Does more trading make a coin less volatile? Turnover, futures and six months of prices

The data

Every cryptocurrency in the site's price history with at least 90 days of clean daily prices and a median of $100,000 or more traded a day: 103 coins for the 90 days to 25 October 2026, 106 for the 180 days. Volatility is the standard deviation of daily price changes, annualised. Turnover is each day's volume divided by that day's market cap, the median over the window. Futures exposure is open interest on Binance, Bybit, OKX, Bitget, Gate, MEXC and Hyperliquid as a share of market cap on 26 October, and Bybit's daily open interest over the window for the 74 coins it lists. Market cap differs by a factor of a million across the sample, and large coins are both calmer and less actively traded, so every result is also checked within size groups.

Turnover: higher went with more volatility

90 days to 25 October 2026. Volatility annualised; largest fall is the deepest drop from a high within the window.

Size does not explain it away. Within each market-cap group, the coins turning over more were the more volatile ones: among coins worth $100 million to $1 billion, 73% under 0.5% turnover, 90% at 0.5 to 2% and 101% above 2%; among coins under $100 million, 76%, 93% and 97%. With size held fixed, the rank correlation between turnover and volatility was 0.43, far from chance.

Nor is it only that volatile weeks bring trading. Turnover in the earlier quarter came before higher volatility in the next one, with size and the earlier volatility held fixed: each tenfold step in turnover, about 16 points more annualised volatility. It had no measurable link to the next quarter's largest fall.

What turnover does show

A thinly traded coin moves further for the share of its supply that trades. Measured as the price change per 1% of market cap traded in a day, the median coin moved:

The measure divides by turnover, so part of the slope is arithmetic, but not all of it: if price moves grew in step with trading the column would be flat. A market cap resting on a trickle of volume is a price set by few trades, and that is what the grade's turnover ladder rewards. It is a measure of how widely the price is set, not of how calm it will be.

Futures exposure: a link, not a forecast

Among coins with perpetual futures listed, more open interest went with more volatility in the same period, and that held with size fixed (rank correlation 0.41 on open interest today, 0.43 on Bybit's 90-day average). Within size groups:

Median 90-day volatility, number of coins in brackets.

But open interest in the earlier quarter did not predict volatility in the next once the earlier volatility was known. Exchanges list futures on coins that already move, and traders open positions where prices move, so the link runs both ways and the data cannot say which way more. The small coins with no perpetual at all were the most volatile group in the sample: no futures did not mean calm.

Monero

Monero is often described as calmer because little of it trades on futures. Its open interest is ordinary for its size: 2.2% of market cap on 26 October, against 2.0% for Bitcoin and 2.2% for XRP, and 0.34% on Bybit's 180-day average against Bitcoin's 0.28%. Its volatility was not low:

180 days to 25 October 2026. Turnover is the median day's volume on the exchanges in the site's price history, so it reads lower than the live figure; open interest is today's, all seven venues.

What set Monero apart was the shape of its move: up 57% over the six months with a largest fall of 29%, the same as Bitcoin's. Among the coins near it in market cap, three rose more: Zcash and NEAR, which fell 44 to 46% along the way, and Hyperliquid, up 90% with a largest fall of 30%.

Both together

The combination, high turnover with little or no futures, could not be tested fairly: the coins in that corner were tiny, a median of about $5 million in market cap, and the most volatile group of all. A cleaner test needs more months, and more coins that are both actively traded and free of futures.

Turnover and the order book

If turnover is not calm, is it liquidity? A follow-up read the order books themselves: on 26 October 2026, every USDT market on Binance, OKX, Bybit, Gate and MEXC, about 4,800 books, matched to 1,262 assets with a market cap (stablecoins left out). For each asset, the dollars resting within 2% of the price on both sides were added up across the five exchanges and set against its market cap.

One snapshot, 26 October 2026; turnover is the 30-day average where the site keeps daily history and the live 24-hour figure otherwise. Medians. Above 100% a day left out.

Up to about 5% a day the answer is yes. A coin trading under 0.2% of its market cap a day had a median of 0.011% of it resting near the price; at 3 to 5%, 0.31%, about thirty times deeper. With market cap and the number of exchanges held fixed, each 1% more turnover went with about 0.45% more depth. A thinly traded coin has a thin book: there is little to buy or sell at the price, which is why each trade moves it further.

