A 48-event arbitrage sweep flagged a 15.52% JCT spread between Gate Futures and Bitget as the standout play, with VELVET, C98, LAB and GWEI rounding out a strong day for cross-exchange traders. Full breakdown of prices, fees, and execution risk inside.
Forty-eight arbitrage opportunities crossed the desk on July 5, 2026, and the spread table was unusually fat at the top. JCT printed a 15.52% gap between Gate Futures ($0.003224 buy-side) and Bitget ($0.003724 sell-side), the kind of number that either means a genuine structural dislocation or a market so thin that the quoted price is barely real. Both are worth discussing, and we will do both below.
Below JCT, the list stays surprisingly rich: VELVET at 11.36%, C98 at 10.25%, LAB at 9.78%, and a cluster of mid-cap names (GWEI, HMSTR, ETHW, MAGMA x2, BTW) all sitting between 6% and 8%. That is a lot of double-digit and near-double-digit spreads for a single day, which usually signals either a broad volatility event moving prices unevenly across venues, or a run of low-liquidity altcoins where quote staleness inflates the apparent gap. The totals feed reported $0.0M across pump volume, dump volume, buy pressure and sell pressure — meaning the raw event stream carried price/spread data but no confirmed depth numbers. Treat every dollar figure below as an execution estimate, not a data-verified fill number. This report is written for traders who will check order book depth themselves before routing size; it is not a green light to fire market orders.
The venue pairings in today's data lean heavily on futures-vs-spot and futures-vs-futures combinations rather than the classic spot-vs-spot dislocation. Gate Futures appears on both sides of the ledger — the buy leg for JCT, the sell leg for VELVET and one of the two MAGMA spreads — suggesting Gate's futures pricing was running persistently out of sync with peers today, whether from funding-rate pressure or lighter arbitrage-bot coverage than Binance or Bitget carry. Binance Futures shows up as a buy-side venue three times (VELVET, LAB, MAGMA), which is notable because Binance Futures is normally the deepest, tightest-priced venue on the board — when it's the cheap side of a spread, that typically means the other exchange moved first and Binance hasn't been arbitraged back into line yet, a window that tends to close fast given how much bot liquidity watches Binance specifically.
Bitget and Bitunix both appear repeatedly as sell-side venues (Bitget on JCT and ETHW, Bitunix on GWEI and MAGMA), which fits their profile as mid-tier exchanges with retail-driven order flow that lags the majors in repricing. OKX shows up once, as the sell side on HMSTR, consistent with OKX generally running tighter to Binance than the smaller venues do — when OKX is on the expensive side of a trade, it's usually because a genuinely fast move happened and OKX's market makers haven't caught up yet, which argues for that spread being real and fast-closing rather than a stale-quote artifact. The overall pattern: Binance and Binance Futures act as the reference price, Gate/Bitget/Bitunix are where the lag lives, and Coinbase's single appearance (C98) stands out because Coinbase almost never lags — when Coinbase is the cheap leg, it's worth extra attention precisely because it's unusual.
Every spread on this list falls into one of two buckets, and mixing up which bucket you're in is the fastest way to turn a paper profit into a real loss. Bucket one: the JCT/GWEI/BTW-style sub-cent micro-cap spreads, where the percentage number is huge but the tradable depth is tiny — these are speed trades. You need resting size on both books before you click, you need to move in and out in seconds to minutes, and you should assume any attempt to push more than a few hundred dollars through will walk the price and eat half your edge in slippage. Bucket two: the C98/LAB-style spreads on larger-cap, better-listed tokens, where the percentage is smaller but the absolute liquidity is real — these can absorb more size, but the round-trip (transfer, confirmation, second leg) takes longer, during which the spread can compress or reverse entirely before you're done.
Position sizing should scale inversely with how thin the underlying token is, not with how wide the spread looks. A reasonable framework: cap micro-cap spread trades (JCT, GWEI, BTW, HMSTR territory) at whatever size you'd be comfortable losing entirely to slippage or a stale quote — often under $500 in practice for tokens trading at fractions of a cent. Reserve larger size for the Coinbase/Binance-class pairs where both venues can absorb a five-figure order without moving the tape. On the futures-vs-futures trades (VELVET, and the Binance/Gate leg of MAGMA), size should be governed by margin requirements and liquidation buffers rather than spot slippage — an 11% mark-price gap can widen further against you if funding keeps diverging, so leave room.
Take the C98 trade as the cleanest worked example, since it's spot-to-spot on two liquid CEXs. Gross spread: buy at $0.012200 on Coinbase, sell at $0.013450 on Binance, a gross gap of $0.01250 per unit, or 10.25%. Trading fees: assume 0.4% taker on Coinbase (a realistic non-VIP taker rate) plus 0.1% taker on Binance, for roughly 0.5% combined round-trip fee drag. Withdrawal: moving C98 off Coinbase to Binance typically runs a flat network fee, call it the equivalent of roughly 0.3-0.6% of a mid-size position depending on trade size and current network conditions — smaller trades eat a proportionally bigger bite. Net: 10.25% gross minus roughly 0.5% trading fees minus roughly 0.4% withdrawal drag leaves an estimated 9.3-9.4% net edge before slippage — still a strong trade, but note that nearly 1 full point of the headline 10.25% evaporates before you've accounted for a single tick of market impact.
Now the JCT trade: 15.52% gross, buy Gate Futures at $0.003224, sell Bitget at $0.003724. Futures venues typically charge 0.02-0.06% maker/taker on the derivatives leg, call it 0.1% total for entering and exiting the futures position, plus roughly 0.1-0.2% taker on the Bitget spot leg, plus a withdrawal/settlement cost that on a sub-cent token can be disproportionately large if the network fee is fixed in native units rather than percentage terms — potentially 1-3% of position value on very small trades. Net estimate: roughly 13-14% after fees on paper, but the real risk isn't the fee line, it's whether $0.003224 and $0.003724 are fillable prices at any meaningful size or just the last-printed tick on a thin book. As a rule of thumb for this desk: don't chase any spread under 3-4% net-of-fees on major-pair spot arbs, and don't chase anything under 6-8% gross on micro-cap or futures-basis trades, since fee drag and slippage compound faster on illiquid names.
The single biggest risk across today's list is withdrawal timing on the micro-cap names — JCT, GWEI, HMSTR, BTW, and MAGMA are all tokens where a network congestion event, an exchange maintenance window, or a simple withdrawal-limit cap can strand your first leg for hours while the spread you were chasing closes completely. Always check the destination exchange's deposit status page before sending funds, not after.
Given how consistently Gate Futures and Binance Futures diverged today, that pairing is worth a standing watch — if funding-rate pressure persists overnight, VELVET and MAGMA-style basis gaps could reappear or widen further at the next funding settlement window. The Bitget/Bitunix corridor also showed up often enough (JCT, ETHW, GWEI, MAGMA) to suggest either a liquidity gap or a listing/news event specific to those venues that hasn't fully worked through yet; watch for continuation on any token that traded through both exchanges today. Best times to watch for fresh dislocations are typically around major funding-rate resets and during the low-liquidity overnight hours for US and EU trading desks, when thinner order books make cross-exchange gaps more likely to open and linger before bots close them. Coinbase-vs-Binance is the pair to check first thing for a clean, high-confidence setup similar to today's C98 trade, since that corridor tends to produce the most reliably executable spreads on this list.
Forty-eight spreads, one clear headline number, and a board that rewarded traders who checked depth before chasing percentages. Watch your fills, watch your withdrawal queues, and don't let a 15% headline talk you into size the order book can't support. Arbitrage Hunter — July 5, 2026.