Arb Desk Report: QNT's Six-Headed Spread and a 48% VELVET Gap — September 10, 2026

🎯 Arb Desk Report

One hundred forty arbitrage opportunities crossed the scanner today, and the top of the board is loud. VELVET printed a 48.34% spread between Binance Futures ($0.045700) and Bitunix ($0.062276) — the kind of number that makes you check your data feed twice before you check your wallet balance. Right behind it, TUT ran a 41.03% gap from KuCoin ($0.015420) to Bitunix ($0.020129), and then the board gets genuinely interesting: QNT shows up not once but six times, spreads ranging from 35.58% to 36.95%, every single leg buying on OKX and selling into either Bybit or Binance Futures. That is not noise. That is a pricing feed on one venue sitting stale or thin relative to the rest of the market for an extended stretch, and it is exactly the pattern an arb desk should be logging for tomorrow, not just today.

Rounding out the top tier, HEI spread 30.39% (Binance Futures $0.098730 to Bitunix $0.116785) and SIREN spread 28.19% (Bitget $0.021390 to Binance Futures $0.023323). Notice the venue that keeps appearing on both sides of the ledger: Bitunix and Binance Futures show up as both buy and sell legs across different pairs today, which tells you liquidity conditions on those two venues were choppy in both directions, not just one. For an arb trader, today's data isn't a lottery ticket — it's a map of where order books were thin and where they weren't. Let's walk the top five in detail, because the headline percentage is the least useful number on this list. The buy price, the sell price, and how fast that window closed are what actually decide whether you made money or donated to a market maker.

🏆 Top 5 Arbitrage Opportunities

1. VELVET — 48.34% spread, buy Binance Futures $0.045700 / sell Bitunix $0.062276. This is the widest print on the board and, honestly, the one I trust the least at face value. VELVET is a micro-cap with thin order books on both venues involved — Bitunix in particular is known for running lighter depth on lower-liquidity alts, so a 48% number here is as likely to reflect a stale quote or a single fat-fingered order absorbing the top of book as it is a genuine, fillable gap. If it was real, the window was almost certainly measured in seconds, not minutes — this kind of dislocation on a sub-cap token gets arbitraged shut by bots faster than a human can open two tabs. Executable for a retail-speed trader with manual execution: unlikely. Executable for someone with API keys pre-loaded on both venues and pre-funded balances: maybe, for a few hundred dollars before slippage ate the edge. This is a watch-and-log entry, not a chase-it entry.

2. TUT — 41.03% spread, buy KuCoin $0.015420 / sell Bitunix $0.020129. Same family of problem as VELVET: sub-cent pricing, low market cap, and Bitunix on the sell side again — that's two of our top three opportunities routing liquidity out through the same venue, which either means Bitunix's book was unusually generous today or unusually stale. KuCoin generally has decent depth for mid-tier alts, so the buy leg is plausible; the risk sits entirely on the sell side. Withdrawal from KuCoin to Bitunix (assuming you don't already hold TUT balances pre-positioned on both exchanges) would typically eat 5-20 minutes depending on the chain and confirmation requirements, and a 41% spread rarely survives that long unless it's structural rather than a one-off print. My take: executable only if you were already holding inventory on both sides. Cross-exchange transfer kills this one for a cold start.

3. QNT — 36.95% spread (the best of the six-pack), buy OKX $49.00 / sell Bybit $66.92. This is the most interesting opportunity on the whole board, not because it's the biggest number, but because it repeated. QNT is a real, liquid, top-100 asset with real order book depth on OKX, Bybit, and Binance — this isn't a thin-book anomaly, it's a pricing discrepancy on an asset that should arbitrage itself shut within seconds under normal conditions. A 36.95% gap surviving long enough to register — let alone six variations of it across two counterparty venues — points to either a data feed issue on one side of the pipe, or a genuine, sustained OKX-side mispricing (possibly a stablecoin/fiat pair quirk, a regional listing price divergence, or an OKX orderbook glitch). Liquidity risk is low here since QNT trades real volume on all three venues; the risk is entirely about whether the price you're reading is the price you can actually hit. This is the one entry on today's list worth calling your OKX contact about before wiring size.

