A record 77 cross-exchange arbitrage windows opened today, headlined by a 19.21% HMSTR spread between Binance Futures and Bitunix. Full breakdown of the top five plays, fee-adjusted profit math, exchange pairing patterns, and tomorrow's watchlist for arb desks.
Seventy-seven arbitrage windows crossed the desk today, and the fragmentation across venues was as loud as it's been in weeks. When low-cap and mid-cap names start printing double-digit spreads between a spot venue and a futures venue on the same day, it usually means one of two things: either liquidity providers are asleep on one side of the book, or a listing/delisting event is scrambling price discovery. Today looked like both. The headline number is HMSTR's 19.21% spread between Binance Futures ($0.000326) and Bitunix ($0.000342) — a gap wide enough that even accounting for double-sided fees and slippage, it left real room for a competent execution desk. Behind it, a second HMSTR window opened later in the session at 10.40%, meaning the pair round-tripped twice in a single day. That's the kind of signal that tells you order books on Bitunix are thin relative to the majors, and market makers aren't keeping the venue tight. Beyond HMSTR, the list runs deep: LAB, C98, TA, DOT, VELODROME, BULLA, ALCX, and TLM all cleared the 9% threshold, which by any conventional arb-desk standard is a green light to at least model the trade. With 77 total events logged, this wasn't a quiet day of stray mispricings — it was a full board. The rest of this report walks through the top five in granular detail, breaks down which exchange pairs are doing the heavy lifting, and runs the fee math so you know exactly what's left on the table after costs.
The clearest pattern across today's 77 events is Bitunix's recurring role as the outlier venue. It shows up as the sell-side leg in the HMSTR (19.21%), LAB (12.62%), and TLM (9.05%) trades, and as the buy-side leg in the TA (11.56%) trade. That's a strong signal that Bitunix's order books, particularly on lower-cap listings, are not being kept in tight sync with the majors — likely a combination of lower market-maker participation and slower price-feed updates relative to Binance, Coinbase, and Gate. For an arb desk, that makes Bitunix a venue worth monitoring continuously rather than opportunistically; the mispricings there aren't random, they're a structural feature of thinner liquidity provisioning. Binance Futures also appears repeatedly, both as a buy-side source (HMSTR, VELODROME, BULLA) and, in HMSTR's second print, as the higher-priced sell side — indicating that Binance's futures book itself can drift from spot reality on lower-cap perpetuals during volatile stretches. Gate Futures shows up twice as the higher-priced sell leg (TA, and implicitly TLM's buy side), suggesting Gate's futures pricing on smaller-cap names tends to run hot relative to spot venues, possibly due to funding-rate-driven premium buildup. Coinbase, notably, only appears as the buy-side (cheaper) leg in both C98 and DOT — consistent with Coinbase's historically more conservative, slower-to-react pricing on altcoins relative to offshore venues that see faster order flow. Bitget shows up split — as a cheap buy-side source for LAB and as a pricier sell-side for VELODROME — which doesn't point to a consistent directional bias for that venue today, more likely reflecting asset-specific liquidity conditions rather than a systemic feed lag.
Every spread on this list falls into one of two buckets, and knowing which bucket you're in should dictate your sizing before you place a single order. Bucket one is the thin-book, high-percentage, low-dollar-depth trade — HMSTR, TLM, BULLA, VELODROME all fall here. These tokens trade at fractions of a cent to a few cents, and a 9-19% headline spread can evaporate to 2-3% net after you push more than a couple thousand dollars through either leg, because you're walking the book. The right approach here is small, fast clips — think $500-2,000 per leg, executed as close to simultaneously as your infrastructure allows, with no attempt to scale into size. Chasing depth on these names just guarantees you eat your own spread through slippage. Bucket two is the deep-liquidity, lower-percentage-but-still-meaningful trade — C98's Coinbase/Binance spread is the clean example. Both venues can handle real size, so this is where a desk should actually deploy capital: $10,000-50,000 or more per leg is plausible without meaningfully moving either book. The tradeoff is time — larger transfers and larger fills take longer to complete, so you're more exposed to the spread closing mid-execution. The practical rule: size up on major-exchange pairs, size down and move fast on thin offshore venues like Bitunix, and treat any spread involving a large-cap asset trading at an obviously distorted price (see DOT above) as a data-integrity check first, a trade second.
Take the HMSTR trade as the walk-through. Gross spread: 19.21% (buy $0.000326 on Binance Futures, sell $0.000342 on Bitunix). On a $2,000 notional per leg, that's a gross profit of roughly $384 before any costs. Now subtract trading fees: assume 0.05% taker fee on Binance Futures and 0.10% taker fee on Bitunix (typical for aggressive fills on both), which is 0.15% combined, or about $6 on $4,000 total volume moved across both legs. Subtract withdrawal/transfer costs if you're physically moving HMSTR between venues rather than running pre-funded balances — HMSTR network fees are typically negligible (well under $1 on most chains it's issued on), but the real cost is time-in-transit risk, not the fee itself. Net result: roughly $377 of the original $384 gross survives to the bottom line, a net spread of about 18.9% — still an outstanding trade if you can execute both legs near-simultaneously with pre-positioned capital. Compare that to the LAB trade: gross spread 12.62% on $11.869 buy-in, meaning $1.498 profit per unit before costs. On the same $2,000 notional, that's roughly $252 gross, minus combined fees of about 0.15-0.20% (higher-priced assets often carry standard percentage fees rather than flat ones), landing net profit around $246, or roughly 12.3% net. As a rule of thumb for this market structure: don't bother chasing anything under about 3-4% gross spread once you account for two-sided taker fees (typically 0.1-0.2% combined), potential withdrawal delays, and the slippage tax on thinner books. Everything on today's list of 77, at a floor of 9.05% (TLM), clears that bar comfortably — which is exactly why this was a strong day for the arb desk.
Given today's pattern, keep Bitunix on permanent watch against Binance Futures and Gate Futures for the rest of this week — the venue showed up disproportionately across today's 77 events, and thin-liquidity venues rarely fix themselves overnight. HMSTR and TLM are the two names most likely to reprint a spread tomorrow given today's double-appearance and the apparent structural liquidity gap versus the majors. Watch the Coinbase-Binance pair on C98 as well; if that gap was feed-lag rather than a one-off news reaction, it could reopen during the next volatility spike, and it's the one trade on this list large-liquidity desks can actually size into meaningfully. The best windows to monitor are typically during the Asia-to-US session handoff (roughly 00:00-04:00 UTC) and again around the US market open, when liquidity provisioning across smaller exchanges tends to thin out fastest relative to the majors. Keep a standing alert on any Bitget-versus-Bitunix pairing too — today's LAB and VELODROME prints suggest that combination is worth tracking even outside the specific tokens that triggered today.
Seventy-seven cracks in the market today, and HMSTR blew the widest one open at 19.21%. Fund both sides of your book, keep your clips small on the thin venues, and go big only where the liquidity can actually take it. Watch Bitunix like a hawk — it's not going anywhere.
Arbitrage Hunter — July 4, 2026