Above 5% the relation weakens. Taking out what size and venue count predict, coins over $10 million reached the depth their size predicts by about 4% turnover and gained little beyond 6%; the 7.5 to 15% row owes its height to a few coins under $10 million whose books are large against a tiny market cap. Up to 6% a day, a day's trading was 10 to 17 times the depth near the price, and from 6 to 8% it jumped to 36 times. What rises instead is the volume set against the book: above 15% a day, a day's trading is about 30 times the depth near the price, and above 25% about 100 times. That much trading against so little resting liquidity is churn, not depth.

Rallies from the all-time low

A second question: does heavy trading near the all-time low come before a recovery? Every coin with three years of daily history was sampled weekly whenever it stood within 25% of its lowest close to date, 2,921 coin-weeks across 84 coins, banded by its average turnover over the previous 30 days, and followed for up to 180 days.

October 2023 to October 2026, weekly samples. Coin counts are for the 90-day horizon.

Heavily traded lows did produce bigger and quicker rallies. Coins turning over 10% a day or more near their low reached a median peak of 35% within 90 days, against about 20% in every other band, and nearly a third of them doubled within 180 days. Trading of 2% or more at the low came before a 50% rally within the next month about 6 points more often than trading under 0.5%, a gap that held when coins were resampled. A coin no one trades at its low rarely rallies hard.

The rallies did not hold. At 90 days the median coin was 15 to 17% lower in every band, and at 180 days 31 to 38% lower, with falls just as deep along the way. The same held for coins that had set a new low within the previous 30 days. Earlier tests on the same history pointed the same way: high turnover near the low did not make the falls that followed smaller, high turnover during a rally came before deeper reversals rather than a longer run, and across ordinary weeks high turnover came before weaker returns. Heavy trading at the low raises the odds of a sharp rebound, not of a lasting floor.

Bitcoin cycles: turnover in downtrends and uptrends

Bitcoin has the longest history and the clearest cycles, so its downtrends and uptrends were measured phase by phase: the last 60 days of the bull run, the downtrend from the peak, the bottom (30 days either side of the lowest close), the first six months of the uptrend and the rest of it. Turnover here is BTC traded ÷ BTC in circulation, with the supply rebuilt from the issuance schedule for each date, so the years compare fairly.

No single exchange is enough. Binance carried 27% of the Bitcoin volume counted here in 2018, 71% in 2023 and 31% in 2026, and from July 2022 into 2024 much of its count was zero-fee trading. The main series adds every Bitcoin pair on the established exchanges (Binance, Coinbase, Bitstamp, Bitfinex, Upbit, Kraken, and Poloniex through 2019), from 2017; the cycle before uses the site's price history (Bitstamp, Bitfinex, Coinbase, Poloniex). Two whole-market aggregators were run beside it: CoinMarketCap, which counts every exchange it tracked, closed ones such as Bittrex, BTC-e and Mercatox included, and Bitcoinity, which keeps per-exchange history for the fiat venues back to Mt.Gox. Both carry the wash trading of their time. CoinMarketCap's Bitcoin volume for 2019 and 2020 comes to 2.3 to 3.3 million BTC a day, 12 to 18% of the supply, and Bitcoinity's 2016 is 2.9 million BTC a day, almost all of it zero-fee trading on Huobi and OKCoin. Their levels are not the real market; how one phase compares with another in the same series still says something.

Mean turnover a day. Peaks: December 2013, December 2017, November 2021, October 2025. Lows: $182 on 14 January 2015, $3,212 on 15 December 2018, $15,781 on 21 November 2022, and $58,625 on 30 June 2026, the lowest close of the current decline so far. For 2025–2026 the uptrend row is only the months since that close.

Turnover stayed high through the downtrends: in 2022 the decline ran at about twice the turnover of the bull top before it on the exchange count, and a little above it without Binance. Buyers stepping in on the way down make the order book that every fall trades into. And in 2015, 2018 and 2022 the bottom was busier still, on every source: Bitcoin bottomed on more turnover than the decline that led to it. The first six months of each uptrend were quieter than the bottom, and turnover came back later, if at all; the bull tops of 2021 and 2025 were among the quietest phases of their cycles, as more of the buying moved to futures and, from 2024, to exchange-traded funds that no exchange count includes.

The current decline does not fit so far. Its busiest stretch was the February 2026 fall from $78,000 to $63,000, and the weeks around the $58,625 close were quieter than the downtrend on the exchange count and on CoinMarketCap. Only the fiat exchanges Bitcoinity tracks, 61% of them Coinbase this year, ran busier at that low than on the way down. Whether $58,625 was the bottom of this cycle is not known yet.

