Bitcoin's annualized three-month futures basis has traded below the two-year U.S. Treasury yield every month since February 2026 — only the second such stretch on record. The carry trade that once returned 20%+ during the 2021 cycle now yields approximately 2–3%, well below the 3.8–4.2% available...
"When carry pays less than T-bills, a dollar in the trade earns less than it would sitting in government paper." — David Lawant, Head of Research, Anchorage Digital
Bitcoin's annualized three-month futures basis has traded below the two-year U.S. Treasury yield every month since February 2026 — only the second such stretch on record. The carry trade that once returned 20%+ during the 2021 cycle now yields approximately 2–3%, well below the 3.8–4.2% available in short-dated government debt. The result: hedge funds that entered Bitcoin ETFs primarily to harvest the basis premium are exiting, while financial advisors and bank wealth desks — buying for client allocation rather than arbitrage — are scaling in.
Q2 2026 13F filings quantify the rotation. Hedge funds shed 39% of their Bitcoin ETF exposure quarter-over-quarter in Q1, and continued trimming in Q2. Advisors now hold $17.4 billion in Bitcoin ETF positions, nearly double hedge fund managers' $9 billion. JPMorgan grew its IBIT stake to $356 million; UBS increased its position 230% year-over-year to $90 million; Morgan Stanley added 23% to reach 16.5 million IBIT shares. The composition of institutional Bitcoin ownership is shifting from short-duration arbitrageurs to longer-duration allocators, with implications for price dynamics, liquidity structure, and the economic value distribution within Bitcoin markets.
The Bitcoin basis trade is a market-neutral arbitrage strategy. A trader simultaneously buys spot Bitcoin (typically via an ETF such as BlackRock's IBIT) and sells Bitcoin futures at a premium, capturing the difference as yield. When futures trade at a persistent premium to spot — a condition known as contango — this premium can be annualized and expressed as a yield.
During 2021, the annualized three-month basis routinely exceeded 20%. By late 2024, it had compressed to approximately 10–12%. As of early August 2026, the annualized basis stood at roughly 2.08%, according to CoinDesk data. On a matched CME check conducted August 7, gross futures carry ranged from 5.69% (December contract) to 7.89% (August front-month), but after funding, margin costs, and execution fees, net yield to institutional traders fell below the 4.19% available on two-year Treasuries.
The compression reflects two forces. First, the proliferation of spot ETFs in January 2024 massively expanded the pool of arbitrageurs, compressing the premium through competition. Second, Bitcoin's price decline from approximately $106,000 in January 2026 to $58,000 by mid-year reduced speculative demand for leveraged long futures, narrowing the contango.
The consequence is mechanical. When a dollar deployed in the basis trade earns less than a dollar parked in government paper, the trade's risk-adjusted return turns negative. Capital exits.
The 13F filings tell the story.
Q1 2026: 13F filers collectively shed approximately 52,500 BTC in exposure, taking total professional holdings from 313,000 BTC to 261,000 BTC — a 17% decline. Hedge funds accounted for the largest share, cutting exposure by 31,400 BTC, a 39% quarter-over-quarter contraction and 42% year-over-year decline. Brokerages reduced holdings by 53% QoQ. According to CoinShares' Q1 2026 Bitcoin 13F Report, hedge funds and brokerages together accounted for 95% of the total exposure reduction.
The most striking individual move came from Brevan Howard, which cut its IBIT position from 36.7 million shares at peak to 5.5 million by year-end 2025 — an 86% reduction, slashing its stake from approximately $2.4 billion to $275 million. The firm was one of the largest basis-trade operators in the ETF market.
Q2 2026: Despite Bitcoin falling 14.2% during the quarter, institutional Bitcoin ETF holdings rose 7.5%, climbing from 498,389 to 535,723 BTC equivalent. The divergence is explained by composition: hedge funds continued trimming, but advisors and banks more than offset the reduction with new buying. The net effect was positive accumulation even as the largest pre-existing category of institutional holder — arbitrageurs — reduced exposure.
Perpetual futures funding rates turned negative on a 30-day average by Q1-end, making the basis trade unprofitable at scale and encouraging further unwinds. The structural incentive that drew hedge fund capital into Bitcoin ETFs in 2024 had, by mid-2026, largely evaporated.
As arbitrageurs departed, a different class of institutional holder expanded.
Financial Advisors now hold $17.4 billion in Bitcoin ETF positions, approximately 150,000 BTC equivalent, up 20% year-over-year. This category — registered investment advisors (RIAs), independent advisors, and wirehouses — buys on behalf of client portfolios, typically with multi-year time horizons and 1–5% portfolio allocations. Edelman Financial Engines, one of the largest U.S. RIA firms, reported a $34 million Bitcoin ETF position in its Q2 filing, exceeding its Amazon stake.
Banks held 15,100 BTC equivalent as of Q2, a four-fold increase year-over-year:
The distinction matters for market structure. Basis traders are price-insensitive on the spot side — they buy the ETF solely to hedge the futures short, and they sell when the trade unwinds. Advisors and banks, by contrast, are making allocation decisions based on portfolio construction rationale. Their holding periods are longer, their flows are less correlated with futures market dynamics, and their selling is driven by rebalancing rather than arbitrage unwind.
