asked questions
+Why don't aggregators (1inch) handle DEX-CEX arbs?
Aggregators only route between DEXs (on-chain venues). CEX-DEX requires off-chain coordination (your CEX account, withdrawal authentication, deposit confirmation). Some institutional platforms (Wintermute, Galaxy) offer 'unified routing' that includes CEX legs, but they're for funds, not retail. The closest retail product is cross-chain bridges (Stargate's auto-rebalancing, which essentially arbs between bridge sides).
+What about high-volume CEX-DEX arb bots — they must work?
Yes, for those who run them — at scale, with capital pre-positioned on both sides, dedicated colocation, and 24/7 monitoring. Estimated profit margins are 0.05-0.2% per trade, requiring millions in inventory to be meaningful. The 'high-volume retail bot' selling courses online is a scam: the math at retail scale doesn't work after fees. If it did, the pros would have already captured it.
+Can I make money arbing on lower-fee chains?
Maybe. Arbitrum + Base have gas costs of $0.05-2 per swap vs Ethereum's $5-50. This shifts the break-even point lower, making smaller arbs viable. But CEX-DEX still requires withdrawal fees + transfer time. The math improves; it doesn't become a free-money machine. Realistic L2 arb profit for retail: ~$5-50 per opportunity, of which you'll find maybe 5-10 per day after monitoring.
+What's 'triangular arbitrage' — different from DEX-CEX?
Triangular: ETH → USDC → DAI → ETH, all on the same DEX. If the round trip results in MORE ETH than you started with, you profit. Modern DEXs have tight cross-pool pricing — triangular opportunities exist mostly on newly-launched pools or during volatility. Bot-dominated; retail rarely competes successfully. The simulator above focuses on DEX-CEX (different venues) rather than triangular (same venue).
+Why does the simulator's 'expected drift' matter so much?
Because it's the LARGEST cost for most retail arbs and the EASIEST to overlook. Gas + fees are visible numbers you see before clicking. Drift is invisible — it only shows up when your CEX deposit confirms 6 minutes later and the price has moved 0.5% against you. Many 'failed arbs' in retail are actually MATHEMATICALLY EXPECTED — the trader didn't account for drift, so they assumed they should profit, and were surprised when they didn't. The simulator makes this cost visible.