Single days tell the same story. Every cycle low since 2014 had a week trading 1.8 times or more of the previous six months' median within 25 days of it: early February 2015, November 2018 at $4,559, 12 March 2020 at $4,800, the FTX collapse of 8 November 2022, and early June 2026. But such weeks also came early. Of nine since 2014 while Bitcoin stood 50% or more below its high, three were followed by another 34 to 48% fall: November 2014, May 2022 and August 2022. Measured from the spike, the following year returned a median 70%, against 66% for any day in the same kind of low. Heavy trading marked the zone where Bitcoin bottomed, not the day.

Falls on heavy trading did bounce harder. On the days Bitcoin fell 5% or more while 20% or more below its high:

2014 to October 2026, site price history. Medians.

A drop on three times the usual trading bounced about twice as far within a week as a drop on ordinary trading, and a month later stood about where it fell, against 5.5% lower. The bounces are the demand that the turnover measures. They did not mark the last fall: the 2014 and 2022 declines had heavy-trading drops that bounced and then gave way. Across other coins, 52,452 such drops, the bounce came whatever the turnover (a best close within 30 days of +14% on ordinary trading, +17% at three times or more), and a month later both stood 6 to 7% lower.

Monthly trend signals

Turnover is a slow signal; so are the trend signals read on monthly closes: the MACD (12, 26, 9) crossing its signal line, and the 9-month average crossing the 26-month one, simple and exponential. Each was read at the month's close, so none uses a later price.

Monthly closes from 2011; the signals need 35 months of history, so they start in 2014. The MACD also crossed down above zero in December 2019 and February 2020, both followed by gains of more than 300% within a year.

The upward crosses confirmed each uptrend late: 5 to 16 months after the low, with the price already 70 to 150% above it, and still followed by gains. The MACD rarely crosses up below zero on a monthly chart; it did once, in July 2023. The MACD crossing down from above zero confirmed four of the six downtrends it signalled, two to ten months after the peak, with another 35 to 57% fall still to come. The 9/26 crosses down were slower still: each came within three months of a cycle low, and the latest printed in June 2026, the month of the $58,625 close. In Bitcoin's history that cross has marked the late stage of a decline rather than its start; four cycles are too few to say more.

Across 127 other coins with five years of history, where the median month was followed by a 36% fall over the next year, only the downward signals held up. A MACD crossing down above zero was followed by a median 55% fall, 19 points worse than the base rate (95% range 3 to 45), and 62% when turnover had run 1.5 times its norm in the six months before. The 9/26 simple average crossing down was followed by 50%. The upward crosses were no better than an ordinary month, with or without heavy trading. For Bitcoin these signals, with turnover, confirmed turns that had already happened; for other coins they confirmed falls, not recoveries.

A combined bottom rule

Put together, the pieces suggest a rule: after a monthly 9/26 death cross, the first month that closes at a new 12-month low on turnover of 1.5 times or more its 12-month median marks the bottom or comes close to it. Tested on monthly closes, with "the bottom" meaning no lower monthly close in the 24 months after:

The cross does filter something out: May and June 2022 were new lows on heavy trading before it, and another 48% and 17% fall followed. Across 127 other coins:

The combination picked better moments than any other kind of new low, +31% two years on against −36% and −2%, but it rarely picked the bottom: 13% of the time, with a median 42% still to fall. On 24 cases that is a lean, not a rule. In Bitcoin the pieces gather around each low without lining up the same way twice, and by this rule $58,625 is not yet confirmed as the bottom.

Weekly signals: faster, noisier

The same crosses on weekly closes turn sooner. The first signal in the right direction after each of Bitcoin's turns, weeks after it and how far the price had already moved:

Weekly closes from Monday, UTC. The 9/26 exponential averages crossed within a few weeks of the simple ones. "—": the monthly averages had not crossed back since the previous turn.

The weekly MACD confirmed most turns within 4 to 10 weeks, often with most of the move still ahead, against 8 to 50 weeks on the monthly chart; the weekly 9/26 crosses took 6 to 26 weeks. The price of speed is noise. Since 2012 the weekly MACD crossed 51 times, 3.6 a year, and the price moved its way over the next 26 weeks only 53% of the time, no better than a coin toss; it crossed up in April 2026 at $73,802 before the fall to $58,625, and after the 2022 low it took 31 weeks because an early cross up was reversed. The weekly 9/26 crosses were right 62% (simple) and 67% (exponential) of the time, about twice a year. The monthly MACD was right in 8 of 11, and the monthly 9/26 crosses, at well under one a year, came so late that they were right only half the time or less. Across 128 other coins no weekly signal did better than an ordinary week over the next 26 weeks.

