A large crypto account posted that the US Treasury had just bought back $6 billion of its own debt, and that this was bullish for Bitcoin. The post was widely shared. The figure in it was correct.
The question worth asking is not whether the number is right but whether the conclusion attached to it has ever been true. Treasury buybacks are not rare, they are scheduled, and there are enough of them now to check.
What a buyback operation actually is
Treasury's buyback program restarted in 2024. In an operation, Treasury offers to repurchase a defined bucket of outstanding securities, dealers offer bonds for sale, and Treasury accepts up to a maximum it has published in advance. Most operations are Liquidity Support, which run on a quarterly schedule and are small. A minority are Cash Management, tied to tax-date cash balances, and those are where the larger numbers appear.
This matters for the claim, because it means the headline dollar figure is usually not news. Across the 161 operations tested here, 103 accepted exactly the maximum Treasury had already announced. The median operation was $2.0bn.
On 1 October Treasury said in advance it would buy at most $6bn. Dealers offered $46.4bn. Treasury accepted $6bn. The only number in that sequence that could have surprised anybody is the one nobody quoted.
The base rate
Bitcoin closed higher on 46.0% of the 161 buyback days, against 49.9% of every day in the same window.
That is a gap of 3.9 points. At 161 observations the margin of error on a hit rate is 3.9 points, and the 95% interval runs from 38.2% to 53.7%. It contains 50% with room on both sides. A permutation test puts the odds of seeing a gap at least this large by chance alone at roughly one in three.
So the finding is a null. Buyback days look like ordinary days.
The same answer turns up at every horizon. The day after an operation, which is the window that actually sits after the result is published, Bitcoin rose 47.2% of the time against 50.2% for all days. Five days out, 46.9% against 51.6%. Three windows, three nulls, all of them straddling the baseline.
The cut that looks like an exception
There is one slice of the data that leans the other way, and it is the slice closest to the claim being made: operations of $6bn or more, the size of the one that prompted the original post. Bitcoin rose on 12 of those 21 days, or 57.1%.
That is worth printing rather than quietly omitting, because leaving it out would be exactly the kind of selective reporting this study exists to push back on. It is also not evidence of anything. Twenty-one observations put the interval at 36% to 79%. Roughly a quarter of randomly chosen 21-day samples would do as well or better. The effect does not persist either: by the next day the average return in that subgroup is negative.
If large operations moved Bitcoin, the size of an operation should carry some information about the size of the day's move. It carries almost none. And it could not be measured cleanly even if it did, because the amount accepted is capped from above by a figure published days earlier, so most of its variation is Treasury's calendar rather than the market's response.
Why the reflex is so appealing
The argument usually runs through liquidity: Treasury pays cash to dealers, dealers hold cash, cash finds risk assets. It is a tidy story and it is not obviously wrong in direction. It is simply too small and too scheduled to show up.
A $6bn operation is a rounding error against the Treasury market, it is announced before it happens, and it swaps one asset for another at a price set in a competitive auction. Buying back debt is not the same as creating money. It is not quantitative easing, and the Federal Reserve's own open-market operations, which some commentary confuses with these, are a different program entirely with different purposes.
None of which proves there is no effect. It explains why an effect that may exist in theory does not survive contact with 161 observations.
What would change the answer
A much larger operation, an unscheduled one, or a change in the program's size relative to issuance would all be reasons to measure this again. So would a run of operations where the accepted amount came in well under the announced cap, which would make accepted size informative rather than mechanical.
Until one of those happens, the honest answer to whether a Treasury buyback is bullish for Bitcoin is that the record does not show it, in either direction, and that the dollar figure in the headline was on the calendar before it was news.