September 8, 2026 delivered 97 flagged arbitrage windows led by a 33.58% CP spread between OKX Spot and Coinbase, and a 22.82% XAN gap between Gate Futures and Bitunix. This report ranks the top five opportunities, breaks down exchange spread patterns across OKX, Bitunix, Hyperliquid, and Coinbase, and runs real profit math after trading and withdrawal fees.
Ninety-seven arbitrage opportunities crossed the tape today, and the top of the board was not subtle. CP printed a 33.58% spread between OKX Spot, where it was bid at $0.023950, and Coinbase, where it traded up at $0.026034. That is not a rounding-error edge you nibble at with a market order — it's wide enough to survive slippage, fees, and a slow withdrawal and still leave real money on the table. Behind it, XAN posted a 22.82% gap between Gate Futures ($0.014512) and Bitunix ($0.015031), with a second, smaller XAN leg later in the session adding another 9.78% between Bybit Spot and Coinbase. SOPH was the volume workhorse of the day, throwing off four separate spreads between 8.70% and 14.52%, mostly pivoting around OKX as the cheap leg against KuCoin, Bitunix, and Hyperliquid on the expensive side. HEMI rounded out the top tier with an 8.61% futures-versus-futures spread between Gate and Binance. For desks running cross-exchange bots, today was a target-rich environment — but the printed spread is the gross number, not the number that actually clears. Everything below closes that gap between what the screen shows and what hits the P&L.
The clearest pattern in today's data: OKX was the cheap leg in five of the top ten prints (CP twice, SOPH three times), consistently undercutting Coinbase, KuCoin, Bitunix, and Hyperliquid. That's not random noise — it suggests OKX's pricing on these specific lower-cap alts is lagging the rest of the market, whether from thinner local order flow or slower price discovery on newer listings. Coinbase, meanwhile, sat on the expensive side of three separate trades (CP twice, XAN once), reinforcing a premium pattern rather than a one-off dislocation — worth watching over the next few sessions to see if it persists. Bitunix showed up as a recurring sell-side venue for both XAN and SOPH, which combined with its historically slower withdrawal processing makes it a venue you want to be already positioned on rather than transferring into mid-trade. Hyperliquid appeared on both sides of SOPH spreads (buy at $0.008350 against KuCoin, sell at $0.012185 against OKX), which is a reminder that as a decentralized perp venue its funding and basis dynamics move somewhat independently of the CEX spot complex — that decoupling is exactly what generates spreads like these. Gate Futures and Binance Futures each appeared once (XAN and HEMI respectively), both plausible cross-venue funding-rate divergences rather than simple spot mispricings.
The tradeoff on a day like today is stark. The SOPH cluster — four spreads between 8.70% and 14.52% — behaves like a fast-decaying opportunity set: thin order books on lower-cap alts reprice within seconds of a bot detecting the gap, so the realized edge shrinks the moment you start filling. These are quick, small-clip trades: think low four-to-mid-five-figure notional per leg, executed by automated systems with latency under a second between exchanges. Slippage on the SOPH pairs specifically should be modeled aggressively — assume you lose 20-30% of the quoted spread to book-walking on anything beyond a few thousand dollars notional per leg. The CP and XAN prints, by contrast, are the slower, larger-size opportunities: a 33.58% or 22.82% gross spread has enough cushion to absorb meaningful slippage and still clear comfortably in the green, which is exactly why they're worth working with limit orders and staged fills rather than racing them. Position sizing recommendation: size SOPH-style trades to what the top three order-book levels can absorb without material price impact, and size CP/XAN-style trades to what your standing inventory on both venues allows — the wider the gross spread, the more slippage budget you have to work with, so let spread size dictate aggressiveness of execution, not the other way around.
Walking through the CP trade on a $10,000 notional per leg: gross spread of 33.58% yields $3,358 in gross profit before costs. Trading fees at a typical 0.10% taker rate on both legs cost roughly $10 on the buy and $10.36 on the sell (fee scales with the larger post-profit notional), for about $20.36 total. Withdrawal costs depend on network — assume a conservative $10 flat equivalent for moving CP off OKX to fund the Coinbase sell, though this can run higher on congested chains or lower if CP trades on a cheap L2. Net profit lands around $3,327.64, or roughly 33.28% net on the trade — barely dented by fees because the gross spread was so wide. Compare that to the smallest print in today's set, CP's second entry at 8.03% (OKX $0.024661 → KuCoin $0.025700): gross profit on $10,000 is $803, minus roughly $20 in trading fees and $10 in withdrawal costs, netting about $773, or 7.73%. The takeaway: at typical CEX fee structures (0.05-0.10% taker per side) plus a modest withdrawal cost, anything under roughly 1.0-1.5% gross spread is not worth chasing — fees and withdrawal costs alone can consume the entire edge, and that's before accounting for slippage on execution. Today's full set of 97 opportunities, if the smallest unlisted ones dip below that floor, means the tail of the distribution likely wasn't tradeable at all — the top ten printed here are where the real money was.
Watch SOPH first thing — four spreads in one session against three different counter-venues (KuCoin, Bitunix, Hyperliquid) signals an actively mispricing order book that hasn't stabilized, and that pattern often bleeds into the next session before liquidity providers catch up. Keep OKX on the buy-side watchlist broadly; it was the cheap leg in half of today's top prints and that kind of lag doesn't usually correct in a single day. The Bybit-to-Coinbase XAN pairing is worth a standing alert given it showed up cleanly today with good depth on both sides. Best windows to watch: the early-session period when Asian-hours OKX and Bitunix flow diverges from US-hours Coinbase and Bybit pricing tends to produce the widest gaps, so set alerts for the first few hours after each region's session open. Keep an eye on HEMI and any other Gate Futures / Binance Futures pairs too — funding-rate divergences on futures-vs-futures crosses tend to cluster around funding settlement times, so check the funding schedule before assuming a repeat print.
Ninety-seven windows, one CP print worth building a whole session around, and a SOPH cluster that rewarded speed over patience. Keep your books deep, your withdrawals pre-funded, and don't let a wide gross spread fool you into skipping the slippage math. See you on the next scan. Arbitrage Hunter — September 8, 2026.