Bitcoin Futures Basis Holds Near 5% After the Selloff: Reading Deribit’s Term Structure

Bitcoin’s dated futures stayed in contango after the selloff, with Deribit basis near 5%. Read the curve, perpetual funding, open interest and limits.

When bitcoin falls hard, one place traders look for signs of capitulation is the futures curve. If leveraged buyers are forced out and nobody wants to pay up for future delivery, dated futures can slide to a discount to spot, a condition called backwardation. That did not happen during the selloff of November 13 to 15, 2026. On Deribit, every bitcoin future expiring from late November 2026 through September 2027 traded above the spot index at retrieval on November 17, and the annualized premium, or basis, rose steadily with maturity from about 4.9 percent to about 5.8 percent.

The perpetual swap, by contrast, traded almost exactly at the index, and funding rates on the largest perpetual markets hovered around zero. This article explains how the basis is measured, sets out the current term structure, compares it with the five weeks before the selloff, and discusses what the gap between a near-zero perpetual premium and a 5 percent dated premium does and does not imply.

Futures and index data come from Deribit’s public application programming interface; perpetual funding, basis and open interest from the public endpoints of Binance and OKX. These venues are named only as data sources. The blog does not compare their products, fees or reliability.

For the contract mechanics behind these comparisons, see the guide to maker fees, taker fees and funding rates and how perpetual-futures funding and liquidations work.

Key takeaways

  • At retrieval, with Deribit’s bitcoin index at $64,247.24, the December 25, 2026, future was marked at $64,582.54, a premium of 0.52 percent, or about 5.0 percent annualized over the 38 days to expiry (calculation).
  • The curve was in contango throughout: from about 4.9 percent annualized for November 27 to about 5.8 percent for September 24, 2027 (calculations). Only the one-day November 18 contract was marked marginally below the index.
  • A daily reconstruction of the December basis from hourly futures prices and Deribit’s daily delivery prices averaged about 5.6 percent annualized from October 10 to November 12 and about 5.2 percent from November 13 to 17 (calculations). The selloff dented the basis for one day, on November 13, but did not invert it.
  • Deribit’s BTC perpetual was marked 0.002 percent below the index, and Binance’s and OKX’s BTC perpetual funding rates averaged close to zero per eight hours from November 15 onward, against a long-run interest component of 0.01 percent per eight hours.
  • Open interest in perpetuals fell during the decline: Binance’s BTCUSDT contract went from 101,152 BTC on November 13 to 92,553 BTC on November 17, a drop of about 8.5 percent in coin terms (calculation).
  • Ether’s curve was flatter: about 3.7 percent annualized for December 25 and 4.3 percent for March 26, 2027 (calculations).

What the basis measures

A dated future is a contract to buy or sell bitcoin, or in Deribit’s case to settle in cash against the bitcoin index, on a fixed date. Its price differs from the spot price because holding bitcoin today and holding a promise of bitcoin later are not the same thing. A buyer who wants exposure for three months can either buy bitcoin now, tying up the full cash amount, or buy a future and keep most of the cash, earning interest on it in the meantime. In an efficient market the future therefore trades above spot by roughly the cost of financing, plus or minus a premium that reflects how keen buyers and sellers of leverage are.

The basis is that difference, expressed as a percentage of the spot price. Because a 0.5 percent premium means very different things for a contract expiring next week and one expiring next year, the basis is usually annualized: the raw percentage is divided by the number of days to expiry and multiplied by 365. A simple annualization is used throughout this article; compounding would change the figures only slightly at these levels.

The annualized basis is what a trader would earn, before costs and with no change in the premium, by buying bitcoin in the spot market and simultaneously selling the future, then holding both until expiry. That position, called a cash-and-carry trade, has no exposure to the bitcoin price; it captures only the premium. When the annualized basis is high, it attracts arbitrage capital, and the selling of futures by that capital tends to push the premium back down. When it is low or negative, the arbitrage runs the other way. The level of the basis is therefore a summary of the balance between demand for leveraged long exposure and the supply of capital willing to provide it.

