For Two Months Money Left the US Bitcoin ETFs. Last Week It Came Back Hard. The Price Has Not Broken Out.

By Kresmion Research, July 22, 2026

For most of the past two months, the story in the US spot Bitcoin ETFs was money leaving. Over the last sixty trading days, 27 of 39 flow days were net outflows, and the typical week was about a 268 million dollar exit. On July 13 the funds lost about 478 million dollars in a single session. Then it turned. In the seven days to July 21 the same funds took in about 684 million dollars net, a top-decile week, with fewer than one in ten trailing weeks since late April stronger. The two most recent sessions did most of it, about 178 million dollars on July 20 and about 197 million on July 21, the second of those the third-biggest single inflow day in two months. Kresmion tracks these creation and redemption flows fund by fund, and how to read spot Bitcoin ETF flows walks through the mechanics.

Bitcoin recovered with the flow. It is back near 66,000 dollars, up about 2 percent on the week and roughly 5 percent off its July 13 low, which was the same day as that big outflow. What it has not done is break out. The price sits at the top of a range it has held in the low to mid 60,000s for weeks, not through it, and the leveraged market is not pricing a move higher. The demand regime flipped, the price bounced back to the top of its range, and positioning still looks unconvinced.

The flow regime flipped

Start with the raw series, because the base rate is what makes the last week mean something. Kresmion aggregates daily creation and redemption flows across the nine US spot Bitcoin ETFs. Over the past sixty trading days, 27 of 39 were net outflow days, and the median trailing week was about a 268 million dollar outflow. Money leaving these funds had been the normal state, with single outflow days worse than 700 million dollars in May and June.

Against that, the week to July 21 stands out. The net was about 684 million dollars in, which sits around the 92nd percentile of rolling seven-day sums over the two and a half months of flow data since late April. The single inflow of about 197 million on July 21 was the third-largest day in two months, and it came right after about 178 million the day before, so two heavy inflow days landed back to back. Independent reporting frames the same shift from the other side: BlackRock's IBIT led the complex to its first daily inflow in weeks in mid-July, snapping a multi-week outflow streak, and by July 20 the funds were on a multi-day inflow run while the price worked against resistance in the mid 60,000s. The turn is real and it is one-directional. It is also narrow: IBIT alone took in about 531 million of the 684 million while Grayscale's GBTC lost about 45 million, so one product is carrying the flow.

The price came back, but not through the range

Bitcoin did respond. Kresmion's derivatives feed marks it near 62,700 dollars on July 13, the day of the big outflow, and near 66,200 dollars now, a gain of about 5 percent off that low and about 2 percent over the past week. So this is not a case of demand arriving and the price ignoring it. The two moved together, and the single biggest inflow day, July 21, was also the biggest up day.

The part that has not happened is a breakout. Near 66,000 dollars, Bitcoin is at the top of a band it has traded in for weeks, roughly the low to mid 60,000s. The recovery retraced the July 13 drop and carried the price back to the ceiling of its range, which is a bounce off a washed-out low as much as a fresh advance. The distinction matters, because a range high on renewed buying is exactly the level where a market either breaks out or gets sold, and nothing in the price alone tells you which.

The leveraged market is not leaning in

If the flow reversal were the start of a durable move, positioning would usually start to lean with it. So far it is not. Kresmion's cross-exchange funding surface has Bitcoin perpetual funding close to flat, mildly positive on average and slightly negative on some venues, which is not the reading you get when a crowd is chasing price with borrowed money. On OKX, recent forced liquidations were mostly long positions being closed into the drift rather than shorts being squeezed, and that is a single-venue sample that should be read as a proxy, not a market-wide total. Longs leading the liquidations while the price grinds is the ordinary pattern, so read it as the absence of a short squeeze rather than a signal on its own. And on Polymarket, a market on whether Bitcoin dips to 62,500 dollars this month still carries about a one in three chance even after fading over the week as the price held. Three reads on positioning, from exchange funding, from realized liquidations, and from a prediction-market crowd, and none is pricing conviction in a breakout. The wrapper is being bought. The people using leverage are not yet paying up to be long.

The evidence that cuts against a skeptical read

Calling this a bounce that positioning distrusts owes the other side its due, and there are real points on it.

The first is that the flow and the price genuinely moved together, which is what a working demand signal looks like. The outflow low on July 13 was the price low, and the two biggest inflow days coincided with the strongest up days. If ETF demand is doing the lifting, a further leg would not be surprising, and the broad liquidity picture supports it: Kresmion's macro regime model reads Neutral with its net-liquidity input a clear positive. A steady bid into a supportive liquidity setting is not a fragile setup.

The second is a matter of scale. About 684 million dollars over a week is a small sum next to a market worth about 1.3 trillion dollars, and a modest share of the tens of billions of dollars that change hands in Bitcoin spot and derivatives trading on a busy day. On that scale, a recovery of about 5 percent off the low is a reasonable response to demand of that size. The inflows are real, the size is modest, and the price did roughly what a move of that size tends to do, which removes any need to call the price action a puzzle.

The third is timing. A one-week reversal, even a top-decile one, is a short window against a two-month regime, and flows into these funds have swung between large outflow and large inflow days all year. A single strong week is a change in direction, not yet a trend, and it would not take much to swing back.

What would confirm or break the read

The test is close and observable. If the inflows keep printing and Bitcoin breaks up and out of the range it has held, the flow reversal was the leading edge of a real turn in demand and the price was following it. If instead the inflows stay strong and the price stalls at the top of the range or slips back, that pattern points the other way: it would mean the steady ETF bid is being met by an equally steady seller, someone large distributing into the demand, which is what a capped price on persistent buying tends to reveal. The near catalyst is the Federal Reserve decision on July 29, which sets the rates picture the whole trade sits inside. Kresmion's flow, derivatives, and prediction-market feeds update daily, so the resolution shows up on its own.

Frequently asked questions

Did Bitcoin ETF flows really reverse?

Yes. After a stretch where 27 of the last 39 trading days were net outflows and the typical week was about a 268 million dollar exit, US spot Bitcoin ETFs took in about 684 million dollars in the week to July 21. That sits around the 92nd percentile of trailing weeks in Kresmion's flow history since late April, and the July 21 inflow of about 197 million was the third-largest single day in two months, right after about 178 million the day before. This describes the flow data, it is not a forecast.

If demand reversed, why is the price only back to its range high?

Because the flow, while a top-decile week, is small relative to the market. About 684 million dollars over a week is a small sum next to a market worth about 1.3 trillion dollars and a modest share of daily Bitcoin trading. A recovery of about 5 percent off the July 13 low, back to the top of the range, is a reasonable response to demand of that size. The open question is whether it carries through the range or gets sold there.

What is the leveraged market saying?

It is not confirming a breakout. Cross-exchange perpetual funding on Bitcoin is close to flat, recent forced liquidations on OKX were mostly long positions being closed rather than shorts being squeezed, and a Polymarket market on a July dip to 62,500 dollars still carries about a one in three chance. Those are three reads on how traders are positioned, and none is pricing strong conviction in a move higher.

What is the single thing to watch next?

Whether continued inflows push Bitcoin up and out of its range. If they do, the flow reversal was leading the price. If the inflows stay strong and the price stalls at the highs or slips back, it suggests a large seller is meeting the ETF bid, which a capped price on heavy buying tends to reveal. The Federal Reserve decision on July 29 is the near catalyst.