Arb Desk Report: 45 Cross-Exchange Windows Fire, CP Rips 26% Between OKX and KuCoin

🎯 Arb Desk Report

Forty-five arbitrage windows crossed the desk today, September 7, 2026, and the top of the stack was not subtle. CP printed a 26.27% spread between OKX and KuCoin — buy side at $0.021300, sell side at $0.022784 — the kind of number that makes you check your data feed twice before you check your exchange balances. That wasn't a fluke either. The same ticker showed up again lower in the stack with a second, independent 14.38% spread against Coinbase, which tells me CP's price discovery was genuinely fragmented across venues today, not just a single stale quote glitching through one API.

Below CP, the book stayed unusually rich for low-cap and mid-cap names: AKE at 12.38% (Bybit to KuCoin), MARSCOIN at 12.28% (Binance Futures to Bybit), and a cluster of names — USELESS, BULLA (twice), KOMA, ticker '4', and FLOCK — all sitting in the 6.8%-8.4% range. That's a lot of double-digit and high-single-digit spreads showing up in one session. For context, a 'good day' on this desk is usually five or six spreads clearing 5%; today gave us ten in the top tier alone out of 45 total events. Liquidity was clearly the swing factor across the board — these are almost all low-float, low-cap tickers, and that's exactly the profile that produces spreads this wide. Deep, liquid pairs on major venues don't dislocate by double digits; thin order books on futures venues and mid-tier spot exchanges do. Volume and pressure telemetry came back at $0.0M across the board on this feed, so treat every window below as a book-depth call, not a volume-confirmed one — more on that in the risk section.

🏆 Top 5 Arbitrage Opportunities

1. CP — 26.27% spread, buy OKX $0.021300 / sell KuCoin $0.022784. This is the standout of the session and, frankly, the kind of spread that should make you suspicious before it makes you excited. A quarter-point spread on a token this size almost never survives contact with real size — the OKX bid-side liquidity at $0.021300 is very likely a handful of levels deep before slippage eats the edge, and KuCoin's sell-side depth at $0.022784 is the mirror problem in reverse. Windows like this on low-cap tickers tend to be measured in seconds to low minutes on the CEX order book itself, and they close the moment one market maker on either venue re-quotes. The real bottleneck isn't the spread, it's the plumbing: if you don't already have pre-positioned CP inventory sitting on both OKX and KuCoin, you cannot execute this — a same-token withdrawal-then-sell cycle takes minutes to tens of minutes depending on the chain, and the spread will be gone long before your transfer confirms. Executable only for desks running pre-funded dual-exchange inventory; not executable as a buy-here-sell-there round trip.

2. CP — 14.38% spread, buy OKX Spot $0.021070 / sell Coinbase $0.022350. The second CP print of the day, and notable because it's against a different counterparty (Coinbase, not KuCoin) at a different OKX price point — strong evidence CP was genuinely mispriced across three venues simultaneously rather than one exchange showing a bad tick. Coinbase's fill quality on a token like this is usually better than a smaller offshore venue, which slightly improves the sell-side execution risk here versus the KuCoin leg above. Still, Coinbase listing depth on lower-cap names is frequently thin outside the top of book, and Coinbase withdrawal/deposit processing for less common tokens can run slower than OKX's. Call this executable in principle for pre-funded inventory, but size it conservatively — the second-best price on the book is rarely close to the quoted level on a spread this wide.

3. AKE — 12.38% spread, buy Bybit $0.011734 / sell KuCoin $0.012603. A sub-penny token, which cuts both ways: tick-size rounding can distort the percentage spread on numbers this small, but it also means slippage in absolute dollar terms is tiny even if it's large in percentage terms. Bybit-to-KuCoin is a well-trodden arb corridor for smaller-cap listings, and both venues generally have workable withdrawal times for standard ERC-20/BEP-20 rails. The risk here is almost entirely book depth — sub-penny tokens frequently have order books that are wide and sparse below the top level, so the effective average fill price on either leg could be meaningfully worse than the quoted $0.011734/$0.012603. Executable for small-to-medium size; scale down hard before assuming linear fills for anything beyond a few thousand dollars notional.

4. MARSCOIN — 12.28% spread, buy Binance Futures $0.164040 / sell Bybit $0.175294. This one's a cross-product spread — futures on one side, presumably spot or futures on the other — which changes the risk profile entirely versus a pure spot-to-spot arb. If you're buying the Binance Futures leg, you're carrying funding-rate exposure and margin requirements on top of the price spread itself, and you can't just 'withdraw and sell' a futures position the way you can a spot balance. This spread is really a basis trade dressed up as an arb. It's executable, but only for a desk already running perpetual futures books on both venues with capital pre-allocated as margin — not a simple buy-low-sell-high round trip. Treat the 12.28% headline number as gross basis, not net arbitrageable edge, until funding rates are netted out.

5. USELESS — 8.39% spread, buy Bitunix $0.193593 / sell Gate Futures $0.209830. Another futures-involved spread, and Bitunix is a smaller, less battle-tested venue than the majors above — that alone bumps counterparty and withdrawal risk up a notch. At $0.19-$0.21, this is a higher unit-price token than the others in the top five, so slippage per unit will bite harder if the book is thin, and Gate's futures order books on lower-cap perpetuals can gap meaningfully around funding intervals. Executable for a desk that already has capital resident on both Bitunix and Gate, with position sizing kept modest given Bitunix's smaller footprint; not a spread I'd chase by moving fresh capital onto Bitunix specifically to catch it.

