Every morning this desk publishes a set of numbered, dated, falsifiable markers — predictions with a bar, an instrument and a deadline — and then grades them in public whether they work or not. This is Wednesday 26 August 2026. Nine markers are live, one settles at 8:30 this morning, and the most interesting thing on the board is not any single marker. It is a divergence between two groups of traders that has just reached its widest point in the entire window Binance publishes.
The find: the two cohorts are now further apart than at any point in 30 days
Binance publishes two positioning series that are easy to confuse and mean opposite things. The global long/short account ratio counts accounts — one vote each, so it is dominated by small retail traders. The top-trader long/short position ratio weights by position size across the largest accounts on the venue. When they move together they tell you nothing. When they separate, somebody is wrong.
They have separated to an extreme. Take the last seven daily readings, 19 August to this morning, over which bitcoin rose 21.9% from $64,725 to $78,876:
| Date (2026) | Top traders (by size) | Crowd (by account) | Spread |
|---|---|---|---|
| 19 Aug | 1.4342 | 1.3776 | 1.041× |
| 20 Aug | 1.5012 | 1.0773 | 1.393× |
| 21 Aug | 1.7206 | 0.9554 | 1.801× |
| 22 Aug | 1.9597 | 1.0387 | 1.887× |
| 23 Aug | 2.0533 | 1.0072 | 2.039× |
| 24 Aug | 2.0859 | 1.0700 | 1.949× |
| 25 Aug | 2.0203 | 0.9448 | 2.138× |
| 26 Aug | 2.2560 | 1.0008 | 2.254× |
Over those seven readings the size accounts increased their long tilt by 57.3% while the crowd cut its own by 27.4%. This morning’s top-trader reading of 2.2560 is the highest in the full 30-day window Binance makes available, and the 2.254× spread between the two cohorts is also a 30-day high — it has now risen on eight of the last nine readings, from 1.002× on 18 August.
What makes it sharper is the candle it happened on. Tuesday was a down day: bitcoin printed $81,272.62 overnight, failed, and closed at $78,539.14, off 0.57%. The largest accounts on the venue responded by adding to longs at the fastest single-day rate of the month (+11.67%). The crowd, which had been net short on Tuesday morning at 0.9448 — a 30-day low — came back to exactly 50.02% long, 49.98% short. Dead flat, to four decimal places.
Two cohorts looked at the same rejection candle. The size money bought it. The crowd went to cash and stopped having an opinion. One of those groups is going to be embarrassed by Friday, and we do not get to know which in advance.
The honest caveat: this is one venue, these are Binance’s own definitions, and a “top trader” category is a disclosure choice, not an audited universe. A 30-day window is also the longest Binance serves on this endpoint, so “30-day high” is a statement about the data available, not about all history. What it is not is ambiguous about direction.
The board
| Marker | Bar | Deadline | Now | State |
|---|---|---|---|---|
| X2 | Brent settle ≥ $95.00 | 28 Aug | $86.99 | Failing, 8.44% away |
| Y1 | 30y close ≥ 5.35% | 28 Aug | 5.183% | Failing, 16.7bp away |
| Z2 | Coin OI ≥ 111,988 BTC | 28 Aug | 106,592 | Failing, 4.82% below |
| Z3 | ETF net 24–27 Aug ≥ $0 | 27 Aug | +$651.9m | Passing, 2 rows in |
| A2 | Aug close ≥ $71,440.63 | 31 Aug | $78,876.00 | Passing, 10.41% cushion |
| A3 | July core PCE ≤ 3.2% | Today 08:30 ET | — | Open, settles today |
| B1 | Any funding > 0.0100% | 28 Aug | Max 0.0100% | Failing, 24 settlements |
| B2 | Aug ETF total ≥ $3,424.9m | 31 Aug | $3,049.8m | Improved, $375.1m to go |
| B3 | Strategy 8-K buys > 0 | 31 Aug | — | Open |
X2 and Y1: grading the premise, not just the marker
X2 requires a Brent settle at or above $95.00 by Friday. Brent for October delivery is at $86.99 this morning, down 1.79% on the session. On 23 August it was $94.39 and the marker was 0.65% away. It is now 8.44% away, and Brent has fallen 7.84% in three sessions. Y1 requires a 30-year Treasury close at or above 5.35%, which would be a new 2026 high; the 30-year is at 5.183%, sixteen and a half basis points short, and further away than it was on Monday.
