How to Read Spot vs Perp Volume

Every number in this series so far has lived on the perpetual. Funding is paid on it, open interest is open on it, the taker ratio and the book and the cascade are all read off it. Underneath that sits an assumption: that the perpetual is where bitcoin actually trades, and spot is the tail. This article measures the assumption.

Binance runs both markets for the same coin. Its spot pair settles in bitcoin — you buy it, you own it. Its perpetual settles in dollars and nobody owns anything; it's a bet on the spot price, held open with margin. Both publish a daily kline with the dollars traded and the number of trades. Divide one by the other and you have the ratio, every day since the perpetual listed on 1 January 2020.

What the number measures

Take one day. Bitcoin's spot pair traded $7.2bn on 5 February 2026 and its perpetual traded $45.4bn. The ratio is 6.3: for every dollar that changed hands in actual bitcoin, 6.3 changed hands in bets on its price. The perp's share of the day was 86%.

Two things the number is not. It isn't open interest — open interest counts contracts still open at the end of the day, this counts everything that traded during it, most of which was opened and closed within hours. And it isn't a measure of leverage on any one trade. A perpetual trade at 1× leverage counts the same dollars as one at 50×. The ratio tells you where the dollars went, not how much margin was behind them.

Finding 1: nine dollars in ten, and it was eight

In 2020 the perpetual took 80% of bitcoin's Binance turnover. This year it takes 90%. Ether went from 72% to 92%, solana from 79% to 89%. The daily ratio has never printed below 1 on ether or solana — not once in 2,450 days — and on bitcoin only 2 times, both inside the fee window. Spot has not out-traded the perp on an honest day in the whole record.

The trade count says who left. Bitcoin's average spot trade was $911 in 2020 and $1,721 at the 2021 peak; this year it's $355. The average perpetual trade has stayed between $2,955 and $4,812 — this year $2,955, 8 times the spot trade. Ether's spot trade is $208 against a $1,545 perp trade. Spot is where the small orders are, and the small orders have got smaller. Size trades the contract.

Finding 2: the biggest move on the chart is a fee

On 8 July 2022 Binance made bitcoin spot free to trade on thirteen pairs, BTC/USDT among them. The day before, spot traded $1.78bn and the ratio was 8.2. That day, spot traded $8.78bn — 4.9 times as much — and the ratio was 2.1. It stayed near 2 for 257 days, and on 7 January 2023 it printed 0.75: the only stretch in the record where actual bitcoin out-traded bets on it.

Then fees came back. 21 March 2023 was the last free day, with spot at $11.8bn and the ratio at 1.6; the next day spot was $6.3bn and the ratio 4.8, and within a month it was back at 8.1. Ether wasn't in the promotion. Its ratio ran 5.1 in the month before the window and 9.7 during it — it went up, not down. Nothing changed about how people wanted to hold bitcoin in the second half of 2022. The price of trading it changed.

That's the first rule for reading this ratio, and it applies to every venue-level volume number in this series. It measures the exchange's pricing before it measures its traders. A dashboard that showed "spot demand surging" through the 2022 bear market was showing a promotion, and the traders who read it as accumulation were reading a marketing budget.

Finding 3: the big day is a spot day

The story goes that the big move is a leverage event — the perpetual is where the cascade happens, so the perpetual is where the volume goes. The two volumes do move together: the day-to-day correlation of their changes is 0.95 on bitcoin and 0.96 on ether, and the weekend cuts both by about half (spot to 52% of a weekday, perp to 51%). But on the day that matters, spot rises more.

On a day bitcoin falls 5% or more, spot volume is 3.0× a quiet day's and perp volume 2.4×; on a 5% rally, 2.6× against 2.0×. Ether is wider: 3.0× against 2.1× on the fall, 2.6× against 1.7× on the rally. The ratio drops from 7.8 on a flat day to 6.3 on a crash. The bigger the day, the larger spot's share of it.

10 October 2025, the day of the largest cascade in the record: $45.1bn on the perpetual, $7.3bn on spot, a ratio of 6.1 against a trailing 8.3. Spot's share that day was 14%, against 11% on an ordinary day. The forced selling happens on the perpetual; the buying of what it sells, at the price it sells it, happens in bitcoin.

Finding 4: spot-led and perp-led forecast nothing

The most common use of this ratio is as a quality label on a move. A rally with the ratio below normal is "spot-led": real buyers, real coins, healthy. A rally with the ratio above normal is "perp-led": leverage, fragile, due a flush. Take every day that rose 3% or more outside the fee window, label it by whether the ratio sat below or above its 90-day median, and look at the next thirty days.

