Normalise open interest: contracts, coins and dollars

Build a small data dictionary

Record venue, instrument, contract type, multiplier, native unit, counting convention and timestamp. For a simple linear product reported in base units, quote notional is quantity multiplied by price. If the feed reports a number of contracts, apply the contract multiplier first. An inverse contract with a fixed quote face value follows a different conversion; do not reuse the linear formula blindly.

Counting conventions can also differ. As checked on 22 September 2026, Bybit's OI endpoint describes openInterest as the sum of both sides and singleOpenInterest as the single-sided value. The June 2026 changelog documents the new single-sided fields. This is a specific API convention, not a reason to divide every exchange's OI by two. Read the field documentation and the history around any series break.

Decompose before interpreting

For a linear base-quantity series Q and price P, notional N = Q × P. Its percentage change satisfies 1 + return(N) = (1 + return(Q)) × (1 + return(P)). Subtracting the two percentage changes is only an approximation because it ignores the interaction term.

For an exact additive decomposition, N1 − N0 = Q0 × (P1 − P0) + P1 × (Q1 − Q0). The first term values the original quantity at the new price; the second values the quantity change at the final price. Other valid decompositions allocate the interaction differently, so state the method when comparing reports.

Notional rises 15.5%, quantity only 5%

Start with 100 BTC of single-sided OI at 60,000 USDT: 6,000,000 USDT notional. Later the quantity is 105 BTC and price is 66,000: 6,930,000 USDT. The 930,000 increase contains 600,000 of price revaluation on the original 100 BTC and 330,000 from the additional 5 BTC valued at the final price.

Quantity grew 5% and price grew 10%; combined notional grew 1.05 × 1.10 − 1 = 15.5%. Calling the whole 15.5% “new positions” overstates quantity growth. When aggregating venues, perform the same normalisation for each constituent and flag unavailable data instead of silently treating it as zero.

A field migration can masquerade as deleveraging

If a data provider changes from a both-sides sum of 600 units to a single-sided count of 300 without documenting the conversion, the chart appears to lose half its OI instantly. No trader needed to close anything. Check release notes, overlapping fields and neighbouring venues before turning a discontinuity into a liquidation story.