The Estimated Leverage Ratio sets open interest in futures against the coin balances held on exchanges. It answers something the open interest figure alone cannot: whether the leverage being carried is backed by a growing pool of collateral, or by the same pool stretched further.

That distinction decides how fragile a market is. Leverage built on top of expanding collateral has room to absorb a shock, and identical leverage built on a shrinking base has nowhere to go the moment price moves against it.

What it actually measures

Two quantities move independently here. Open interest can climb while exchange balances fall, and the reading rises faster still when both happen at once, which is the condition that has preceded most cascades.

The level is banded against its own history rather than read as an absolute. Exchange balances trend over long periods as custody habits change, so what counted as heavy leverage in one era is ordinary in another, and the bands adjust for that.

Falls in the reading are as informative as rises. A sharp drop is leverage being removed, usually involuntarily, and the deepest of those drops line up with the moments when this market was carrying the least risk.

Collateral is the half of this nobody quotes

Open interest gets reported constantly and exchange balances rarely do, which leaves most commentary tracking one side of a two-sided question. Record open interest against a growing collateral base is a completely different market from the same figure against a shrinking one.

The second case is the one that breaks. When collateral leaves an exchange while positions stay open, the buffer between an adverse move and a forced closure thins without anything in the headline number changing. This view exists to make that thinning visible while it is happening.

ELR by Venue: the average venue is not the one that gives way

The spread between venues is the reading rather than any single bar. Venues clustered together describe a system carrying leverage evenly, and one bar standing far above the rest describes a concentrated point of failure.

An aggregate reading describes a market that does not exist. Nobody trades at the average venue, and the cascade that eventually arrives will start at whichever specific book ran out of room first, not at the mean of all of them.

That makes the spread more informative than the level for this particular question. A market at a comfortable average with one venue at an extreme is carrying a risk the aggregate is actively concealing, and the concealment is worst exactly when it matters most.

What it does not tell you

Exchange balances are estimates built from address clustering, and clustering is a judgement rather than a fact. A custody reshuffle can move the reading without a single position changing hands.

Collateral held away from the visible chain does not appear here at all. Positions margined in stablecoins, or backed by assets sitting elsewhere, are counted on the exposure side and missing from the backing side.

A crowded reading describes a condition and not a trigger. Leverage has stayed crowded for months at a time, and something outside this view has to arrive before the condition turns into an event.

How to read it

Extreme. Leverage sits far above anything the collateral base has previously supported.

Crowded. Positioning is heavy against the collateral available, and the buffer underneath it is thin.

Building. Leverage is accumulating within the range this market has carried before.

Deleveraged. Exposure has been cleared out and the collateral base is comfortably ahead of it.

Estimated Leverage Ratio has a panel of its own on the Futures dashboard, as do ELR by Venue, Total OI and Cascade Detector.

Common questions

Why set open interest against exchange balances?

Because leverage only means anything against what backs it. The same open exposure is comfortable when collateral is plentiful and dangerous when it is not, and one figure on its own cannot separate the two.

Is the reading an exact measurement?

No. Exchange balances are estimated from address clustering, which is inference rather than observation. The direction and the standing against history are dependable; the precise level is not.

What happens when the reading drops sharply?

Leverage has been removed, and usually not by choice. Sharp declines coincide with liquidation events, and the deepest of them have historically left this market carrying unusually little risk.

Does stablecoin collateral show up here?

Not on the backing side. Positions margined in stablecoins contribute to open interest and add nothing to the coin balances, so the reading overstates leverage wherever that practice is common.

Can a crowded reading persist?

For months, yes. It describes how much slack the system has rather than when that slack runs out, and it needs a catalyst from outside the view before anything follows.

Why does the per-exchange figure matter?

Because positions fail somewhere in particular. Whichever exchange has stretched its collateral furthest runs out of margin first, and the selling that follows sets the print everybody else is forced out at.

How much can these figures be trusted?

Loose. Balances are deduced from chain activity and attribution is never exact, so ranking one exchange against another, and watching a bar move, hold up far better than reading any bar as a number.