Both markers will fail, and this desk said in print on Tuesday that the shared premise underneath them — sanctions escalation transmitting to bitcoin as an energy and inflation shock — looked wrong. It is now worse than wrong; it is inverted. On Monday 24 August the Treasury launched what it called Operation Economic Outcast against Iran, with roughly sixty designations and five sectoral determinations under Executive Order 13902, digital assets among them — the first time any OFAC sectoral determination against any country had named the sector. Since that announcement, crude is down 7.84% and the long end is down, not up.
That matters well beyond two markers, because the same premise sits underneath a large share of the commentary currently attached to this rally. If you are told bitcoin is rising because sanctions are stoking an energy-driven inflation impulse, the oil price is the cheapest available refutation, and it is refuting.
B1: twenty-four settlements, and a ceiling that has not been touched
B1 asks for a single Binance BTCUSDT perpetual funding settlement above 0.0100% by Friday 16:00 UTC. Since 21 August 16:00 UTC there have been 24 consecutive settlements, and the maximum across every one of them is exactly 0.0100% — the venue’s baseline. Not one has printed above it. Two printed below: 0.0058% on Tuesday morning and 0.0077% at this morning’s 00:00 UTC settlement.
Bitcoin covered roughly $18,700 of range inside that window, made a three-month high and gave it back, and the price of holding a long position never went up once. Whatever has been bidding this market is not paying for leverage to do it. B1 is our test of whether a leveraged bid ever arrives, and with three settlements left before the deadline it is on course to fail cleanly.
Z2 is the companion, and it is this desk’s own falsification marker: if coin-denominated open interest reclaims the 15 August peak of 111,988 BTC, the “no leverage in this rally” thesis is dead and we said so in advance. It read 106,592 BTC at this morning’s 00:00 UTC snapshot, 4.82% below the peak, and the live figure has since slipped to 106,088 — 5.27% below. (Those two percentages are the same series eight hours apart, and we label which is which because last week we did not.) Dollar open interest fell from $8.465bn to $8.368bn overnight. Tuesday’s small build did not continue. The thesis survives another day, which is a fact about the market, not a compliment to us.
Z3 and B2: the flow markers, and a correction to our own odds
Z3 asks only that the sum of ETF net flows for 24–27 August be at or above zero on completed rows. Two rows are in at +$337.6m and +$314.3m, a running +$651.9m. It passes unless something breaks.
B2 is the one that has moved, and we owe a revision. Yesterday we called it “roughly a coin flip,” needing $689.4m across four sessions to clear October 2025’s $3,424.9m. After Tuesday’s $314.3m, August stands at $3,049.8m and the requirement is $375.1m across four remaining sessions — 26, 27, 28 and 31 August — or $93.8m a session. The seven-session streak has averaged $367.1m. B2 is no longer a coin flip; it needs roughly a quarter of the current run-rate, and it should be graded as favoured. Calling it a coin flip yesterday was correct with yesterday’s row and wrong by this morning.
One thing worth holding onto while that plays out. Tuesday’s $314.3m was 90.5% IBIT — the narrowest single session of the streak, against a prior range of 53.8% to 83.0%. FBTC fell from $104.6m to $15.4m. B2 measures the total; it does not measure whether the total is coming from one fund or ten. If August clears October 2025 on the back of a single issuer, the record will be real and the breadth behind it will not be.
