Two thirds of funding rates are a constant

Funding is the most-watched number in perpetual futures and the least understood. It gets called a sentiment gauge, a crowding signal and a leverage thermometer. The mechanics are published, and the published mechanics make the number much less informative than it looks.

I pulled the funding rate of every market Hyperliquid quotes, straight from its public info API, at 05:23 UTC on 5 October 2026. The result is a distribution with a very strange shape.

Of 234 quoted markets:

• about two thirds report funding of exactly +0.00125% per hour
• 56 report exactly 0
• a couple of dozen report anything else

Two thirds of the venue's funding rates are the same number to six decimal places.

The exact split moves during the session, because markets drift in and out of the clamp as their books thin and deepen. Across fetches an hour apart the pinned group ran between 153 and 156 of 234. The share is the durable part; the counts are a snapshot.

That number is a constant, not a reading

Hyperliquid's funding documentation states the interest component in plain terms: "interest rate component is predetermined at 0.01% every 8 hours, which is 0.00125% every hour, or 11.6% APR paid to short."

The published formula is:

F = Premium Index + clamp(interest - Premium Index, -0.0005, 0.0005)

Two things follow. The interest rate is fixed by the venue, not discovered from the market. And the premium term is clamped inside plus or minus 0.0005, which is the same magnitude as the interest term itself.

So when a market's premium is small, the clamp holds the result at the interest rate. The rate you see is the venue's cost-of-carry assumption sitting still.

BTC is one of the 156. At the time of the snapshot it showed a mark of $85,738, funding of +0.00125% per hour, which annualises to +10.95%. Open interest was $3.22bn on $1.82bn of 24-hour volume. Nothing about that reading tells you whether longs are crowded. It tells you Hyperliquid charges 0.01% every eight hours.

What a funding heatmap is actually showing

Aggregate the same data and the distortion is obvious. Over 200 of 234 markets show positive funding. That sounds like broad long crowding.

It is not. It is 156 markets reporting a fixed positive rate, 56 reporting zero, and 22 reporting something else. A heatmap coloured by funding sign would paint two thirds of the board the same colour for a reason that has nothing to do with positioning.

This is a general problem, not a Hyperliquid quirk. Every venue sets some baseline carry component, and every dashboard that colours by sign inherits it. The fix is not to distrust funding. It is to read the deviation from the venue's own baseline, and to check what that baseline is before treating a number as a signal.

How to check it yourself

The audit takes one request. The venue's info endpoint returns the universe and the context for every market in a single payload:

curl -s -X POST https://api.hyperliquid.xyz/info \
  -H 'Content-Type: application/json' \
  -d '{"type":"metaAndAssetCtxs"}'

Count how many markets share the most common funding value. If the mode holds the majority, the mode is the venue's, not the market's. Then read the venue's own funding page and find the interest component it publishes.

Do that once per venue and you know which markets on it can carry a real funding signal at all. On this venue, that is roughly 25 out of 234.

The same week, in volume

Funding is a positioning read, so it is worth pairing with what the market actually did. Over the week to 5 October, on KuCoin's XBT perpetual, the weekend produced a large close on a small amount of trading.

| Date (UTC) | Close | Turnover | Share of Friday | |---|---|---|---| | Friday 2 October | $84,473.50 | $340.8m | 100% | | Saturday 3 October | $84,712.80 | $50.9m | 14.9% | | Sunday 4 October | $86,469.10 | $107.6m | 31.6% | | Monday 5 October, partial | $85,726.90 | $41.4m | 12.1% |

Sunday closed +2.36% against Friday on 31.6% of Friday's turnover. Saturday, the thinnest session, ran at 14.9%. Monday had faded 0.86% from the Sunday close.

Across the last 30 daily bars, mean turnover ran $262m on Fridays against $99m on Sundays and $84m on Saturdays. In this sample, weekend prices move on a fraction of the volume that sets the week's reference levels.

The honest reading: a 2.4% move on under a third of normal turnover is a thin-book move. Weekend highs are cheap to print.

What this does and does not license

It does not mean funding is useless. On the 22 markets where the premium is doing real work, funding is a genuine measure of positioning, and extreme values on those markets are worth acting on. XMR was showing +0.0052% per hour on $72.9m of open interest at the time of the snapshot, which is roughly four times the fixed rate and therefore a real signal.

It also does not generalise. The clamp and the fixed interest component documented above are Hyperliquid's. Other venues set different baselines and this audit has to be run against each one separately.

And one venue over one week is a sample, not a rule. The turnover pattern is suggestive. It is not a finding about how markets behave.

The discipline worth keeping

Two habits come out of this, and neither needs a vendor or a subscription.

First, take the number apart before you interpret it. A rate with a documented formula has terms, and one of those terms may be constant.

Second, know when you are looking at thin volume. A price move on a third of normal turnover tells you about the order book, not about conviction. Both checks cost one request.

Sources

Every figure in this piece traces to one of these:

• Hyperliquid funding documentation: the predetermined interest component, the 0.01% per 8 hour figure, the 11.6% APR statement and the full formula with its clamp.
• Hyperliquid info API, request body {"type":"metaAndAssetCtxs"}, fetched 2026-10-05T05:23:25Z: the 234-market universe, every funding rate, mark prices, open interest and 24-hour volume.
• KuCoin futures klines, fetched 2026-10-05T05:52:03Z: daily open, high, low, close and turnover. Response shape confirmed against 200 of 200 OHLC-valid bars.

Cryptocurrency trading, leveraged perpetual futures, and automated algorithmic strategies carry significant risk of rapid and total financial loss. Never risk funds you cannot afford to lose completely.