Onchain spot demand for bitcoin stays negative despite a $2.39 billion ETF week

The two series use different units. Farside counts net dollar subscriptions into US spot products by trading day, while CryptoQuant estimates the change in coin demand across the spot market over a rolling month ^[10]. Neither source gives a bitcoin price, so the September dollars cannot be converted into coins and netted against the October 1 reading ^[1][3].

An investor buying an existing ETF share on an exchange generally trades with another holder or a market maker, and the fund need not issue a share or buy a coin ^[8]. Net creations do change what a fund holds. Even those can drift from a same-day spot purchase because of timing, authorized participant inventory and issuer reporting ^[9]. Coins also reach the market from outside the funds: miners receive new ones, older holders decide whether to spend, and traders move inventory onto exchanges ^[12]. According to crypto.news, a fund can add exposure while those holders sell more coins than the new buyers absorb ^[12].

The September money was front-loaded. Its first two sessions made up about $1.71 billion, or 72% of the five-day sum, and each later session was smaller than the one before ^[22][2]. Three funds took about 91% of the opening day ^[23].

The following week depends on when Farside's table was read. The provisional total included $31.7 million for that Friday ^[5]. The later report records October 2 as a $189.9 million inflow ^[13]. Substituting that row lifts the week to about $241.1 million ^[18]. Even the higher figure is about a tenth of the September week ^[19]. Within it, BlackRock's IBIT drew $292 million over four sessions while Fidelity's FBTC had $167.9 million of net outflows, Grayscale's GBTC $54.6 million and Bitwise's BITB $38.6 million ^[6]. FBTC, the third-largest taker on September 21, was being redeemed a week later ^[7][6]. Ethereum funds went from $689.8 million of inflows to $118 million of outflows over the same two weeks ^[14].

One explanation for the gap is lag. The October 1 reading still spans early September, and it improved by 81,000 BTC in seven days ^[20]. At that pace it would reach zero in about nine days ^[25]. A rolling window also sheds old days, so that projection is a straight line and not a forecast. Another explanation is concentration. On this reading, late September came from a few large allocators in three products ^[23], and the alternating daily flows since then are the normal state. A third is measurement. CryptoQuant's figure is built from onchain supply behavior, and it is neither a census of buyers nor a reading of order-book bids ^[11].

I think concentration fits the completed rows best, because Farside shows a strong burst followed by daily flows that switch sign ^{[2][5][13][17]}. The counter-case is the 81,000 BTC improvement itself. Fund buying working its way through a rolling window would produce exactly that ^[20]. A positive 30-day reading by mid-October, with creations continuing and coin accumulation, exchange demand and price action agreeing, would prove the concentration reading wrong ^[24][25].