Boring Boris runs the numbers on 38 arbitrage flashes across the CEX complex, headlined by a 15.32% GPS spread between OKX and Binance Futures, plus recurring PORTAL windows worth pre-funding capital for.
Thirty-eight arbitrage flashes crossed the scanner today, spanning a headline 15.32% gap on GPS down to a still-solid 4.87% print on that same ticker later in the session. That's a busy day — a quiet session rarely clears a dozen prints above 3%, and today logged more than three times that, concentrated in repeat offenders: GPS, PORTAL, and BTW alone account for over half of the top ten.
GPS tops the board at 15.32%, buy-side OKX against sell-side Binance Futures — the kind of gap that shows up when a thin order book lags the pack by minutes, not milliseconds. BTW follows at 8.90% (Bitget to Bybit), then two separate PORTAL prints, 8.21% and 6.90%, across two different venue pairs. That repetition matters more than any single number: a token mispricing once is noise, mispricing three times in one session across different corridors is a pattern worth watching live.
One flag before the breakdown: today's feed returned $0.0M across pump volume, dump volume, buy pressure, and sell pressure. Every spread below is priced off top-of-book only — treat the percentages as real, but confirm actual executable size yourself before sizing a position.
GPS — 15.32% spread, buy OKX $0.009515, sell Binance Futures $0.010132. The board's biggest gap, and unsurprisingly it's on a sub-cent token — a few ticks either way swings the percentage hard, and thin books like this lag price discovery. No volume came through the feed, so check depth past level one on both venues before sizing; small caps like GPS often show a fat top-of-book quote and little behind it. No window duration was tracked either — assume this snapshot is already closing. The real complication: the sell leg is Binance Futures, a short position, not a spot sale, so funding rate and margin risk stack on top of the arb if it isn't closed instantly. My take: executable only with capital pre-funded on both venues, hitting both legs together. A cold-start transfer here is a losing trade before it begins.
BTW — 8.90% spread, buy Bitget $0.390761, sell Bybit $0.404822. A meatier unit price than GPS means less tick-noise, so this one likely reflects a real, if temporary, liquidity gap rather than rounding. No volume figure came through, so verify the book before committing size. Bitget-to-Bybit is a well-trodden retail arb corridor — fast, reliable rails, but also watched by plenty of other bots, so the window is short. Main risk is Bitget's withdrawal queue, historically slower than Bybit's during busy periods. My take: executable for a mid-size clip if pre-funded on both sides, and worth chasing even cold-start given the corridor's track record — though don't expect the full 8.90% to survive a 10-15 minute transfer wait.
PORTAL — 8.21% spread, buy Bybit Spot $0.015340, sell Binance $0.016599. The cleanest print on the board — both legs spot, no futures-margin complication. Sub-two-cent price still means small-cap slippage risk, so check both books directly since no depth data came through. Bybit Spot to Binance has solid withdrawal infrastructure for a token like this when the network isn't congested, though PORTAL has a history of congestion spikes tied to airdrop activity, which can stretch confirmation times right when speed matters most. My take: the most approachable trade on this list for a desk without balances pre-funded everywhere — the corridor and asset combination give the best odds of the spread surviving a live transfer.
PORTAL — 6.90% spread, buy Bitunix $0.017010, sell Binance Futures $0.017895. A second PORTAL window on a different corridor — Bitunix on the buy side this time. Two PORTAL prints in one session on two different pairs suggests fragmented price discovery for the token right now, not a one-off fluke. Bitunix is the thinner-liquidity name here, so expect wider effective spreads on size than quoted; no volume data came through, so verify before sizing. As with GPS, the sell leg is Binance Futures — a spot-long/futures-short basis trade, not a simple buy-sell, requiring active management of funding and margin. My take: executable for desks already running basis trades with futures infrastructure; not a good first trade for manual spot-only arb.
ALICE — 6.80% spread, buy Binance $0.136700, sell Coinbase $0.146000. Textbook Coinbase premium — US retail demand runs ahead of the offshore market. Highest unit price on the board, which usually means tighter books, though no volume figure came through today. The catch is the exit: Coinbase's withdrawal and fiat off-ramp process is slower and more deliberate than the other venues here, and moving ALICE onto Coinbase to sell needs either an existing balance or a transfer eating into the window. Coinbase premiums are also famously fickle, compressing in minutes once other bots catch up. My take: executable only with capital already sitting on Coinbase — not worth a cold transfer, the premium has a habit of vanishing before deposits clear.