4. HEI — 30.39% spread, buy Binance Futures $0.098730 / sell Bitunix $0.116785. Sub-10-cent token, futures-to-spot cross venue — this adds a wrinkle most of today's list doesn't have: you're comparing a derivatives price to a spot price, which is not a clean arbitrage unless you're also managing basis and funding rate exposure on the futures leg. If you bought the perp outright rather than delta-hedging, you're not capturing 30.39% risk-free — you're taking directional exposure with a discount, which is a very different trade. Liquidity on Binance Futures for a name like HEI is typically fine; Bitunix on the sell side is the same question mark as VELVET and TUT. Executable as a pure arb: only if you can actually settle the futures position into spot inventory to deliver against the Bitunix sell, which most retail accounts can't do cleanly. Treat this as a signal, not a trade.

5. SIREN — 28.19% spread, buy Bitget $0.021390 / sell Binance Futures $0.023323. Smallest spread of the top five but arguably the cleanest setup: Bitget spot has reasonable depth for mid-cap alts, and Binance Futures is about as liquid as it gets on the sell side. The spot-to-futures wrinkle from HEI applies here too, but the smaller gap means smaller basis-risk consequences if you're wrong about convergence timing. Withdrawal friction between Bitget and Binance is generally on the faster end (both are Tier-1 CEXs with quick confirmation SLAs for major chains), so the practical window was probably longer than VELVET or TUT's. My take: this is the most plausibly executable trade in the top five for someone running a semi-automated setup with pre-funded balances on both venues.

📊 Exchange Spread Patterns

The standout pattern today is OKX vs. Bybit/Binance Futures on QNT — six separate prints, all buying on OKX between $48.81 and $49.66, all selling on either Bybit ($66.92-$67.40) or Binance Futures ($66.63-$66.75). That's not six independent coincidences; that's one persistent pricing gap being sampled six times as it drifted slightly. When you see the same buy-venue/sell-venue pair repeat across a session with a stable-ish spread band (35.58%-36.95%, a tight ~1.4 point range), the read is a structural mispricing — most likely an OKX-side quote issue — rather than a transient liquidity air pocket. Compare that to VELVET, TUT, and HEI, which each show up once: those look like one-off dislocations on thin books, the kind that self-correct within a candle or two and don't repeat.

Bitunix is the recurring sell-side venue across three of today's top five (VELVET, TUT, HEI), which is worth flagging as its own pattern: either Bitunix's order books were running consistently rich relative to the rest of the market today, or its price feed was lagging. Binance Futures shows up on both sides of the ledger — buy side for VELVET and HEI, sell side for QNT and SIREN — which tells you it wasn't a directional Binance issue, just a venue with enough volume to appear everywhere. No Hyperliquid prints made the top ten today, which is a change from sessions where DEX-vs-CEX basis dominates the board; today's story is entirely CEX-to-CEX, and specifically concentrated on OKX's pricing relative to its peers.

⚡ Speed vs Size Analysis

There's a real tradeoff embedded in today's board between the sub-cent micro-caps (VELVET, TUT) and the liquid mid-cap (QNT). The micro-cap spreads are bigger on paper — 48.34% and 41.03% — but they're only bigger because the books are thinner; try to push $20,000 through a VELVET or TUT arb and you'll walk the order book on both legs, and the effective spread you actually capture could be a fraction of the quoted number. These are trades for small size, executed fast, or not at all. QNT is the opposite profile: real depth on OKX, Bybit, and Binance means you could push meaningfully larger size — $50,000-$100,000+ — without moving the market materially on either leg, assuming the mispricing itself is real and not a feed artifact. The rule of thumb: quoted spread and executable size are inversely correlated on this list. Size into the liquid QNT-style setups, keep the micro-cap plays small and fast, and never assume you'll get the full quoted spread on anything under a few million in daily volume.

Slippage compounds in both directions on a cross-exchange arb — you eat it buying in, and you eat it again selling out, and if you're not already holding inventory on both venues, you eat transfer time in between where the spread can move against you before you ever place the second leg. For anything in the HEI/SIREN futures-to-spot category, add basis risk on top of slippage: the futures price can converge toward spot faster than you can transfer and sell, compressing your edge before settlement. Position sizing recommendation: cap single-leg size at roughly 1-2% of the thinner venue's visible top-of-book depth for the micro-caps, and size QNT-style liquid setups based on your total capital allocation to arb strategies, not the order book — the book can absorb it.

