EVA Help

Long/short ratio

What it is

The long/short ratio is how many accounts (or positions) are long for each one that is short. "Crowd" means all accounts; "large" means top traders by position size, as the exchange counts them. It is not the market's direction but the participants' sentiment.

More: how we calculate it

"all · by accounts" — the share of accounts in longs on Binance, Bybit, OKX and Gate, averaged with OI weighting and converted back into "longs per one short".
"large · by position size" — top traders by position size on Binance and OKX, aggregated the same way. "large · by account count · Binance" — Binance only.
Polling — every 5 minutes; exchange data covers 5 minutes (for "large · by account count · Binance" — an hour). Data counts as fresh for 15 minutes.
What 1.5 means: for every account in a short there are one and a half in a long. These are accounts and positions, not money.
The moments below were found using Binance data (crowd and large by positions) — their history is longer than the aggregated one (since 01.10).

What it looks like in EVA

**BTC · 1h · 09.10 07:00 MSK** — "Long/Short" pane: all accounts, large by position size, large by account countShow on chart

How it is read

Typical patterns: how traders read them and use them, and for each one a real moment from our history where it played out as a "textbook case". Moments where the same pattern failed are below, in "When it does not work".

1. The crowd is long more than usual

How traders read it: almost everyone is on one side. Many trade against the crowd: when everyone has bought, there is no one left to buy.

textbook caseETH · 1h · 08.10 03:00 MSK — crowd long/short 3.363 — in the coin's top 5%; then as expected: 1% down after 4 hShow on chart

2. The crowd is short more than usual

How traders read it: the crowd is betting on a drop. Many expect a short squeeze upward.

textbook caseBTC · 1h · 02.10 08:00 MSK — crowd long/short 0.8954 — in the bottom 5%; then as expected: 1% up after 7 hShow on chart

3. Large traders and the crowd diverge

How traders read it: the crowd moves to one side, large traders to the other. Many follow the large traders.

textbook caseSOL · 1h · 08.10 06:00 MSK — over 12 h crowd +5.7%, large −5.1%; then as expected: 1% down after 1 hShow on chart

When it does not work

No pattern works every time: the ratio explains better how accounts are positioned than it tells where the price will go.

The same patterns — moments from our history where they failed.

pattern failedETH · 1h · 16.09 12:00 MSK — crowd long/short 3.3497 — in the coin's top 5%; then the opposite: 1% up after 2 hShow on chart

pattern failedETH · 1h · 02.10 11:00 MSK — crowd long/short 2.202 — in the bottom 5%; then the opposite: 1% down after 6 hShow on chart

pattern failedETH · 1h · 05.10 05:00 MSK — over 12 h crowd −11.3%, large +5%; then the opposite: 1% down after 2 hShow on chart

Common traps

Accounts, not money: a thousand small longs outweigh one large short in the ratio.
Each exchange has its own method: every exchange computes the ratio from its own clients; our aggregated line covers four exchanges with OI weighting, while "large by account count" is Binance only.
An extreme can last — the crowd is sometimes right for weeks.
Hedges: some shorts are insurance for spot buys rather than a bet on a drop.