The curve at retrieval

ContractMark price ($)Days to expiryPremium to indexAnnualized basisOpen interest ($ million)
Perpetual64,245.88—−0.002%—689.7
Nov 1864,242.020.96−0.008%−3.1%1.1
Nov 2064,267.532.960.032%3.9%30.2
Nov 2764,333.599.960.134%4.9%69.5
Dec 464,394.6016.960.229%4.9%6.2
Dec 2564,582.5437.960.522%5.0%300.6
Jan 29, 202764,923.5572.961.053%5.3%9.0
Mar 26, 202765,489.81128.961.934%5.5%111.6
Jun 25, 202766,425.45219.963.390%5.6%72.7
Sep 24, 202767,399.35310.964.906%5.8%16.6

Mark prices and open interest are from Deribit’s book summary; Deribit’s futures open interest is quoted in U.S. dollars of contract value. Days to expiry count from the retrieval to each contract’s official expiry time, which Deribit’s instrument list gives as 08:00 UTC. Premiums and annualized figures are calculations against the index price of $64,247.24 returned at the same retrieval.

The shape is a gently upward-sloping curve. The basis rises by about 0.9 percentage points between the nearest monthly contract and the furthest quarterly, a pattern that is typical when the market is calm about financing and modestly willing to pay for longer-dated exposure. There is no kink around the holiday-shortened week of November 26 and 27, and no sign in the near contracts that traders were prepared to sell bitcoin for future delivery at a discount. The negative figure for the November 18 contract reflects a premium of about $5 on a contract with less than a day to run, which is within the range of normal pricing noise and carries no information once annualized.

Open interest is concentrated in two places: the perpetual, with $689.7 million, and the December 25 quarterly, with $300.6 million. Dated futures together account for $618.8 million, or about 47.3 percent of Deribit’s bitcoin futures open interest (calculation). Trading is far more concentrated: over the 24 hours covered by the retrieval, the perpetual traded about $1.74 billion, against about $48.9 million in the December contract and $13.4 million in March.

Did the selloff move the basis?

A single snapshot cannot say whether the basis is high or low relative to recent weeks. To build a history the blog compared Deribit’s hourly futures prices with the exchange’s published daily delivery prices for the bitcoin index, using the hourly close that ends at the delivery time on each day. The results for the December 25 and March 26 contracts are summarized below.

PeriodDec 25 annualized basis, averageDec 25 rangeMar 26 annualized basis, averageMar 26 range
Oct 10 – Nov 12 (34 days)5.59%3.66% – 7.63%5.47%4.73% – 6.42%
Nov 13 – Nov 17 (5 days)5.20%3.07% – 6.17%5.25%4.83% – 5.81%

All values are calculations. The daily readings for the most recent days were, for December: 3.07 percent on November 13, 5.54 percent on November 14, 5.19 percent on November 15, 6.17 percent on November 16 and 6.02 percent on November 17. For March they were 4.83, 5.00, 4.84, 5.81 and 5.76 percent.

The first observation is that the basis did not collapse. November 13, the day bitcoin’s Coinbase daily candle fell 4.4 percent and broke below $70,000, produced the lowest December reading in the series, but even that was positive, and the reading recovered the next day. The March contract barely moved. Over the whole period the December basis drifted down only from the high fives to the low fives on average, a change smaller than its normal day-to-day variation.

The second observation concerns method. The reconstruction compares an hourly futures close with a delivery price that Deribit calculates as an average of the index over a short window, so the two are not taken at exactly the same instant. The same comparison applied to the perpetual, which should trade very close to the index, produces daily differences between −0.27 percent and +0.26 percent during November (calculation). For a contract 38 days from expiry, a measurement error of 0.2 percent of price translates into about 2 percentage points of annualized basis, which explains why the reconstructed December figure for November 17 (6.02 percent) differs from the snapshot figure (5.02 percent).

Single-day readings in the history are therefore noisy; averages over several days, and the longer March contract, are more reliable. The conclusion that the basis stayed positive and broadly stable survives that noise. A claim that it rose or fell by a specific fraction of a point on a specific day would not.

The perpetual tells a different story

Perpetual swaps have no expiry, so they cannot carry a term premium in the same way. Instead, exchanges use a funding payment, exchanged between longs and shorts at fixed intervals, to keep the perpetual price close to the index. On Binance and OKX the BTC-USDT perpetuals settle funding every eight hours. A positive rate means longs pay shorts; a negative rate means shorts pay longs.

DayBinance BTCUSDT mean funding (% per 8h)OKX BTC-USDT-SWAP mean funding (% per 8h)
Nov 30.01000.0100
Nov 60.00820.0032
Nov 90.00860.0058
Nov 120.00620.0076
Nov 130.00510.0025
Nov 140.00480.0025
Nov 150.0022−0.0009
Nov 160.00190.0015
Nov 17 (2 of 3 settlements)−0.00030.0030

Daily means are calculations from each venue’s settlement history, grouped by the venue’s calendar day. Binance’s premium-index endpoint lists an interest component of 0.01 percent per eight hours for the contract, and both venues printed exactly that rate on several days in early November, which is the level funding settles at when the perpetual trades at a small, steady premium. A rate of 0.01 percent per eight hours compounds to roughly 11 percent a year if paid at every settlement (calculation, simple annualization over about 1,095 settlements). The rates since November 15, near zero, correspond to an annualized cost for longs of between about −1.0 and +3.3 percent (calculations).

