Funding

An ordinary future has an expiry date: on that day its price converges with spot because the contract settles against spot. A perpetual future has no date — it can be held indefinitely. So nothing forces its price to meet spot on a given day, and with a skew in demand it could drift arbitrarily far from spot.

The anchor is funding — a regular payment between holders of longs and shorts. When the perpetual trades above spot (more people want leveraged longs), the rate is positive: longs pay shorts. Holding a long becomes more expensive, holding a short more rewarding; some participants close longs or open shorts, and some buy spot while shorting the perpetual, collecting funding. All of this pulls the contract price back toward spot. When the contract is below spot, everything mirrors: the rate is negative and shorts pay.

Three properties beginners confuse:

• The payment is between participants, not to the venue. The venue takes its trading fee, while funding moves from the longs' pocket to the shorts' (or the other way).
• Only position holders at the accrual moment pay. On most venues accrual happens at set times: whoever holds a position at that minute pays or receives for the whole interval. Close a minute before accrual — you pay nothing for that interval; open a minute after — the same.
• The sign of the rate is not the direction of price. Positive funding says leveraged longs are more expensive to hold, i.e. demand is skewed toward longs. Who turns out right does not follow from the rate.

Illustrative example of how the anchor works. BTC spot is $100,000, the perpetual is $100,150: basis +0.15%. More people want leveraged longs, the rate comes out positive, say +0.05% per 8 h. Now a trader has a clear calculation: buy 1 BTC on spot for $100,000 and open a 1 BTC short in the contract. The price can go anywhere — the two positions offset each other — while the short collects funding: $100,150 × 0.0005 ≈ $50 per accrual. The more such participants, the more selling in the contract and buying on spot, and the gap narrows.

In practice it costs more than on paper: entry and exit fees on two markets, the next interval's rate is unknown in advance, capital is tied up. So the gap does not vanish to zero but stays within narrow bounds, and funding rarely stays far from a small positive value for long.

The funding rate is a percentage per one accrual interval. Intervals differ between venues: most often 8 hours (three accruals a day), for some pairs 4 hours, and on Hyperliquid every hour. So raw rates of different venues cannot be compared: 0.01% on a venue with an 8 h interval and 0.01% on Hyperliquid are payments that differ eightfold. Hyperliquid's hourly rate looks tiny, though over a day it adds up to as much or more.

To compare, rates are normalised to one interval. EVA normalises to 8 hours:

rate per 8 h = rate × 8 / interval in hours.

• Hyperliquid, 0.0015% per hour: 0.0015 × 8 / 1 = 0.012% per 8 h.
• A venue with a 4 h interval, 0.025% per interval: 0.025 × 8 / 4 = 0.05% per 8 h.
• A venue with an 8 h interval, 0.01%: stays 0.01%.

Keep the annual scale in mind too: 0.01% per 8 h is three accruals a day, 0.03% a day, about 10.95% a year of the position's value. A “normal” rate looks negligible, but over a month of holding it is noticeable, and an extreme one (0.1% per 8 h) is 0.3% a day, over 100% a year.

Where rates live in EVA and how they differ:

• “Market funding” — a panel line: average funding over the venues where we have a rate (Bybit, Hyperliquid, Bitget, Gate), weighted by each venue's open interest and normalised to 8 hours. Next to it — the number of venues in the average. Positive — longs pay, negative — shorts; the line turns yellow from |0.05%| and up. Binance and OKX are not in this average.
• “Cost of leverage” — a band under the chart: funding and basis from one exchange, Bybit. Here the rate is taken as is, without reading the interval. So “how much per day” cannot be derived from this band: multiplying by three only works if you know the interval is 8 hours, and pairs also have 4 and 1 hour intervals.
• “Overview” → “Funding by exchange” — a table across all six venues, rates normalised to 8 hours (more in the fourth lesson).

The payment is computed from the position value, not from the margin:

payment = position value × rate per interval.

Position value is coins × price (venues use their fair, mark price for this, but the current price is fine for an estimate). This is where leverage makes funding noticeable. Illustrative example: margin $1,000, 10× leverage — the position is worth $10,000. At +0.01% per 8 h the long pays $10,000 × 0.0001 = $1 per accrual, $3 a day. Trivial in dollars, but relative to margin it is 0.1% per accrual and 0.3% a day, about 9% of margin a month.

The mistake nearly everyone makes: computing from margin. $1,000 × 0.01% = $0.10 — ten times less than the real payment. The higher the leverage, the bigger the miss.

A handy formula for the share of margin: payment in % of margin = rate × leverage. At 0.1% per 8 h (extreme but real) and 20× leverage, one payment is 2% of margin, three a day — 6% of margin a day. After a week of holding, funding would have eaten about 42% of margin even if the price stood still. So high funding on the side you hold is a direct cost that belongs in the trade plan just like the entry and exit fees.

The flip side: at a negative rate of −0.05% per 8 h a $20,000 short pays $10 per accrual, and a $20,000 long receives those $10. Receiving funding is pleasant but not free income: the position still carries the full price risk, and a single 2% move ($400) is worth more than a week of such payments (21 × $10 = $210).

The “Overview” → “Funding by exchange” tab is a table: coin × venue (Binance, Bybit, OKX, Bitget, Gate, Hyperliquid). Each cell holds the rate normalised to 8 hours; the cell colour shows how far the venue's rate is above or below the average across venues. Under the venue name is its next accrual; a coin with a different time shows it right in the cell; an asterisk means the time was computed from the interval because the exchange did not send it. A rate older than half an hour is marked “stale” and excluded from the average. Columns sort on click, and clicking a row opens the coin's chart.

Rates of different exchanges for the same coin sometimes diverge noticeably. Example: the average is +0.01%, while one venue shows +0.06% — a deviation of +0.05 percentage points. It is a fact that the long skew is stronger on that venue: more people want leveraged longs there, or fewer shorts stand against them.

The temptation is “short where they pay more, buy where they pay less, and pocket the difference”. That is not a ready trade. Venues have different fees, intervals and accrual times, the next interval's rate may change, and two positions on two exchanges are two margins and two places where something can go wrong. A 0.05 pp difference per 8 hours on a $10,000 position is $5 per accrual; a single round trip of entries and exits on two exchanges can easily cost more.

Funding is best read together with open interest from the previous section. A high rate on small OI means less than an average one on large OI: in the first case a handful of positions hold the skew, in the second a lot of money does. A high rate with rising OI — new positions open despite costly holding; a high rate with falling OI — the crowded side is already leaving. Both pictures describe what is happening now and what it costs participants, but neither says how it ends. The downside of leverage — forced closing of positions when margin runs short — is the next section.