August, closed

AssetAugust changeMonthly buy share
Solana+42.35%51.3%
Ether+32.70%50.2%
XRP+30.47%48.5%
Bitcoin+25.36%50.7%
Dogecoin+19.45%50.1%
Tokenised gold+9.90%49.4%

Bitcoin ranged from $62,275 to $81,479 across the month, a span of 30.84%, on $33.5 billion of spot volume. Gold gained 9.90%, so this was not purely a crypto story, but the majors ran three to four times harder.

Look at the XRP row before going further. It gained 30.47% over the month on a buy share of 48.5%, which is below the halfway line. A third of a year's worth of return arrived on a month in which more value crossed the spread selling than buying. That is not an error in the data and it is not a contradiction. It is the thing this site has spent two weeks documenting, showing up on a monthly candle.

Eight months, and almost no movement in the number

Here is 2026 so far, month by month, with the buy share beside each return.

MonthBitcoin returnMonthly buy share
January10.16% lower48.7%
February14.94% lower48.1%
March+1.96%49.5%
April+11.81%50.6%
May3.50% lower50.2%
June20.43% lower48.4%
July+7.27%49.5%
August+25.36%50.7%

The returns column swings across 45.8 percentage points, from June's 20.43% loss to August's 25.36% gain. The buy share column swings across 2.6.

That is not a quirk of 2026. Across all 99 completed months in our data the monthly buy share has never once printed below 47.1% or above 54.4%, a total historical span of 7.3 percentage points. Monthly returns over the same 99 months span 98, from a 37.29% loss to a 60.75% gain. One variable moves thirteen times as far as the other.

Note also that May, which fell 3.50%, had a higher reading than July, which gained 7.27%. At the monthly scale the ordering is not even reliably preserved.

The two Novembers

Widen the window to all 99 completed monthly candles in our data, going back to May 2018, and the point sharpens into something close to a paradox.

MonthReturnBuy share
November 201836.55% lower50.8%
November 2020+42.82%48.9%

A month that lost more than a third of its value registered 50.8%. A month that gained more than two fifths registered 48.9%. The crash had the higher reading by nearly two points.

That is not cherry picking two odd months to make a point, and the aggregate confirms it.

GroupMonthsMedian buy share
Gained more than 20%1849.9%
Lost more than 20%548.8%
Difference1.1 percentage points

Eighteen months that gained more than a fifth of their value sit 1.1 percentage points above five months that lost more than a fifth. Within the up group the readings run from 48.5%, on a month that gained 24.05%, to 53.5%, on a month that gained 20.96%. The month with the strongest reading in that entire group was not the month with the strongest return.

The ladder, completed

On Thursday we published a guide with percentile tables for five minute, hourly and daily candles. The monthly figures finish the ladder, and the pattern is almost mechanical.

Timeframe10th percentile90th percentileSpan
5 minute29.1%71.2%42.0 points
1 hour39.9%60.8%20.9 points
1 day45.7%52.4%6.7 points
1 month48.1%51.2%3.2 points

The span roughly halves at every step up. Forty two points on a five minute candle becomes three on a monthly one. The cause is averaging and nothing more mysterious: a five minute window can be dominated by a single participant, while a month contains millions of independent decisions in both directions and pulls the result to the middle.

One thing does not simply continue, and it is worth flagging because it breaks the tidy story. The correlation with returns does not hold up at the monthly scale. It runs 0.500 on five minute candles, 0.355 hourly and 0.492 daily, then falls to 0.160 monthly. So the metric is not merely quieter over a month, it is also carrying less information about direction. We did not expect that and we are reporting it rather than smoothing it over.

The practical reading stays what it was on Thursday. Judge a number against its own timeframe. August's 50.7% is in the 83rd percentile of monthly readings, which is genuinely strong for a month. Read against a five minute chart, where 71% is routine, it would look like nothing happened.

Today's session, and the flows behind the month

Bitcoin sits at $78,838, effectively unchanged over 24 hours at +0.04%, having ranged 2.40% between $77,392 and $79,250. The session dipped to $77,930 in the 12:00 hour on a 39.2% buy share and recovered through the afternoon, with the 17:00 hour reading 60.4%. Perpetual open interest is flat at 0.09% higher across 26 hours and the long/short account ratio has drifted to 1.0060, back to almost exactly even.

On the flows, August will finish as the strongest month for US spot Bitcoin ETFs since October 2025, with roughly $3 billion of net inflows. Fund assets are reported near $99.05 billion.

And the figure that belongs beside it, which we also carried on Friday: even after this month, the funds remain net negative for 2026 by about $2.26 billion. August cut the year's outflows by more than half. It did not erase them. A 25% month and a year still in deficit are both true.

Where things stand

AssetPrice24h24h range
Bitcoin$78,838+0.04%$77,000 to $79,250
Ether$2,473.871.08% lower$2,387.28 to $2,504.46
XRP$1.38561.23% lower$1.3352 to $1.4062
Solana$103.811.35% lower$100.31 to $105.44
BNB$690.561.21% lower$679.09 to $699.88
Dogecoin$0.083192.63% lower$0.08081 to $0.08567
Tokenised gold$4,4490.38% lower$4,397 to $4,468

Bitcoin is flat and every alt is lower, which is the reverse of the month's leadership. Total market capitalisation is $2.670 trillion, down 2.03% over 24 hours, with Bitcoin dominance at 59.19%. The price sits 13.60% above the 200-day moving average of $69,400.

Funding settled at 0.0100% for the last two periods, back at the venue ceiling. Fear and Greed reads 62, Greed, down from 69 yesterday and 73 on Friday.

The ETH to BTC ratio is 0.03136, down 1.17%. That is the ninth consecutive day we have reported it: up, up, down, up, down, up, up, up, down. Nine observations, no trend, and we will keep saying so until there is one.

What we are not saying

We are not saying the monthly reading predicts September. The monthly correlation with same month returns is 0.160, and we have not tested a forward version at that scale because 99 observations would not support one.

We are not saying a 50.7% month is bullish. It is the 83rd percentile of monthly readings, which describes August and nothing after it.

We are not saying the metric failed in August. For Bitcoin it tracked well: an 85th percentile return alongside an 83rd percentile reading. For XRP it did not, and we said so.

We are not drawing anything from the two Novembers beyond what they demonstrate, which is that the sign of a large move is not reliably encoded in the taker balance at monthly scale.

What we are saying is that August closed up 25.36% with a buy share 2.3 points above the month that fell 20.43%, that the readings compress predictably as the window lengthens, and that the correlation weakens at the top of the ladder in a way we did not anticipate. Every figure came from public endpoints and every one is reproducible.