Binance’s daily perpetual basis series, which compares the contract price with its index, was slightly negative on every November day in the retrieved file except November 9 (+0.02 percent), and between −0.03 and −0.06 percent on each day from November 10 to 16. The latest premium-index reading showed the BTCUSDT mark price at $64,275.09 against an index of $64,320.33 and a last funding rate of −0.0020 percent. On Deribit the perpetual was 0.002 percent below its index.

The result is a split. In the perpetual market, where most leveraged trading happens, the cost of holding long exposure has fallen to about zero. In the dated market, a buyer of the December contract is still paying a premium equivalent to about 5 percent a year.

Why the two can diverge

If a long position can be carried in the perpetual at close to zero cost while the December future charges 5 percent annualized, an arbitrageur could in principle buy the perpetual, sell the December future and collect the difference. The fact that the gap persists says something about who uses each market and what each trade costs.

First, the perpetual and the dated future serve different users. Short-term traders favor perpetuals because they never have to roll. Investors who want exposure over weeks or months, including some that may be unable or unwilling to hold perpetual contracts, tend to use dated futures. A selloff that flushes short-term speculative longs can push perpetual funding to zero without much affecting the demand of longer-horizon users of dated contracts.

Second, the arbitrage is not riskless. Funding can rise sharply in a rally, turning a cheap carry into an expensive one in a matter of days, while the dated future’s premium is locked in only if the position is held to expiry. A trader who sells the December future and buys the perpetual is exposed to that funding risk for five weeks.

Third, the dated basis reflects the cost of dollar funding. A cash-and-carry trade requires cash to buy bitcoin or a perpetual margin balance, and that cash has an opportunity cost. A 5 percent annualized premium on bitcoin futures should be compared with what the same dollars could earn elsewhere. Readers can check current short-term dollar rates, such as Treasury bill yields, from official sources; the blog has not included a specific rate here because it is outside the data retrieved for this article.

Open interest: who left during the decline

The perpetual open interest series shows that positions were closed during the selloff rather than added.

DateBinance BTCUSDT open interest (BTC)Binance BTCUSDT open interest ($ billion)OKX BTC contracts open interest ($ billion)
Nov 1098,5797.4263.859
Nov 1199,5457.3433.847
Nov 1296,6267.1483.866
Nov 13101,1527.2263.459
Nov 1498,3896.7173.244
Nov 1597,9516.2813.032
Nov 1693,5246.1092.986
Nov 1792,5536.003—

Binance figures are daily records from its open-interest history endpoint; the latest is incomplete. OKX figures are from its contracts open-interest endpoint, which aggregates the venue’s BTC contracts and is quoted only in dollars. Binance’s coin-denominated open interest peaked at 101,152 BTC on November 13 and fell to 92,553 BTC in the latest record, a decline of about 8.5 percent (calculation). In dollar terms the drop from November 13 is about 16.9 percent (calculation), the difference being the price decline. OKX’s dollar open interest fell about 22.8 percent from November 12 to November 16 (calculation).

This pattern helps explain the funding data. Closing longs reduces the demand that keeps funding positive; if shorts had been added aggressively, coin-denominated open interest would have risen. Binance’s taker buy/sell volume ratio, which compares aggressive buy volume with aggressive sell volume on the contract, was below 1 on each day from November 12 to November 16 (0.912, 0.898, 0.915, 0.998 and 0.952), consistent with more aggressive selling than buying but close to balance by November 15.

The dated futures open interest on Deribit cannot be compared over time from the snapshot alone. The December contract’s $300.6 million is large relative to the other dated contracts, but the retrieved data do not show how much of it was opened before or after the selloff.

Ether’s curve, for comparison

Deribit’s ether futures curve was also in contango at retrieval, but flatter and lower than bitcoin’s. With the ether index at $1,818.04, the November 27 contract was marked at $1,819.82 (about 3.6 percent annualized), December 25 at $1,824.96 (about 3.7 percent), March 26, 2027, at $1,845.33 (about 4.3 percent) and September 24, 2027, at $1,892.86 (about 4.8 percent); all annualized figures are calculations. The reconstructed December basis for ether averaged about 4.4 percent from October 10 to November 12 and about 4.6 percent from November 13 to 17 (calculations), again with considerable day-to-day noise.