Across the top ten, Binance and Binance Futures sit on the sell side four times (GPS x2, PORTAL x2), while OKX, Bitget, and Bitunix consistently show up as the cheaper buy-side venue. That's not random — it points to Binance's larger retail flow pulling price up faster than smaller venues can match on lower-cap alts. Bybit is the interesting outlier, appearing on both sides: buy-side for PORTAL (8.21%) and APR (5.27%), sell-side for BTW (8.90%) and PORTAL (5.04%). It isn't systematically cheap or expensive — its position depends on the asset's flow that day, so watch it both directions rather than assuming a fixed lean. Coinbase's single appearance (ALICE, sell-side) is the cleanest case here, the classic US-premium trade. Exchange51 and KuCoin each appear once as the pricier sell leg against Bybit and Bitunix respectively, suggesting these smaller venues lag price discovery on illiquid names rather than run a structural premium. Bottom line: watch the OKX/Bitget/Bitunix-to-Binance corridor first — it produced four of today's ten best prints.
Every spread here trades speed against size, and today skews fast-and-small. Sub-two-cent books like GPS and PORTAL are wide relative to their spread — the quoted percentage often only holds for the first few hundred to low-thousands in notional before slippage bites. Push five figures into a GPS-sized book and both legs get walked, turning 15.32% into something much smaller by the time you're filled. BTW and ALICE, at higher unit prices, tolerate more size before slippage hurts, but neither is deep enough to treat like a major-pair arb. Practical guidance: probe with a few hundred dollars per leg on any new small-cap print to measure realized slippage before scaling, and cap small-cap size at whatever you're comfortable losing if the fill goes wrong. On the larger-notional trades the constraint flips — more size available, but a faster-closing window, since better-capitalized desks are watching the same liquid pairs and will compress the gap within minutes.
Take PORTAL at 8.21% (buy Bybit Spot $0.015340, sell Binance $0.016599) as the cleanest worked example — spot-to-spot, no futures leg. On a $5,000 position, gross profit runs roughly $410. Strip out trading fees at a realistic 0.08-0.10% taker per leg on both exchanges — about $8-10 total — and you're still over $400 net before transfer costs. The part that actually decides the trade is withdrawal: cold-starting (no pre-funded Binance balance) means moving PORTAL from Bybit first, at a flat fee typically in the $0.50-$3 range — cheap in dollars, but the real cost is the 5-20 minutes of transfer time during which the spread often closes. If it compresses to, say, 3% by the time funds land, that $410 theoretical profit becomes roughly $150 after fees — still positive, but a third of the headline. Pre-funded execution, both legs filling within seconds, keeps essentially the full ~$400 net. That's the whole game: pre-funded capital captures the quoted spread, cold-start capital captures whatever's left after the market corrects. Rule of thumb — don't bother chasing anything under roughly 1% if pre-funded and near-instant; fees and minor slippage eat most of a sub-1% edge. Cold-transfer trades need considerably more buffer, generally 3%+ on liquid names and more on anything as thin as today's sub-cent prints, just to survive the mean-reversion during your transfer window.
A few flags apply across nearly every print today, not just one name:
GPS and PORTAL each printed three times today across different venue pairs — the strongest signal for tomorrow. Put the OKX-to-Binance and Bitunix-to-Binance corridors on the watchlist first. The broader pattern — Binance and Binance Futures sitting expensive against OKX, Bitget, and Bitunix — is worth pre-positioning for: keep working capital funded on the cheap-side venues so you're ready to hit the Binance sell leg the moment a gap opens, rather than starting a transfer after the fact. Thin small-cap spreads like today's tend to open widest during low-volume windows — late US evening into Asia pre-market, roughly 22:00-02:00 UTC — when market makers thin out and one venue's book drifts from the pack. Watch Bybit specifically; it sat on both sides of today's board and isn't a fixed-direction venue right now, so whichever way it leans tomorrow will likely pair with Binance or Bitget again.
That's the board for today — fifteen-plus percent on GPS, three separate PORTAL windows, and a Coinbase premium doing exactly what Coinbase premiums do. Fund your accounts, check your depth, and don't trust a spread that hasn't survived the time it took you to read this sentence.
Arbitrage Hunter — August 17, 2026