Open Interest

Bitcoin open interest is the total USD notional of futures contracts that are open — entered and not yet closed or settled — summed across venues each day and stacked, one band per venue, with the thin line on top as the aggregate. The y-axis runs logarithmically so the COVID-era book around $3B stays legible against today’s $49B, and a muted price line sits behind for cycle context.

The single distinction that the rest of this page leans on: open interest is a stock measured at an instant, not a flow measured over an interval. Volume counts every trade that crosses; OI counts only positions still held. The same dollar of OI can sit untouched for months while billions in volume turn over inside it, and a leveraged blow-off can print falling OI (positions closing) alongside exploding volume (those closures are themselves trades) on the same day. Confuse the two and the chart reads backwards.

Today the top three venues by notional are Binance, CME, MEXC, and the regulated CME leg holds 14.4% of the stack — a share that has risen sharply since the spot-ETF approvals of early 2024, and the reason a single OI number is no longer comparable across cycles without knowing its composition.

Two inputs flow in: each venue’s reported USD open interest at the daily close, and the deterministic Bitcoin circulating supply on that day. Provenance is on the data sources page; the per-venue normalisation and the OI/market-cap derivation are spelled out on methodology. Three computations sit behind the chart:

OI_total(t) = Σv OI_v(t) across the venues we cover, with everything outside the top five rolled into one Other bucket so the stack stays legible.
market_cap(t) = price(t) × circulating_supply(t), where supply is the closed-form value at block-height t — no pricing-feed dependency.
OI/mcap(t) = OI_total(t) / market_cap(t) — the leverage-vs-spot proxy that drives the regime classifier.

Two reconstruction notes that most hosts leave implicit. First, OI is reported in three equivalent forms across venues — contract count, USD notional, coin notional — and a coin-margined contract’s USD value moves with spot even when no contracts open or close. We pull USD notional wherever we can and accept a small mechanical price-sensitivity on the coin-margined slice. Second, the regime thresholds are OI/mcap > 2.5% for “Leveraged” and < 1.5% for “Deleveraged.” They are descriptive cuts on the post-2020 distribution, not bright lines — the content is the regime, not the exact threshold.

The instructive read pairs OI with price simultaneously. A single OI number tells you the size of the book; the OI move tells you positioning is being added or removed; only the direction of the matching price move tells you which side. The cleanest published version sits in the Glassnode leveraged-positioning note, which spells out the four cells: When price and OI rise together, traders are adding long exposure into an upward trend (long opens); A declining price with increasing OI suggests traders are opening short positions; A simultaneous drop in both price and OI indicates long positions are being closed; Rising prices alongside falling OI point to short positions being exited. Mapping each cycle move to one of those four cells is the core read this chart supports.

Seven dates since the start of our daily series sketch the cycle. Aggregate open interest compounded from the COVID flush near $3B to a post-ETF book above $50B; the path between them is the story this chart tells. Spot prices are nightly closes from our pipeline; OI cells come from the same daily snapshot powering the chart. The two November 2022 rows are the case study the next section unpacks.

The cleanest case study in the series sits across two rows of the table above. On 2022-11-08, the day before FTX paused withdrawals, aggregate OI sat near $13.1B. By 2022-11-30 — three weeks after the Chapter 11 filing — total OI had fallen to $10.6B. The instinctive read is “the market deleveraged twenty percent.” That is the wrong frame. The missing notional did not unwind — the venue carrying it disappeared. Positions on the FTX perpetual book were socialised into the bankruptcy estate; aggregator totals stepped down because one cell of the stack went to zero, not because every other venue’s holders closed out.

Kaiko’s Alameda Gap analysis put FTX’s peak derivatives market share near 15%; the share didn’t drift, it vanished in one weekend. Surviving venues absorbed the flow over the following months — the aggregate climbed back through the 2023 banking-crisis chop and into the spot-ETF approvals — but the November 2022 step is the cleanest reminder that aggregate open interest is not a neutral leverage gauge across venue regimes. Any cross-cycle comparison has to ask which venues were inside each total.

The composition of the stack changed permanently after the January 2024 spot-ETF approvals. Through the 2021 cycle the largest band in the stack was an offshore perpetual venue, and the regulated CME leg sat third or fourth. By November 2024 CME had crossed 218,000 BTC ($21.3B) of OI — the first time at that scale. Today the regulated leg holds 14.4% of total open interest, well above its low-single-digit pre-ETF share.

That flip matters for how you read the chart. A 2021-era $24B total was retail-led, dominated by unregulated perpetual leverage with weekly liquidation cascades. The post-ETF $49B is structurally different — a meaningful slice now sits inside cash-settled, calendar-expiry exposure that does not produce the same weekend wicks, much of it the short leg of basis trades hedged against ETF spot. The OI/mcap ratio can read broadly similar numbers cycle to cycle, but the volatility a given reading produces has fallen as the regulated band thickened. A read of open interest that ignores the composition shift is reporting last cycle’s indicator on this cycle’s book.

Open interest is one of the most-quoted derivatives numbers and one of the most mis-quoted, because three structural shifts change what the number means without changing the formula that produces it. None of these is a bug; each is a reason the same reading carries a different message in different years.

Venue discontinuity rewrites the total without anyone trading. November 2022 is the canonical case: aggregate OI fell from $13.1B to $10.6B chiefly because FTX’s book left the dataset, and reading the 8th against the 30th records a 19% “deleverage” that the Alameda Gap research shows conflates two unrelated things. The same hazard runs the other way: when an aggregator adds a venue, the total steps up with no new positioning. Whenever the stack has a composition step-change, treat the level as discontinuous across it.

The regulated-leg buildout broke the cross-cycle leverage read. The OI/mcap ratio was designed as a retail-leverage gauge, and through 2021 that is what it measured. After early 2024, a growing share of the book is CME basis exposure — long ETF spot, short CME futures — which is market-neutral, not directional leverage. A $60B book that is one-quarter cash-and-carry is far less flush-prone than a $60B book that is all perpetual longs, yet the ratio prints the same. Reading the “Leveraged” band the way it behaved in 2021 over-states fragility on today’s composition.

Coin-margined OI moves with spot on its own. A non-trivial slice of perpetual OI is coin-margined — quoted in BTC, displayed in USD after multiplying by current spot — so its USD reading falls in a drawdown even if no contracts close. During the worst of a liquidation cascade, coin-margined OI shrinks faster than the position count, exaggerating the visible “deleveraging.” The effect is small in calm markets and large precisely when you most want a clean read.

The defence against all three is the same: never read the aggregate level in isolation. Read the OI move against the simultaneous price move for the four-quadrant signal, weight the regulated-vs-perp composition before invoking the leverage regime, and treat any total that spans a venue change as two different series glued together.

The four-quadrant phase diagram is the page. Locate today on it — using the OI change and the price change together — and treat the cell as a regime, not a trade. Long-side exhaustion in the “long openings” cell with funding north of +0.05% per 8h is the canonical leverage-flush precursor; short openings into a falling tape with a crowded short book is the squeeze setup. Pair the read with funding rate for the price of crowded leverage and the liquidation heatmap for the geometry of forced flow.

The OI/mcap ratio is a coarse backdrop, not a daily input. Above 2.5% the chart reads “Leveraged,” the regime where wicks have historically been violent; below 1.5% it reads “Deleveraged,” where rallies grind. The bands move on the order of weeks, so daily readings are noise — and on this cycle’s book the regulated-leg caveat above means a “Leveraged” print is a weaker fragility signal than the same number was in 2021.