1. Basis is a single subtraction.
Basis = contract price − spot price. Same coin, same moment: take the price on the contract side, subtract the price on the spot side, and what's left is the basis. Positive means the contract is more expensive than spot; negative means it's cheaper. That is the whole calculation. There is no second step.
Here it is with real numbers. At 15:49 UTC on September 22, 2026, OKX public market data showed BTC-USDT spot at 86,501.6 and the BTC-USDT-SWAP perpetual at 86,440. The funding rate on that perpetual at the same moment was 0.00606%.
| Measure | Value at that moment |
|---|---|
| Spot, BTC-USDT | 86,501.6 USDT |
| Perpetual, BTC-USDT-SWAP | 86,440 USDT |
| Basis (perp − spot) | −61.6 USDT |
| Basis rate (basis ÷ spot price) | ≈ −0.071% |
| Funding rate at the same moment | +0.00606% |
Written out: 86,440 − 86,501.6 = −61.6, so the perp sat 61.6 USDT below spot. That absolute figure can't be compared with anything on its own. Coins at different price levels, or BTC in different years, aren't on the same scale, so the usual next move is to divide by the spot price and turn it into a percentage: −61.6 ÷ 86,501.6 ≈ −0.071%. One subtraction and one division; plug in the two prices on your own screen and you get the reading for your own moment.
These numbers are a snapshot of a single second. Basis moves with every trade, so by the time you check, it almost certainly won't be −61.6, and the sign may well have flipped. Everything said about these numbers below describes the price structure at that moment on September 22.
2. Contango and backwardation: which side is in a hurry.
A contract trading above spot is in contango (at a premium); below spot, it's in backwardation (at a discount). Both terms come from traditional futures markets, and crypto kept the vocabulary. For more terms like these, see our Crypto indicator glossary.
| Shape | Price relationship | What the market is saying at that moment |
|---|---|---|
| Contango (premium) | Contract > spot | The contract side is bidding harder: someone is willing to pay above spot to get leveraged long exposure |
| Backwardation (discount) | Contract < spot | The contract side is selling harder, or leveraged longs are pulling out, and someone would rather sell below spot |
| Flat | The two nearly coincide | The gap between the two sides has been worn away, common in quiet markets |
A perpetual swap never expires, so its price has no natural point to converge on. What pulls it back toward spot is the funding mechanism: while the perp stays above spot, longs pay, holding costs rise and buying thins out, and the same works in reverse. That's why perp basis spends most of its time swinging in a narrow band around 0, and the −0.071% from September 22 is exactly that kind of near-zero reading. Whether it's high or low within this pair's own distribution is something you can only see by pulling the basis rate over a stretch of time. A single snapshot can't settle it.
Dated (quarterly) futures, which do expire, run on different logic. Their basis converges toward 0 as delivery approaches, because at expiry they settle against the index price (most exchanges use an average of the index over a window before expiry; the details are in each exchange's contract rules), so the contract price has to meet the index in the end. The basis moving toward 0, in itself, is forced by the settlement rules and says nothing about sentiment cooling; to read sentiment from dated futures you look at the annualized level, not at how far the basis still is from 0. For the same reason, a contract three months from delivery and one three days out naturally carry basis on different scales, and comparing the two tells you nothing. Dated-futures basis is also usually annualized by the days remaining (basis rate × 365 ÷ days to expiry). A perpetual has no expiry, so that conversion doesn't apply to it.
3. Negative basis, positive funding.
Take the snapshot again: basis −0.071%, the perp cheaper; funding +0.00606%, positive, so longs pay shorts at settlement. One reads as if the contract side is cool, the other as if longs are paying up. Intuitively, only one of them should be true.
Both are true, because they measure different stretches of time.
Basis is two prices subtracted at the instant you look at the screen; its resolution is one second. Funding doesn't take any single second. It takes the perp's premium over the spot index across a period, time-weights it into an average, and applies that at a settlement point. The snapshot in hand is one second inside that window, and in that second the perp happened to be 61.6 cheaper. An average over a period being positive while one second inside it is negative involves no mathematical conflict at all.
Go one step further: a positive rate doesn't even prove the perp was pricier than spot on average across the window. The formula contains more than the premium term — there is an interest term, and a clamp on top. Take OKX's version: the interest rate is fixed at 0.01%, so on 8-hour settlement an average premium of −0.01% leaves interest minus premium at 0.02%, inside the ±0.05% clamp, and the rate still comes out at +0.01%. A positive rate tells you that premium plus interest, after the clamps, came out positive; on its own it can't tell you the average basis was positive. How long the sampling window is, how it's weighted, how wide the caps are: each exchange sets this out in its own contract rules, the methods differ, and you check against whichever exchange you actually use. Settlement timing and what positive and negative rates mean are covered in more depth in Funding Rate Guide: Cost, Crowding and Carry.
This misreading is everywhere among retail traders: see a positive funding rate, assume the perp is at a premium right now, then conclude that "the contract side is chasing longs." The mistake is that middle step. The only way to know whether it's contango or backwardation right now is to pull up the two prices and subtract, and funding can't stand in for that subtraction. Five crypto indicators people misread first covers a few more leaps of this kind.
4. When this number misleads you.
The calculation itself is clean. Where basis misleads is in which two numbers you subtract, and when.
