I pulled live funding rates this morning for every coin listed on all 5 of Bybit, Binance, Bitget, Gate and MEXC — 431 coins with full coverage. For each one I checked whether four of the five exchanges landed on one of the known floor values (0%, ±5.48%/year, ±10.95%/year — the fixed interest-rate component I wrote about on the 26th) while the fifth exchange sat somewhere else entirely, more than 20 annualized percentage points away.
GLM: four exchanges say 5.48%, one says -127%
The cleanest example in the whole set. Bybit, Binance, Bitget and MEXC all priced GLM's funding at exactly 5.48%/year this morning — the standard floor, meaning their premium index for GLM was sitting at or near zero on all four venues. Gate priced the same coin, at the same moment, at -127.35%/year. Same asset. Same 8-hour funding window. A 132.8 percentage-point gap between one exchange and the other four, who all agree with each other down to the decimal.
That's not four venues disagreeing with each other and one being average. It's four venues independently landing on the identical default, and a fifth showing a real, locally-priced number — which means Gate's order book for GLM had genuine, sustained short pressure this morning that never reached the other four books.
24 of 431 coins show the same pattern
GLM isn't unique, just the most extreme case. Running the same test — 4 exchanges on a floor value, 1 diverging by 20+ points a year — across all 431 coins found 24 of them (5.6%) fitting the pattern this morning:
| Coin | Odd exchange | Its rate | Other 4 agree on | Gap |
|---|---|---|---|---|
| GLM | Gate | -127.35% | 5.48% | 132.8pp |
| O | Binance | 70.14% | 5.48% | 64.7pp |
| BB | Bybit | -43.46% | 5.48% | 48.9pp |
| OPG | Gate | -40.41% | 5.48% | 45.9pp |
| ERA | Gate | -38.11% | 5.48% | 43.6pp |
| WAXP | Gate | -35.37% | 5.48% | 40.8pp |
One exchange shows up way more than it should
If the odd-exchange-out were random, each of the 5 exchanges would show up in roughly a fifth of the 24 cases — about 5 each. That's not what happened. Gate was the diverging exchange in 15 of the 24 cases (62.5%). Bitget and Bybit each turned up as the outlier 4 times. Binance showed up once. MEXC never did — every one of its readings agreed with the floor whenever the other three did too.
My best read on this: Gate's order books for a lot of these lower-volume names are thinner than the big three, so a modest amount of one-sided flow moves its premium index further than the same flow would move Bybit's or Binance's book for the same coin. That's a guess about mechanism, not a measured fact — I haven't pulled order book depth to confirm it. What I do know for certain is the pattern itself: it repeats, and Gate is disproportionately the exchange doing the diverging.
What this is worth, honestly
Two days ago's post on the raw cross-exchange gap found 28 coins clearing a 100%/year spread out of the full 624-coin universe on 7 exchanges. This filter is stricter and smaller on purpose — it throws out cases where multiple exchanges disagree with each other for no clear reason, and keeps only the cases where a consensus default is being broken by exactly one venue. That's a cleaner signal that something specific is happening on one order book, not just noisy dispersion.
It's still not free money. Capturing GLM's 132.8-point gap means holding a long on one of the four agreeing exchanges and a short on Gate at the same time, paying taker fees on four fills to open and close both legs, and carrying the liquidity and basis risk of a coin thin enough to produce a gap like this in the first place. Run the real numbers through the funding arbitrage calculator before treating any of this as a trade idea — on a coin this illiquid, the fee and slippage side of that calculation usually matters more than the headline spread.
What I actually do with this
I've added the 4-agree-1-diverge check to how I read the live funding rates page day to day — a coin sitting on a round floor number everywhere is background noise, and a raw big gap between two random exchanges is often just two illiquid books disagreeing with each other. The pattern worth a second look is the specific one from today: everyone on the same default, and exactly one exchange somewhere else. It's rarer, and when Gate is the one doing it, I now know that's the expected default, not a surprise.
Before you read a cross-exchange funding gap as an opportunity
- 24 of 431 coins showed 4 exchanges agreeing on a floor value while 1 diverged by 20+ points a year this morning — a stricter, cleaner signal than a raw biggest-gap ranking.
- GLM was the extreme case: 4 exchanges at 5.48%/year, Gate at -127.35%/year — a 132.8 percentage-point gap on the same coin, same moment.
- Gate was the odd one out 15 of 24 times (62.5%) — Bitget and Bybit 4 each, Binance once, MEXC never.
- Fees and liquidity eat into the raw spread fast on coins thin enough to produce a gap this size — check the real math before assuming it's capturable.
→ Funding arbitrage calculator · → Live funding rates · → Why funding rate floors exist · → Learn
FAQ
What does it mean when one exchange's funding rate diverges from all the others?
Checking 431 coins listed on all of Bybit, Binance, Bitget, Gate and MEXC this morning, 24 of them showed a specific pattern: four of the five exchanges sat on the exact same funding-rate floor value (the default interest-rate component, not a live positioning read) while the fifth exchange diverged by more than 20 annualized percentage points. When 4 out of 5 venues agree on a flat default and one doesn't, the one that doesn't is showing a real, locally-priced signal, not noise.
Which exchange most often diverges from the pack on crypto funding rates?
In this morning's snapshot, Gate was the outlier exchange in 15 of the 24 coins that showed the 4-agree-1-diverge pattern (62.5%), followed by Bitget and Bybit with 4 each and Binance with 1. MEXC never appeared as the odd one out in this sample.
What was the biggest single-coin funding divergence found?
GLM. Bybit, Binance, Bitget and MEXC all priced GLM funding at exactly 5.48%/year (the standard floor value) this morning, while Gate priced it at -127.35%/year — a 132.8 percentage-point gap on the same coin, at the same moment, on venues that otherwise agreed.
Can you actually trade a funding rate divergence like this for profit?
Not easily, and rarely for free. Capturing a cross-exchange funding gap means holding a long on the cheap venue and a short on the expensive one simultaneously, paying taker fees on both legs to enter and exit, and carrying basis and liquidity risk on what are usually thin, low-volume perpetuals — exactly the kind of pair likely to gap or de-list. The gap is real. The math after fees and slippage on an illiquid coin usually isn't as good as the raw number implies.