27 arbitrage windows fired across the board today, led by an 8.91% STAR spread between Gate Futures and Binance Futures. GPS dominated the tape with five separate opportunities scattered across Bitget, Binance Futures, Aster, OKX and Bybit — a textbook sign of fragmented, thin-book liquidity worth watching closely tomorrow.
Twenty-seven arbitrage opportunities crossed the scanner today, and the board had a distinctly fragmented flavor. This wasn't one clean dislocation on one pair — it was a scattershot of spreads across a dozen venue combinations, with one token in particular, GPS, showing up in five of the ten largest windows. That's the kind of signature you see when a mid-cap token's liquidity hasn't consolidated yet: order books on Bitget, Binance Futures, Aster (Exchange51), OKX and Bybit are all pricing the same asset slightly differently at any given moment, and none of them are deep enough to arbitrage the others away instantly.
The headline print was STAR at an 8.91% spread — buy on Gate Futures at $0.127000, sell on Binance Futures at $0.138310. That's a big enough gap to survive fees, slippage, and a bit of execution lag with room to spare, assuming the size was there to fill it. Close behind, NIL posted a 7.62% spread between Binance Futures ($0.048020 buy side) and Bitunix ($0.051680 sell side) — notable because NIL also showed up later in the session on the opposite side of the trade, which tells you the price gap between those two venues is oscillating rather than settling. GPS, meanwhile, printed five separate spreads ranging from 4.44% to 6.03%, touching five different exchanges. If you're running a cross-exchange bot today, GPS is where your fill logs are going to be busiest.
One caveat up front: today's feed came back with $0.0M on pump volume, dump volume, buy pressure, and sell pressure across the board. That means we've got clean price-spread data but no confirmed depth behind these prints. Treat every number below as a real, observed quote gap — not a guaranteed fill size. More on that in the risk section.
Binance Futures was the busiest single counterparty on the board today, showing up in six of the ten largest prints — and notably on both sides of the trade. It was the sell-side premium venue against Gate Futures (STAR) and against Bitget (GPS, 6.03%), but it was also the buy-side discount venue against Bitunix (NIL), against Bitunix again (GPS, 4.69%), and against Bybit (BTR). That two-way behavior is the signature of a high-volume venue whose price occasionally leads and occasionally lags smaller venues depending on which side of the market is moving faster — Binance Futures isn't structurally cheap or expensive today, it's just liquid enough to be first to reprice in both directions.
Bitget showed up three times and told a more one-directional story: sell-side premium against Exchange51/Aster (GPS, 5.43%) and against Bitunix (NIL, 4.74%), but buy-side discount against Binance Futures (GPS, 6.03%). Bitget is behaving like a venue that lags the fast movers slightly — useful as a sell target when the market's moving up, since its book hasn't caught up yet. Bitunix showed the opposite personality within the same session: it was the sell-side premium venue against Binance Futures for NIL early on, then flipped to the buy-side discount venue against Bitget later. That kind of whipsaw usually means Bitunix's book is thin enough to overshoot in both directions rather than tracking the broader market smoothly.
Exchange51 (Aster) appeared twice with an inconsistent role — cheap buy-side against Bitget on one GPS print, expensive sell-side against OKX on another. Given Aster is a newer mapping in our feed with liquidity still building out, that inconsistency is more likely a symptom of a maturing order book than a genuine repeatable edge. KuCoin and Bybit each made a single appearance (buy-side for the token labeled '4' against Binance Futures, and sell-side for BTR against Binance Futures respectively) — not enough data points today to call a pattern, but both are worth keeping on the watchlist given they're generally reliable, decently deep venues when they do show a spread.
There's a real tradeoff on today's board between chasing the single big print and working the repeatable small ones. STAR's 8.91% spread is the fattest edge available, but it's a one-shot opportunity on a thinner futures venue (Gate) — you get one bite, and if your fill drags because the book can't absorb your size, the effective spread you capture shrinks fast. Compare that to GPS, which threw off five separate spreads between 4.44% and 6.03% across five different venue pairs in the same window. That's not one big trade, it's a pattern you can work repeatedly and in parallel — smaller edge per leg, but more total surface area, and if one leg's book runs dry you can rotate to another GPS pair instead of sitting on your hands.
Slippage matters more than the headline percentage on every one of these. A 6% quoted spread on a venue pair with real depth (Binance/Bybit) is often more capturable in full than an 8.9% spread on a pair involving a thinner book (Gate Futures, Bitunix, Aster) — because market impact on the thin side eats disproportionately into the edge as size scales up. My rule of thumb for position sizing on prints like these: cap any single leg at a size that would move the thin-side book by no more than roughly 10–15% of the quoted spread itself. If you don't have live order book depth in front of you, err smaller and use limit orders on the illiquid leg rather than market orders — you give up some fill certainty, but you protect the edge that made the trade worth doing in the first place.
Let's walk the STAR trade end to end, since it's the cleanest structure (futures-to-futures, no on-chain transfer needed) and the biggest gross edge on the board.
Now the NIL trade, which involves a smaller, less liquid venue (Bitunix) on the sell side and may require an on-chain or exchange-internal transfer if you're not pre-positioned on both books:
The general takeaway: round-trip costs on a two-leg CEX arb typically run 0.3–0.9% once you stack fees, realistic slippage, and any transfer costs — more if a leg touches a thin book like Gate Futures, Bitunix, or Aster, less if both legs are on top-tier depth like Binance or Bybit. That puts my personal floor for 'worth chasing' at around 1.5% gross spread for a clean same-collateral futures pair, and closer to 2.5–3% gross for anything requiring an on-chain or cross-venue transfer where withdrawal delay is also eating into your capital efficiency. Every opportunity on today's board cleared that bar comfortably — the lowest print was 4.42%.
The single biggest flag on today's data: pump volume, dump volume, buy pressure and sell pressure all came back at $0.0M. That means every spread above is a confirmed price dislocation, but none of it comes with verified depth behind it. Don't assume a quoted spread is fillable at meaningful size just because it's on the board — verify live order book depth on both legs before committing capital, especially on the thinner venues.
GPS is the name to watch first thing tomorrow. Five spreads across five venues in one session strongly suggests this token's liquidity is still fragmenting rather than consolidating, and that pattern rarely resolves in a single day — expect fresh GPS windows to open between Bitget, Binance Futures, Aster, OKX and Bybit again, particularly around the Asia/US session overlap when volume picks up and books get repriced fastest. NIL is the second name worth a standing watch: it flipped direction between Binance Futures/Bitunix and Bitunix/Bitget within today's session alone, which means the underlying price relationship between those venues is unstable and likely to throw off another window.
For timing, keep an eye on the funding settlement windows (00:00, 08:00, and 16:00 UTC) — futures-involved spreads like STAR and BTR tend to open or close around those marks as positioning adjusts. And keep Aster/Exchange51 on a shorter leash than the established venues; as its liquidity matures, spreads against it should get both more frequent in the short term and less reliable to actually fill — worth tracking whether that trend holds tomorrow.
Twenty-seven windows, one standout at 8.91%, and a GPS liquidity story that's just getting started. Size your clips to the thin side of the book, respect the funding clock, and don't let a fat headline spread talk you into skipping the depth check. Stay sharp out there.
Arbitrage Hunter — August 18, 2026