The two sides may not be the same kind of price. The contract side offers at least three numbers: last traded price, mark price and index price. Mark price has been smoothed; index price is a composite of several spot sources. Subtract spot's last price from the contract's last price and the gap carries noise such as order book depth and whether the final trade hit the bid or lifted the offer; switch to mark price and part of the gap has already been smoothed away. Either method works, but you need to know which one you're using, and not calculate it one way today and another way tomorrow and then compare the two.
At this size it's just noise. −0.071% on BTC comes to sixty-odd USDT. Taker fees at the major exchanges generally run around 0.05%–0.1%, and a single leg plus slippage can easily swallow a gap that small; check the fee page in your account to see where your own tier sits. Basis this shallow can't support any structural conclusion. It only means buyers and sellers weren't lined up in that second. Do not read a story into wiggles in the third decimal place; that is over-interpreting this indicator.
Subtract across platforms and you get a platform gap. Exchange A's spot price minus Exchange B's perp price carries both venues' liquidity, fee levels and user base. That number has its uses (cross-exchange spreads are a business in their own right), but it doesn't mean "the market is in backwardation." To describe the market, take both sides from the same exchange at the very least, or use multi-exchange composite indices on both sides.
Dated-futures convergence gets mistaken for sentiment. Basis shrinking toward 0 as delivery nears is forced by the settlement rules. A September contract's basis narrowing in the last week of September is not, by itself, evidence that bullish enthusiasm is fading; for sentiment over that stretch, check whether the annualized level fell as well. Plotting the basis of contracts with different times to expiry on one chart and comparing their heights is comparing apples to oranges.
In extreme moves it briefly breaks down. During the few minutes of a sharp drop, matching, liquidations and system load on the spot and contract sides don't move at the same pace. The gap can be stretched to levels you never normally see, then snap back within minutes. Recording the basis in those minutes is worthwhile, because it shows which side liquidity broke on, but treating it as a normal reading will lead you astray. Moments like these are better read alongside open interest; Open Interest Guide: The Leverage Still on the Table covers how to read it.
5. What it can and can't do for you.
Basis answers a descriptive question: right now, is the contract side bidding more aggressively or more cautiously than spot? It can't tell you direction. Backwardation doesn't mean "it's about to rise," and contango doesn't mean "time to get out." Both shapes have turned up plenty of times in rallies and in sell-offs, and a basis reading on its own has no predictive power.
Read it against its own distribution, not as an absolute value. The normal range for the same pair shifts between market phases, and one range certainly can't be carried across different coins. Chart the basis rate for the same pair over a recent stretch and see where today's number falls within its own distribution; that tells you more than staring at −0.071% itself.
Pair it with open interest to see who is pushing. Contango with open interest rising and contango with open interest falling are two different markets. The first looks more like new leveraged longs coming in; the second looks more like shorts closing out and pushing the price up. Basis only gives you the spread. The push has to come from the open interest line.
It is what cash-and-carry is priced on. Buy spot and short an equal amount of the perp at the same time, and the two legs offset each other's moves; what you earn is the basis reverting, plus whatever funding you collect while the rate is positive. The structure is real, but whether it pays only shows once you subtract these: four fees, for opening and closing both the spot and the contract leg; a funding rate that can turn negative at any time, after which you're the one paying; margin on the short leg, which has to be kept topped up or the short gets liquidated, and once it is, the spot leg left behind is naked one-sided exposure; plus the risk of holding funds on an exchange at all. When the basis rate is down in the third decimal place, those costs added together are often larger than it is. Work out the costs first, then decide whether the structure is worth opening.
Everything this number needs is on an exchange's public market data pages: subtract spot from the perp price, then divide by spot. Check it a few times a day, and within a week or two you'll have a feel for the normal range of the coin you follow.
6. FAQ.
Is there any arbitrage left when basis is zero?
Without basis reversion, all that's left to earn is funding. The fees are paid once, across the four legs to open and close, while funding is booked every settlement period, so what matters is how long you hold: settlement periods to break even ≈ sum of the four fee rates ÷ funding rate per period. Taking the 0.00606% from the snapshot in this article, and assuming the four fees add up to 0.2% (purely as an example), funding would have to stay at that level for about 33 settlement periods in a row just to cover the fees; if it turns negative along the way, the numbers have to be redone. Margin on the short side isn't a cost. It's capital that's tied up, and it's where the liquidation risk sits.
How does a change in basis affect a perp long I'm holding?
Work back from the definition: perp price = spot price + basis. So the price a perp long follows is spot's move with the change in basis layered on top; if spot doesn't move but basis narrows from positive to negative, the perp price still goes down. When funding is positive, longs also pay, which is a separate holding cost. Which price your PnL and liquidation are calculated on is set out in your exchange's contract rules.
Can I calculate basis the same way for altcoin perps?
The calculation is identical: the same exchange's perp price minus spot, divided by spot. The difference is in how you read it. Last traded prices already carry noise from order book depth, and the thinner a coin's order book, the larger that noise is likely to be, though that last point is inference rather than a measured result. Normal ranges don't transfer between coins either, so measuring a small coin against BTC's scale means nothing. Build up a stretch of its own basis rate first, then talk about high or low.