Ether’s perpetual funding was weaker than bitcoin’s. OKX’s ETH-USDT perpetual had negative daily average funding on every day from November 13 to November 17, between about −0.001 and −0.007 percent per eight hours (calculations), while Binance’s ETHUSDT averaged roughly zero from November 15. The lower ether basis and weaker funding fit the fact that ether fell further than bitcoin during the selloff: Coinbase’s ETH-USD close declined about 24.9 percent from November 4 to November 16, against about 19.4 percent for BTC-USD (calculations).

What a healthy-looking curve does not prove

It would be easy to treat a positive, upward-sloping curve as evidence that the selloff is over. The data do not support that inference. Contango is the normal state of bitcoin futures; the reconstructed December and March readings were positive on every day of the retrieved period, through both up and down days. A curve inverts only under unusual stress, when spot sellers overwhelm futures demand or when funding markets seize up. Its absence shows that the November decline was not that kind of event. It does not show that the price will rise.

Nor does the 5 percent basis mean that “institutions are bullish.” The basis compensates whoever supplies leverage; it rises when demand for leveraged long exposure exceeds the supply of arbitrage capital, and falls when the reverse is true. A stable basis during a sharp decline suggests that the balance between those two groups did not change much in the dated market. The speculative excess, to the extent there was one, was in the perpetual market, and that is where it was cleared.

What to watch in the curve

Perpetual funding returning to its baseline

If funding on the large perpetuals climbs back toward 0.01 percent per eight hours, the gap with the dated basis will close from below. That would indicate renewed demand for short-term leveraged longs.

A flattening or inversion at the front

If the November 27 and December 4 contracts move to a discount to the index, that would be a sign of stress not seen in this episode. Readers can check it from the same book-summary endpoint used here.

Roll activity into December

As November 27 approaches, positions in that contract will either expire or roll. Growth in December or March open interest at a stable premium would suggest that longer-horizon users are maintaining exposure.

A practical reading

For a holder of spot bitcoin, the basis is the price at which the market would buy that bitcoin for delivery later; at about 5 percent annualized, the market is not offering a large reward for selling forward, but it is not demanding a discount either. For a leveraged trader using perpetuals, the near-zero funding means the cost of carry is unusually low for the moment, which can change quickly. Neither observation is a signal to trade. Both are measurements of conditions that anyone can recheck from the public endpoints listed below.

Frequently asked questions

What is the bitcoin futures basis right now?

At retrieval on November 17, 2026, Deribit’s December 25 future carried a premium of 0.52 percent to the index, about 5.0 percent annualized (calculation). Longer contracts were higher, up to about 5.8 percent for September 2027.

Did the bitcoin futures curve go into backwardation during the selloff?

No. Every dated contract with more than a day to run was above the index at retrieval, and the reconstructed daily December basis stayed positive throughout November 13 to 17.

Why is perpetual funding near zero when the futures basis is 5 percent?

The two markets serve different users and the arbitrage between them carries funding risk. The selloff appears to have cleared leveraged longs from perpetuals without much changing demand in dated futures.

How is the annualized basis calculated?

The future’s premium to the index, as a percentage, is divided by the days to expiry and multiplied by 365. This article uses that simple method.

Sources and research limits

Bitcoin and ether futures marks, open interest and volume are from Deribit’s public bitcoin futures book summary and ether futures book summary; index prices from its index-price endpoint; expiry times from its instrument list; the basis history from its hourly chart data and delivery prices. Funding rates are from Binance’s funding-rate history and OKX’s funding-rate history. Binance’s perpetual basis, premium index, open interest and taker ratio are from its basis, premium-index, open-interest history and taker volume endpoints. OKX open interest is from its contracts open-interest endpoint. Spot price changes use Coinbase’s public BTC-USD daily candles.

Limits: CME bitcoin futures, which are a large part of the regulated futures market, were not retrieved, so this article describes Deribit’s curve only. The basis history is reconstructed with a timing mismatch described above and should be read as approximate. All premiums, annualized rates, averages and percentage changes are the blog’s calculations. This article is for information only and is not investment advice. Futures and perpetual swaps are leveraged products that can cause losses larger than the initial margin.

Disclaimer: This content was generated with the assistance of artificial intelligence (AI) and has been reviewed by our editorial team. It is intended for informational purposes only and should not be construed as financial, investment, or legal advice. Cryptocurrency investments involve significant risk.
KCEX BLOGKCEX BLOG
Previous 19 hours ago
Next 2026-10-25 02:27

Related Posts

SHARE
TOP