On August 10, 2026, CoinDesk reported that hedge funds trading Bitcoin futures on the CME had shifted from structural net short to net long positioning — described as the first such flip in years. The "leveraged funds" category in the CFTC Commitments of Traders report had maintained a persistent net short position throughout 2024 and 2025, a direct consequence of the basis trade: the futures short leg.
CryptoQuant CEO Ki Young Ju initially reported the flip but subsequently clarified that CFTC Total Reportables, not leveraged funds specifically, were modestly net long. The distinction is important: the "leveraged funds" category in the COT report includes, but is not limited to, hedge funds running basis strategies.
Regardless of the precise categorization, the directional shift is notable. With the basis at 2%, maintaining a futures short purely for carry makes no economic sense. Some funds appear to be re-deploying toward directional long exposure, while others are exiting futures markets entirely and reallocating to Treasuries.
This coincided with Bitcoin open interest falling to a five-month low of 587,600 BTC on August 25, down from 645,760 BTC on August 14. Total futures open interest across the market fell 3.08% quarter-over-quarter to $103.2 billion. The rally from $58,000 to $81,000 through August was driven more by short liquidations than by new leveraged longs, according to market data — a dynamic consistent with basis unwind rather than speculative accumulation.
With the basis trade compressed, institutional participants are exploring alternative approaches to generating yield on Bitcoin exposure.
Covered-call writing has attracted significant attention. Anchorage Digital's research unit, led by David Lawant, published a comprehensive study using more than 37,000 backtests across the Deribit implied-volatility surface from October 2021 to April 2026. A standard 20-delta, 30-day covered-call strategy generated a net yield of 5.5% on the underlying BTC position from April 2025 to April 2026, during which spot BTC fell 19.4%. The overlay offset approximately one-third of the drawdown, and portfolio annualized volatility fell from 40.6% to 35.0%.
However, full-cycle results were less favorable. Extended across the entire October 2021 to April 2026 period, the same strategy produced a negative yield of 0.5%. The core limitation: if Bitcoin rallies through the strike, upside participation is capped. Bitcoin options notional open interest grew roughly ten-fold over five years, briefly exceeding $100 billion at end-2025 before settling near $60 billion — larger than the entire BTC futures market.
Market-making has also emerged as a yield source. CoinDesk reported on August 28 that crypto market-making firms were generating returns from Bitcoin's August rally through non-directional strategies — capturing bid-ask spreads and liquidity provision — rather than taking explicit price bets.
The rotation from arbitrageurs to allocators has several structural consequences.
Reduced selling pressure during basis unwinds. As the proportion of ETF shares held for basis purposes declines, future episodes of futures contango compression should produce smaller forced selling in the spot market. The 2026 sell-off saw hedge funds dump 52,500 BTC in a single quarter; with their share of total holdings shrinking, the magnitude of future unwind-driven liquidations should diminish.
Lower futures volumes, tighter spot spreads. CME trading volume dropped from a peak of $1.47 trillion in February 2026 to approximately $880 billion in July. As basis traders exit, futures-spot volume ratios decline, but the remaining spot liquidity is increasingly provided by long-term holders, which tends to produce more stable order books.
Shift in fee revenue distribution. The basis trade generated substantial fee revenue for futures exchanges (CME), prime brokers, and clearing firms. As this volume migrates to spot ETFs and potentially covered-call overlays, the fee beneficiaries shift toward ETF issuers (collecting management fees) and options exchanges (Deribit, CME options). The economic value generated by institutional Bitcoin participation is being redistributed across the infrastructure stack.
Higher ETF asset stickiness. BlackRock's IBIT captured 81% of August's $853 million weekly inflow peak, with total August inflows exceeding $3 billion — the strongest month of 2026. With IBIT's holder base increasingly composed of advisors and banks rather than hedge funds, the product's AUM should exhibit lower volatility and higher retention rates.
The collapse of the Bitcoin basis trade is not a temporary dislocation. It reflects the structural maturation of a market that has attracted sufficient arbitrage capital to compress its own inefficiencies. The carry trade's 20%+ yields in 2021 existed because the market was inefficient; 2–3% yields in 2026 exist because it is not.
The consequent ownership rotation — from hedge funds that held ETFs solely as arbitrage legs to advisors and banks that hold them as portfolio allocations — has meaningful implications for how economic value flows through the Bitcoin ecosystem. Fee revenue shifts from futures exchanges to ETF issuers and options platforms. Selling pressure becomes driven by portfolio rebalancing rather than arbitrage unwind. Holder duration extends.
For Bitcoin's price dynamics, the near-term implication is that the basis-driven selling overhang is largely cleared. The 52,500 BTC liquidated by hedge funds in Q1 has been absorbed by buyers with longer holding periods. Whether this creates a more stable price floor or simply a different set of risks depends on the duration and conviction of the new buyer base — and on whether sub-Treasury yields persist long enough to keep arbitrageurs on the sidelines permanently.