The noise gathers at the turns, where a weekly signal is most wanted. Counting every weekly cross, either way, from 26 weeks before each turn until the monthly MACD confirmed it:

One cross means a clean turn; each pair beyond it is a signal reversed. Crosses in the 26 weeks before a turn count, because a cross can come early.

Only the 2017 top gave one clean weekly cross on both charts. Before the 2018 bottom the weekly MACD crossed up three times in September 2018, near $6,500, and the price then fell to $3,212; in 2026 it crossed up in April, down in June and up again in July. The weekly 9/26 averages flipped less, once at four of the ten turns, but four or five times around the 2013 top, the 2015 bottom and the 2020 crash.

Checking a weekly MACD cross against the monthly chart helped. Of its 44 crosses from 2014 on, the 18 that pointed the same way as the last monthly MACD reading were followed by a move their way over 26 weeks 67% of the time; the 26 that pointed against it, 42%. The same check does not work as a filter for the weekly 9/26 crosses: those against the monthly MACD were right 73% of the time (15 crosses), those agreeing with it 38% (8), because the weekly cross that matters at a turn comes before the monthly chart turns. For those, the monthly signal is a later confirmation, not a test at the moment of the weekly cross. Across 128 other coins, agreeing with the monthly MACD made no weekly signal better than an ordinary week.

Weekly signals suit an earlier, provisional read of a turn, and they can flip several times before the monthly signals give the later, firmer one. As of October 2026 the weekly MACD has been up since 26 July and the weekly 9/26 averages since 13 September, at $76,842; the monthly signals have not turned.

The whole sequence

The combined rule asks the steps to come close together. In practice they spread out, so the looser version sets no deadline between them, only an order: (1) the monthly MACD crosses down while above zero; (2) the monthly 9/26 average, simple or exponential, crosses down within two years of it; (3) at any time after, until the averages cross back up, a capitulation week, seven days that fall 15% or more on twice the usual trading (the median of the previous 180 days, counted in coins so the price does not inflate it); (4) the faster confirmation after it, the weekly 9/26 simple average crossing up while the weekly MACD is above its signal line. A price drop and heavy trading are not guaranteed at step 3, so a setup can stop there.

Only 2022 ran through every step in order, and there the capitulation week landed within 1% of the bottom. The steady part was the death cross: in every cycle the lowest close of the next two years was at most 21% below it, and in 2015 and 2019 the cross came after the cycle low itself. In 2026 it came on the lowest monthly close so far, and no capitulation week has followed.

Across 128 other coins with five years of history the sequence started 85 times:

For other coins the sequence came much earlier than for Bitcoin: the death cross a year before the bottom on a median, and the capitulation week rarely the last one. The weekly confirmation picked better moments than an ordinary day, but its lows seldom held, and 28 of the 53 came in 2023's recovery and were up a median 51% a year later; the 9 of 2024 and the 4 of 2025 were followed by median falls of 64% and 73%.

What changes in the grade

The method page described low turnover as the source of sharp price swings. That claim was taken out the same day. The turnover ladder stays, described for what the data supports: how much of the supply the price is set by. Futures exposure stays as a measure of paper supply beside the real one, described as exposure, not as a cause of volatility. Both are published for every asset each minute at /api/v1/liquidity.json.

The order-book test set the top of the ladder. The grade scores the 30-day median of daily turnover, because one day's figure changed a coin's score on almost half of all days. Since method version 10, 4 to 6% a day scores 10, the band where books have reached their depth and trading is still backed by them, and the score falls by the same amount for each tenfold step away from it in either direction. Above 100% a day, where the volume and the market cap cannot both be right, turnover is left out and the asset leaves the volume rank. Since method version 20 (27 October 2026) the top is a single point, 5% a day, where the books stopped deepening.

Limits

One six-month window, in one market, on about a hundred coins. Daily volume comes from the exchanges in the site's price history, not every venue; circulating supply is held at today's figure, which overstates the earlier market cap of coins still issuing. Open-interest history is one exchange's and reaches back 200 days. The order books are one snapshot on one day, on five exchanges, and depth beyond the levels each exchange returns is not counted. The rally test covers only coins still listed today, which leans it toward recoveries, over a period when most coins other than Bitcoin fell; the 10%-and-over group is 18 coins. The Bitcoin cycles rest on four declines; the exchange count misses closed venues such as Bittrex and all exchange-traded fund volume, and the aggregators that include closed venues also include their wash trading. The combined bottom rule fired in one Bitcoin cycle and 24 times across other coins; the looser sequence ran its full course in one Bitcoin cycle, and more than half of its confirmations in other coins fell in 2023. The scripts are in the site's repository under scripts/research/turnover-volatility, and the test will be rerun as the history grows.

Questions