Position 2026-08-04: short (-0.20, confidence 0.40)

Our view for the session is a small net short tilt in XBTUSD, held with low conviction. Bitcoin closed at $63,467 after an inside-ish day (high $64,018, low $62,237), essentially unchanged on the session, down about 0.4% over the past week and up roughly 0.6% over 30 days. That is a market going nowhere at a level roughly 50% below the $126,200 all-time high, with 30-day realized volatility compressed near 29% annualized. The lean is bearish because the demand side of the tape is deteriorating faster than price is; it is small because nothing in the data argues for an imminent break. The clearest signal is institutional flow. The most recent daily US spot-ETF print was a net outflow of roughly $265 million, and the issuer breakdown shows it was broad-based rather than a single-fund artifact: IBIT -$123M, FBTC -$55M, GBTC -$53M, BITB -$18M and ARKB -$18M, with no issuer posting a creation. Over the trailing week net flows are slightly negative (about -$62M) against a still-positive 30-day total near +$467M, so this is a stalling bid rather than a rout — but it is the marginal buyer that carried this cycle stepping back. Corroborating evidence sits alongside it: the Coinbase premium has been persistently negative (about -0.09 latest, -0.10 on a 7-day average), pointing to soft US spot demand, and dry powder is shrinking, with USDC supply down roughly $684M on the week and USDT down about $4.0B over 30 days. Macro is the second pillar, and it is bearish mainly by omission. Equities put in a strong week — S&P 500 +2.5%, Nasdaq +3.9% — while VIX fell 15% to 15.9 and the dollar index dropped 1.5%. On its trailing 60-day co-movement with the Nasdaq and its negative dollar beta, Bitcoin should have participated in that tape; instead it went sideways to slightly lower. Underneath the calm equity surface, the bond market is less comfortable: the MOVE index is up 23% over 30 days, high-yield credit spreads have widened about 3% over the same window, the 2s10s curve has steepened sharply, and 5y5y forward inflation expectations have crept up to 2.31%. Kalshi's next-FOMC ladder prices essentially no chance of a target rate below 3.50% in September, so there is no near-term easing impulse to lean on for a liquidity-sensitive asset. The supporting reads all point the same direction, modestly. Options are paying up for downside — 25-delta skew at 5.73 versus a 5.15 weekly average and the put/call ratio at 0.52 above its 0.50 30-day mean — while dated-futures basis slope has flipped negative against a positive weekly average, a sign that carry demand is thinning even as retail long/short ratios sit crowded long (Binance 1.94, Bybit 1.59) and aggregate open interest slipped. On-chain, 56.6% of supply is underwater and the median UTXO sits fractionally below its cost basis, with a heavy 6-12 month age band that bought higher; that is overhead supply into any rally, even though a $53,235 realized price and 60.5% of supply older than a year keep the long-term base intact. Sentiment, by contrast, gives us nothing. Fear & Greed at 28 is virtually identical to its 7-day average, Google search interest is flat near its weekly mean, and the news tape is idiosyncratic rather than market-moving. A typical systematic or ML approach would tend to sit close to flat in a tape like this, with trend and mean-reversion components largely offsetting and low realized volatility inviting size; we share that caution but think the combination of a broad-based ETF redemption day, a negative spot premium, shrinking stablecoin float and a failure to participate in a genuine risk-on week justifies a small tilt to the downside rather than none at all. The low conviction is deliberate: with options gamma and open interest clustered between $62,000 and $65,000 and the Kalshi year-end implied median at $62,500, the market itself is pricing drift, not a decisive move.