Bitcoin: +4.7% after a spot-led rally, +5.1% after a perp-led one. That's the same number. Ether says perp-led paid — +9.0% against +1.6% — and solana says the opposite, +16.6% for spot-led against +9.6%. Three contracts, three answers.

Split by year and the answers fall apart further. Ether's edge for perp-led rallies is mostly 2022: −22.3% after a spot-led up day that year against +2.2% after a perp-led one, in a bear market where the perp-led rallies happened to land near the bottom. From 2023 on, spot-led wins 2 of 4 years on ether and the largest gap either way is 4%. A label whose sign depends on the year isn't a label; it's a coin.

Finding 5: a high ratio is a fingerprint of a fall

One thing does show up, and it's worth being precise about what it is. Band each day by its ratio against the 90-day median and look both ways — at the month before and the month after.

Days with the ratio 25% or more above its median came after a losing month — −3.3% on bitcoin, −9.1% on ether, −3.1% on solana — and before a better one: +5.2%, +12.3% and +9.9%, up 60%, 69% and 61% of the time. Read from the left, that's the mechanism: after a fall, spot volume dries up faster than perp volume does, so the ratio rises. Read from the right, it's just what the month after a fall has looked like on average. The ratio isn't forecasting froth. It's describing a drawdown that has already happened, and drawdowns in this sample were usually followed by recoveries.

How to read it yourself

• Expect 8, not 1. Bitcoin's perp trades about 8 dollars for every spot dollar on Binance; the middle 90% of days sit between 2 and 11. A reading of 3 is not "spot-led", it is a fee promotion or a different exchange.
• Check the fee schedule before the chart. A step change in the ratio on one pair, with the others unmoved, is pricing. The 257-day block in 2022–23 is the template.
• Judge it against its own last 90 days. The level has more than doubled since 2020, so an absolute threshold from one year is wrong in the next.
• Don't read a crash's ratio as a leverage measure. Spot's share rises on the big day. If you want the forced selling, read open interest and the cascade article; this ratio will point the other way.
• Don't grade a rally with it. Spot-led and perp-led up days were followed by the same month on bitcoin, and by opposite months on ether and solana.
• A ratio well above its median is a sign the last month was bad, not that the next one will be. The average next month in this sample was a recovery, and the average is doing a lot of work in that sentence.

Method, and what this cannot tell you

Each UTC day's spot volume and perpetual volume are the quote-asset volume (dollars traded) on Binance's BTCUSDT, ETHUSDT and SOLUSDT spot pairs and their USDT-margined perpetuals, from the day each perpetual listed. The ratio is perp dollars over spot dollars; the perp share is perp dollars over the two combined; the average trade is dollars over the kline's trade count. The relative ratio is the day's ratio divided by the median of the previous 90 days, and a day is spot-led below 1 and perp-led at or above it. The surge multiples compare the median volume on days in each return bucket with the median on days that moved under 1%. The led and band tests score the 30 days after each day by the spot close, and they exclude the zero-fee window (8 July 2022 to 21 March 2023) because every one of its days is spot-led by construction. The weekend test compares the median volume on Saturdays and Sundays with weekdays, outside the window. Years are calendar years; 2026 ends on 15 September.

• One exchange. Binance is the largest venue on both sides, but Coinbase, Kraken and the ETFs are spot volume this study does not see, and Bybit, OKX and CME are derivatives volume it does not see. The ratio here is Binance's, and Binance's spot share of the world is not the same as its perpetual share.
• USDT pairs only. Bitcoin also trades on Binance against USDC, FDUSD and fiat on spot, and as a coin-margined perpetual. Those flows are left out on both sides, and the fee promotions on some of them (FDUSD in 2023-24) would move a broader ratio the way the USDT one moved in 2022.
• Reported volume, not verified volume. Exchange volume can be inflated by wash trading and by fee rebates, and a zero-fee pair is the cheapest one to inflate. The fee window is named for that reason; it is not the only place the number may be padded.
• A 'trade' is one match. An order that sweeps three resting orders counts as three, so the average trade size understates the average order. The comparison across years and between venues is still fair because both sides are counted the same way.
• The led and band tests overlap. Consecutive days share most of their forward window, so the effective number of independent observations is a fraction of the day counts, and the by-year rows are small. Read the signs, not the second decimal.
• Six and a half years, one full cycle. The share has risen through a bear market and a bull market, which says it is structural rather than cyclical, but whether a spot ETF era reverses it is not something this sample can answer.