Executive summary

The market is transitioning, and the honest reading is that a corrective advance is closer to its end than its beginning. Confidence is moderate rather than high, and the reason is stated plainly below: the evidence on who bought this rally is strong, the evidence on what pattern it belongs to is a judgement rather than a measurement.

Bitcoin has risen more than a third from the July low and more than a quarter from last Monday. The striking feature is not the size of the move but its financing. Open interest was at its highest point of the entire advance on the very first bar, and fell throughout. This was not new money arriving with leverage. It was old money leaving — shorts buying back what they had sold, and a persistent spot bid underneath them. That is a rally with a natural end point, because every buyer a squeeze produces is a seller who has just stopped selling.

What the data is telling us

The financing. From the 62,679 low on 17 August to the 79,500 high this morning, open interest fell from 111,234 to 107,853. The peak was the first bar. The trough came mid-ramp on the 19th. Positions closed on the way up, not down — and the taker flow ran 60.7 per cent to the buy side, 32,800 units bought against 21,201 sold. Falling open interest alongside dominant buy-side flow is consistent with shorts closing, and it is the clearest mechanical evidence a squeeze leaves behind. It is not proof: aggregate open interest cannot label an individual transaction, and a long taking profit reduces it just as a short covering does. What the buy-side skew establishes is the direction of the closing on average, not bar by bar.

The timing. Open interest fell 4.33 per cent on the 17th alone — the day the move began — and barely moved afterwards. The squeeze was spent inside twenty-four hours. Everything since has been carried by spot: roughly 3.8 billion dollars traded on the Coinbase book across the advance, of which about 620 million was net buying at market. Seven bars in that window carry a zero on one side of the volume split and are excluded as unreliable; the remainder reconciles to total traded volume to within a third of a percentage point.

A note on where these numbers come from. The price, volume and premium figures are Coinbase spot. The open-interest and basis figures are not — spot markets have no open interest — and come from the perpetual futures market. Reading the two together is the standard way to separate cash demand from leveraged demand, and it is what makes the argument above possible. It also means the comparison is across venues rather than within one, and any reader who wants to reproduce it should know that before they start.

The venue. The Coinbase premium ran above its own median on 74 per cent of the bars and sat around the 76th percentile of its history. Perpetual futures basis sat near the 24th. American spot demand was consistently strong; leveraged demand was consistently absent. Whatever this rally is, it was not built on borrowed money.

And a figure that resists the enthusiasm. Cumulative delta improved from −25,346 to −15,327 across the whole advance. It recovered roughly two-fifths of the selling that preceded it, and remains firmly negative. The market has bought back some of what it sold. It has not bought back most of it.

The structure before it. The advance is retracing a decline of 30 per cent from the May high, itself part of a fall of 54 per cent from the October top. Price closed below its 200-day average for 101 consecutive sessions before yesterday. The rally has now recovered 77.6 per cent of the May-to-July leg and 28.4 per cent of the whole decline.

Structurally, this behaves as a completing correction — the third leg of a zigzag, or the closing section of a more complex sideways structure. Both readings share the same consequence: what is unfolding is the market finishing something, not starting it.

A historical check that supports the caution. Every previous occasion where price poked above a declining 200-day average while that average sat above the shorter ones was pulled out of eleven and a half years of daily data — twelve episodes — and examined blind, with dates and outcomes hidden so the structure could be judged on its shape alone. Nine of the eleven resolved cases were not impulsive. The base case for this configuration is a corrective advance.

Two cautions belong with that finding, and they are not footnotes. Eleven cases is a small sample. And in that same record, corrective advances still finished higher six times in nine. Corrective does not mean falling.

Key drivers

• The forced buying that started it is largely exhausted. The whole of the open-interest contraction happened on the first day. Whatever short position was being closed has mostly been closed, and that particular source of demand cannot repeat.
• Spot, not leverage, has carried it since. That makes the advance more durable than a leveraged move — there is no crowded long position to be flushed — but it also means there is no leveraged fuel left to extend it.
• The rally stalled at a level that had held before. This morning's high stopped twenty-three points beneath the April swing high, on volume more than double the session median, with momentum at its highest reading of the day. A climax tag of old resistance, rejected.
• The structure is corrective on both the live reading and the historical base rate. The pattern in force is a completing third leg; the historical record of this exact configuration is nine-in-eleven non-impulsive.
• And the level where the argument ends is close. A full retracement of the May-to-July decline sits a little over seven per cent above the current price, in a band where five independent measures converge. Above it, the description “correction” stops applying, and something more serious than a squeeze is under way.

Risks to this view

What would break it. A daily close above the convergence band in the low eighty-thousands. That is a complete retracement of the leg being corrected, and a correction cannot retrace everything it corrects and remain one. Above there, the reading changes from “completing a rebound” to “reversing a trend”, and it changes regardless of how the advance was financed.

What would break it sooner. Open interest expanding rapidly while futures basis climbs out of its low percentiles. That would mean leverage has taken over from spot and a different, faster market has arrived — one which can extend much further than a squeeze, and unwind much harder.

What would confirm it. Loss of the shelf immediately beneath the current price, where the value area, the session volume-weighted average and the short-term average all sit within a few hundred points of each other. Below that the ladder thins markedly and the market has little to lean on until several thousand points lower.

And the honest weakness in our own case. The structural claim rests on a judgement about pattern, not a measurement. The historical base rate that supports it comes from eleven observations. Neither is the kind of evidence that should carry a position on its own, and neither is presented as such here.

The View From The Edge

There is a particular kind of rally that feels stronger than it is, and this has been one of them.

A quarter added to the price in five sessions is the sort of move that changes minds. It ended a run of a hundred and one closes beneath the 200-day average. It reclaimed levels that had looked distant a week ago. And yet the mechanism underneath it was mostly people covering a losing bet, followed by patient spot buying into a futures market that never showed up.

That is not a criticism of the move. Squeezes are real, and the price they produce is real money changing hands. But a squeeze is a finite event. It has a fixed amount of fuel — the size of the short position — and when that is exhausted, the bid it created simply stops. The tell is the timing: the entire open-interest contraction occurred on the first day. The spot bid is still present — that much the data shows plainly — but it is now working alone, without the forced buying that gave the move its initial violence.

The structure agrees with the flow. What preceded this advance was a decline of 54 per cent, and what has followed it looks like the closing act of a correction of that decline rather than the opening act of anything new. The blind test of every comparable episode in eleven years of daily history returned nine non-impulsive structures out of eleven, which is as close to a base rate as this configuration allows.

None of which makes this a market to sell. Correction is a description of shape, not a forecast of direction, and the same record that says these advances are usually corrective also says they usually keep going for a while. Six of nine finished higher. The mistake would be to hear “corrective” and reach for a short.

The disciplined position is the uncomfortable one. This is not the moment to sell into strength, and it is a poor moment to be adding. The advance has stopped at a level that mattered, on volume that suggests exhaustion rather than accumulation, financed by a source of demand that has mostly run its course. What follows is either a pause before the last push towards the low eighty-thousands, or the high of the whole rebound. The market will say which within a few sessions, and the level that separates them is close enough to watch rather than predict.

Educational only — not financial advice. Levels are areas of interest, not trade instructions.