💰 Profit Calculations

Take the best QNT print as the walkthrough: buy OKX at $49.00, sell Bybit at $66.92, 36.95% gross spread. On a $10,000 position: $10,000 / $49.00 = 204.08 QNT acquired. Sold at $66.92 = $13,656.98 gross proceeds. Gross profit: $3,656.98. Now the fees. Assume 0.10% taker fee on the OKX buy leg ($10.00) and 0.10% taker fee on the Bybit sell leg (~$13.66) — total trading fees roughly $23.66. Then the withdrawal leg: moving QNT off OKX to fund the Bybit sell typically costs a flat network fee, call it 0.5-1 QNT on the ERC-20 rail, or roughly $25-$50 at these prices — say $30 as a working estimate. Net profit: $3,656.98 − $23.66 − $30.00 ≈ $3,603.32, or a net return of about 36.0% on capital deployed for that leg. Even after fees and withdrawal, a spread this wide barely gets dented — which is exactly why a 36.95% print surviving on a liquid top-100 asset is the headline of the day, not VELVET's flashier 48.34%.

Now flip to a realistic worst case: a 2% spread on a similarly liquid pair. $10,000 position, 0.10%+0.10% trading fees (~$20) plus a flat withdrawal fee — on many chains that's $1-$5 for stablecoins or majors, but can run $10-$30+ for less common tokens or congested networks. On a 2% gross spread ($200 gross), fees alone can consume 10-25% of the edge, and that's before you account for the price moving against you during transfer time. The floor for this desk: don't bother chasing anything under roughly 0.5%-1% net-of-fees expected spread unless you're running a fully automated, pre-funded, both-sides-pre-positioned setup where withdrawal time is removed from the equation entirely. Anything above 5% on a genuinely liquid pair, like today's QNT cluster, is worth real size. Anything above 25% on a micro-cap, like VELVET or TUT, is worth investigating for data quality before it's worth wiring capital.

⚠️ Risk Alerts

Bitunix appeared as the sell-side counterparty on three of today's five biggest spreads (VELVET, TUT, HEI) — that concentration is itself a flag. When one exchange keeps showing up on the rich side of every trade, check whether its price feed is simply lagging the market before you commit capital, because arbing into a stale quote means your 'sell' fill may not clear at the price you saw. On the withdrawal side, none of today's opportunities are viable for a cold start without pre-positioned balances on both venues — TUT and VELVET's windows almost certainly closed before a KuCoin-to-Bitunix or Binance-to-Bitunix transfer would have confirmed. The QNT cluster is the exception: real liquidity, real venues, but verify OKX's quote against a second source (Binance spot, CoinGecko aggregate) before treating a 36%+ spread on a top-100 asset as real money rather than a feed glitch — that combination is unusual enough to warrant a sanity check every single time it shows up.

General low-liquidity warning applies to every sub-cent token on this board: VELVET, TUT, HEI, and SIREN all trade in the sub-$0.13 range, which is exactly the price band where a single large market order can move the book 10-20% and manufacture a fake arbitrage signal that vanishes the moment you try to size into it. Always check live order book depth, not just the last-traded price, before committing capital to any spread under a few million dollars in daily volume. And a standing reminder for anyone running futures-to-spot legs like HEI and SIREN: funding rate exposure and basis convergence risk are real costs that don't show up in the headline spread number.

🔮 Tomorrow's Setup

Top of the watch list: QNT. A spread that repeated six times in one session on a liquid, top-100 asset, consistently routed through OKX as the buy side, is the single most likely candidate to reappear tomorrow — either because the underlying feed issue hasn't been fixed, or because whatever regional/liquidity dynamic caused it is still in play. Check OKX QNT pricing against Bybit and Binance first thing; if the gap is still there, it's worth sizing up. Second watch: Bitunix broadly — with three of today's top five routing sell-side liquidity through it, keep an eye on whether that venue's book normalizes or keeps running rich against KuCoin, Binance Futures, and other peers on low-cap alts.

Best times to watch: the Asia-to-Europe handoff (roughly 00:00-04:00 UTC) and the US-open volatility window (13:00-15:00 UTC) tend to produce the thinnest cross-venue liquidity alignment and the widest transient spreads, especially on OKX-heavy pairs given its Asia-weighted volume base. Exchange pairs to prioritize for monitoring tomorrow: OKX vs. Bybit and OKX vs. Binance Futures (given today's QNT pattern), plus Bitunix against any Tier-1 venue for low-cap alts. If the QNT gap has closed by tomorrow's session, that confirms it was a transient feed issue rather than structural — log it either way, because a repeat next week would be a much stronger signal to build a standing monitor around.

Sign Off

Six QNT prints on the same buy-sell axis isn't luck, it's a pattern worth watching closer than any single flashy percentage on this board. Log it, verify it, and don't let a 48% number on a sub-cent token distract you from the boring, liquid, repeatable edge sitting quietly in the middle of the list. That's the trade. Arbitrage Hunter — September 10, 2026.