A3 settles at 8:30, and it is closer than we said
A3 asks for July core PCE at or below 3.2% year on year, published by the BEA at 8:30 a.m. ET this morning. June printed 3.3%. Yesterday this desk wrote that consensus had “consolidated at 3.3%,” pointing the marker at a fail. That overstated it. FactSet carries 3.2%; several other houses carry 3.3%; the Cleveland Fed nowcast sits at roughly 3.29%, which rounds to 3.3% but sits close enough to the boundary that a soft month-on-month print flips it. A3 is a genuine coin flip to one decimal place, and we should not have implied otherwise.
The release also carries the second estimate of Q2 GDP with corporate profits, and July advance durable goods, in the same 8:30 window. Two days later, Chair Warsh gives his first Jackson Hole keynote at 10:00 a.m. ET on Friday, under a symposium theme of “Financial Innovation: Implications for Payments and Policy.” The full agenda does not publish until Thursday at 8:00 p.m. ET, so anyone telling you what Warsh will say is guessing at a document that does not exist yet.
A dated bull call, published before the high — Chart Champions, 20 August 2026, six sessions before the $81,272.62 print and the rejection that followed. Included with its date so you can grade it rather than absorb it.
https://www.youtube.com/watch?v=nansyVfrM0MThe hashrate still will not agree with the price
A marker-adjacent note that keeps earning its place. Bitcoin is up 25.42% this month, and the network’s next difficulty adjustment is currently projecting a 3.08% cut, 1,561 blocks out and 22.6% of the way through the epoch. The retarget just recorded was −1.31%. A 3.08% cut would be the largest since the −5.00% at height 957,600, three retargets ago. Network hashrate reads roughly 858.6 EH/s and transaction fees are pinned at the 1 sat/vB floor.
Miners are switching machines off into a 25% monthly price rally, and nobody is bidding for block space. That is not a prediction of anything by itself. It is simply the third independent series — alongside funding and open interest — declining to confirm the story the price is telling.
The call that aged well, for balance — Gareth Soloway, 14 July 2026, arguing that bears would be squeezed near-term. Six weeks later they were. We publish the ones that worked alongside the ones that did not.
https://www.youtube.com/watch?v=Xu2MSbIwf4UThree new markers
- C1 — bitcoin trades above $79,563.71 (Tuesday’s US cash-session high, Binance BTCUSDT) during a US cash session, 13:30–20:00 UTC, on or before Friday 28 August. Written because Tuesday’s high was set at 02:00 UTC and no American session came within 2.10% of it. This tests the time zone, not the level.
- C2 — the Binance top-trader position ratio holds at or above 2.0000 at every daily reading through 28 August 00:00 UTC. Written because the size accounts have just gone to a 30-day high on a down candle; this tests whether they stay there when it is tested.
- C3 — the next difficulty adjustment (height 965,664, estimated early September) prints a cut larger than 2.00%. Current projection −3.08%; the last actual was −1.31%. This tests whether the hashrate decline is real or an artefact of variance inside one epoch.
What settles next
Today, 8:30 a.m. ET: A3, on July core PCE. Tonight: Wednesday’s ETF row, which feeds Z3 and B2. Thursday 8:00 p.m. ET: the Jackson Hole agenda. Friday is the heavy board — X2, Y1, Z2 and B1 all expire, alongside Warsh’s 10:00 a.m. keynote, and this desk will publish a single multi-grade review rather than four separate notes. Monday 31 August: A2 settles on the monthly close, B2 on the final ETF row, and B3 on Strategy’s next 8-K.
Running honestly: of the nine markers on the board this morning, three are failing, three are passing, and three are open. Two of the three failures — X2 and Y1 — share a premise that we have now graded as wrong in print for three consecutive days. That is the point of writing them down.
Disclaimer: This article is journalism and market analysis, not investment advice. Bitcoin and other digital assets are volatile and you can lose the entire amount you put in. Nothing here is a recommendation to buy, sell or hold any asset, and no part of it is tailored to your circumstances. The markers described below are this desk's own published, falsifiable predictions, graded in public precisely because many of them fail. Figures are sourced and timestamped in the text; prices move after publication. Do your own research and consider speaking to a licensed financial adviser before making any investment decision.