📊 Exchange Spread Patterns

Binance Futures showed up on both sides of the ledger today — sell side against Bybit (MARSCOIN) and Exchange51 (BULLA), buy side against Gate Futures (KOMA, ticker '4') and against Bitunix implicitly through the USELESS/Gate cross. That two-way appearance is the clearest pattern in the data: Binance Futures pricing on smaller perpetual listings is lagging or leading Gate Futures and Bybit inconsistently, which is exactly the kind of cross-venue futures dislocation that shows up when a token gets added to one venue's perp board before liquidity catches up. Gate Futures vs Binance Futures was the single most repeated corridor in the mid-tier of the list (KOMA, ticker '4', and one BULLA leg), which suggests Gate's futures market makers are consistently a step behind Binance's on newer or thinner perpetual listings — worth setting a standing watch on that pair specifically. OKX vs KuCoin and OKX vs Coinbase both appeared via CP, reinforcing OKX as a source of stale or fast-moving spot quotes on lower-cap tickers relative to the bigger venues. Exchange51 and Bitunix, the newer/smaller names in the mix, both showed up exclusively on the buy side of spreads — consistent with smaller venues generally posting the more favorable (i.e., stale-low) price when a token is repricing fast elsewhere.

⚡ Speed vs Size Analysis

The honest tradeoff on a session like this: the biggest headline spreads (CP at 26.27% and 14.38%) are also the ones with the shortest realistic execution windows and the worst depth-to-headline-price ratio, because a mispricing that large gets arbitraged out fast by anyone else watching the same feed. The 6-8% cluster (BULLA, KOMA, '4', FLOCK) is the more interesting trade set for anyone sizing up — smaller headline edge, but historically these mid-single-digit-to-high-single-digit spreads on futures-vs-futures corridors (Gate/Binance) persist longer because they're driven by structural funding/liquidity differences between venues rather than a single bad quote. Slippage math matters more than the headline number here: a 26% spread that costs you 15 points of slippage on each leg nets to roughly zero, while an 8% spread on a deeper futures book might cost you 1-2 points total and clear net positive. Position sizing recommendation: scale size to the shallower of the two order books' visible depth at the quoted price, not to your available capital — on these low-cap names, the visible depth is almost always the binding constraint, not your wallet.

💰 Profit Calculations

Take AKE as the clean worked example: buy on Bybit at $0.011734, sell on KuCoin at $0.012603, gross spread 12.38%. Trading fees: assume 0.1% taker on the buy leg and 0.1% taker on the sell leg (standard spot taker rates absent VIP tier discounts) — that's 0.2% off the top, leaving roughly 12.18% gross-after-trading-fees. Withdrawal fees are the real variable: moving AKE off Bybit to KuCoin (or pre-funding both and skipping the transfer) costs either a flat token withdrawal fee — call it the equivalent of 0.5%-1.5% of trade size on a sub-penny token given typical flat-fee-on-small-notional drag — or zero if you're running pre-funded balances on both venues and just rebalancing later. With pre-funded inventory: net profit lands around 12.0% after fees. With an active withdrawal-and-transfer cycle: net drops to roughly 10.7%-11.7%, and you've also eaten the transfer time during which the spread can (and often does) close. Now compare to the CP spread at 26.27%: same 0.2% trading-fee drag gets you to ~26.07% gross-after-fees, but if you have to actually move CP between OKX and KuCoin, withdrawal time alone likely kills the trade before settlement. The rule of thumb on this desk: below roughly 1.5%-2% gross spread, it's not worth chasing once both fee legs and any withdrawal cost are counted — that's your real floor, not zero.

⚠️ Risk Alerts

Every single spread in today's top ten is on a low-cap or micro-cap ticker (CP, AKE, MARSCOIN, USELESS, BULLA, KOMA, '4', FLOCK) — that's the single biggest structural warning on the board. Thin order books mean the quoted price is rarely the fillable average price for anything beyond small size. Volume and pressure data on this feed all came back at $0.0M, meaning there's no independent volume confirmation behind any of these prints — treat every spread here as a snapshot of quoted price, not a validated tradable size, until you check live depth yourself. Futures-involved spreads (MARSCOIN, USELESS, KOMA, ticker '4') carry funding-rate and margin risk on top of the price spread — don't read the headline percentage as pure arbitrage edge. Smaller venues in the mix — Bitunix, Exchange51 — carry higher counterparty and withdrawal-reliability risk than the majors; confirm withdrawal status isn't paused on a given token before committing capital, since smaller exchanges are more prone to freezing withdrawals during volatile repricing exactly like what's shown here. Finally, two tickers repeating in the top ten (CP twice, BULLA twice) with different counterparties each time is consistent with a token that's actively repricing across the board — that cuts both ways, since it means more windows, but also means the 'stale' side of the spread could just as easily be the side that's about to move against you instead of toward you.

🔮 Tomorrow's Setup

Keep the Gate Futures vs Binance Futures corridor on a standing watch — it produced three of today's ten top spreads (KOMA, ticker '4', one BULLA leg) and looks structural rather than incidental. If CP keeps showing dislocation across OKX, KuCoin, and Coinbase, it's worth flagging as a token with a listing or liquidity event in progress — that kind of triple-venue fragmentation usually persists for a session or two before market makers catch up, not just one print. Best times to watch: the first hour after Asia-session open and the first hour after US futures markets open tend to be when smaller-cap perpetuals reprice fastest relative to spot, which is exactly the pattern behind today's futures-involved spreads. Exchange pairs to prioritize tomorrow: OKX/KuCoin, OKX/Coinbase, Bybit/KuCoin, and Gate Futures/Binance Futures — all four produced real edge today and none of them look like one-off noise.

Sign Off

Forty-five windows, a 26% headline, and a market that's clearly still sorting out where CP actually trades. Watch the depth, not the headline — that's the whole job. Stay boring, stay funded on both sides, and don't let a pretty percentage talk you into a trade your order book can't support.

Arbitrage Hunter — September 7, 2026