A reader's guide to derivatives data

Digital asset derivatives publish more about positioning than almost any other market. Open interest, funding and liquidation data are free, continuous and widely misread. A practical guide to what each measures and where each stops being informative.

Conventional futures markets report positioning weekly, with a lag, in aggregated categories. Digital asset perpetual markets publish equivalent information continuously and at venue level. This is a genuine analytical advantage, and it is routinely squandered by treating positioning data as directional forecast.

Open interest: how much is at stake

Open interest is the total notional value of outstanding derivative contracts. It measures the size of the market's collective bet, not its direction — every contract has a long and a short.

What makes it informative is its change relative to price. Rising open interest with rising price indicates new long positions entering, which is a market building exposure. Rising open interest with falling price indicates new shorts entering. Falling open interest in either direction indicates positions closing, which is risk leaving regardless of what price does.

Total open interest stood at $84 billion in January 2026 and $53.15 billion in early September, the latter up 9.7% over the preceding thirty days.^1,2 The composition is unusual: more than 96% sits in perpetual contracts rather than dated futures, a ratio without parallel in conventional derivatives markets.¹

Funding: what one side pays the other

Perpetual contracts have no expiry, so funding payments keep them tethered to spot. Every eight hours, one side pays the other at a rate proportional to the gap between perpetual and index price.

Funding measures crowding, not direction. Persistently positive funding means leveraged longs are paying to hold exposure available more cheaply in spot — a statement about impatience, not about being wrong. The reason it matters is what it implies about the exit: positions opened in the same window at similar prices share maintenance margin levels, and a large cluster of them moves together when triggered.

Two readings deserve attention. The 46-day negative stretch between roughly 1 March and 15 April 2026 was the longest since November 2022, with shorts paying 0.5–1% weekly while open interest rose about 12% — new shorts entering and paying heavily to do so.³ And the September reading of roughly 3.1% annualised describes a market with no strong positioning in either direction.

Liquidations: the mechanical layer

Liquidations occur when a leveraged position's margin falls below maintenance requirements and the venue closes it automatically. They are mechanical events, not decisions, which is precisely what makes them informative — they reveal where leverage was actually sitting rather than where participants said it was.

The 2026 record includes a $3.2 billion single-day cascade, the largest recorded in the asset class, and $1.86 billion in twenty-four hours on 3 June.⁴ The composition also matters: by early September short liquidations accounted for 65.5% of futures liquidations, the mechanical footprint of the spring's crowded short positioning being unwound during the August rally.²

Liquidation clusters can be estimated ahead of time from open interest, leverage and entry prices. In April 2026 the short cluster sat between $80,000 and $85,000; bitcoin traded above $80,000 in late August.³ Reading a cluster is not forecasting that price reaches it — it is establishing that if price does reach it, the move through will be amplified rather than damped.

Positioning data describes the shape of the distribution of outcomes. It does not describe which outcome occurs, and the gap between those two readings is where most errors with this data originate.

Long/short ratios and basis

Long/short account ratios offer a crude sentiment read. January 2026 figures showed roughly 1.5× for bitcoin, 1.8× for ether and 2.7× for Solana, the last normalised from peaks near 3.3×.¹ Values near balance indicate room for positioning to build in either direction; extremes indicate crowding.

Basis — the spread between futures and spot — describes the term structure. A flat curve across 30, 90 and 180 days, as observed in early 2026, indicates balanced expectations without strong directional conviction. Contango with an upward slope, as observed in ether, indicates a market willing to pay for forward exposure.¹

Where each measure fails

Every one of these degrades under specific conditions, and they degrade together.

During genuine liquidity events, funding stops describing sentiment and starts describing plumbing — suspended withdrawals, a venue's price detached from the market, one side unable to arbitrage. The number still prints and no longer means what it usually means.

In thin markets, a single large account can set the funding rate for everyone, and open interest changes reflect one participant rather than a market.

Over long horizons, all of it degrades. These are reads on days to weeks of positioning, not on quarters. Any framework treating funding as a slow structural indicator is using it far outside its useful range.

Across fragmented venues, single-venue readings mislead. Funding spreads between venues reached 6.37% for one asset in early 2026.¹ One venue running hot while others sit flat is usually a venue story — a concentrated book, a market maker stepping back. Elevated readings in the same direction everywhere at once is a market story. Confusing the first for the second is the most common error made with this data.

Using it well

The practical summary: open interest tells you how much is at stake and whether risk is entering or leaving. Funding tells you which side is paying and therefore how uncomfortable it is. Liquidation data tells you where leverage actually sat, after the fact, and where clusters remain. Together they describe the mechanical structure of the market — how a move is likely to propagate if one occurs.

None of them describes whether a move occurs. That distinction holds across every reading in this note, and it is the whole of what separates useful analysis of positioning data from expensive analysis of it.