Position 2026-08-05: long (+0.20, confidence 0.42)

Our stance for 5 August 2026 is a small long tilt in XBTUSD — a modest fraction of full size, held with low conviction. Bitcoin closed 4 August at $64,054, up 0.93% on the day but essentially flat over both the last week (+0.30%) and the last month (+0.71%), and still roughly 49% below the $126,200 all-time high. That is not a market delivering a trend; it is a market grinding sideways in the low-$60Ks with 30-day realized volatility around 29% annualized. The case for leaning long here is not momentum — it is that most of the non-price evidence has quietly turned constructive while price itself has not. The macro backdrop is the clearest tailwind. The dollar index has fallen 1.49% over seven days to 99.87, and BTC's trailing 60-day co-movement with DXY has been solidly negative (a descriptive relationship, not a forecast). Equities are running hot — the S&P 500 +4.14% and the Nasdaq +6.87% over the week — with VIX down 9.4% to 16.5 and high-yield credit spreads tight at 2.78 and narrowing. Prediction markets price the September FOMC as a coin-flip between two adjacent target-rate ranges, so no policy shock is being discounted. The one discordant note is bond volatility: the MOVE index is up 18.6% over 30 days, and the 2s10s spread has steepened sharply to 0.43. Rates markets are less calm than equity markets, which is part of why we are sizing this small. Spot ETF flows have stabilized after a rough late July. The latest aggregate print was +$170.1M, the trailing week +$120.2M and the trailing month +$415.4M against $51.5B of cumulative net creations. Just as important, the most recent per-issuer day was broad-based rather than a single-desk artifact: eleven issuers reporting, +$58.7M net, led by FBTC at $33.4M with several smaller sponsors also positive. On-chain, the picture is one of a cheap, sticky holder base — MVRV at 1.19 against a realized-price cost basis of $53,225, 60.5% of supply unmoved for over a year, and a median coin sitting almost exactly at its own cost basis. Roughly 52% of supply is underwater, which caps upside enthusiasm but also means the marginal seller is realizing a loss rather than taking a gain. Positioning and sentiment both argue that the froth has already been paid for. Perpetual funding is close to flat across venues — slightly positive on Binance and Bybit, hovering near zero on BitMEX and Deribit — so nobody is being charged much to be long. Deribit's DVOL has slipped to 34.0 from a 35.8 weekly average, the 25-delta skew premium has eased from its recent average, and the 24-hour liquidation mix was near-balanced (long/short notional ratio 1.08). Meanwhile the Fear & Greed index sits at 25 with retail search interest near the floor, and exchange long/short ratios on both Binance and Bybit have fallen below their weekly averages — crowded longs have already been shaken out, not built up. The honest counterweight is price structure itself, and it is bearish. A 49% drawdown, a flat 30-day tape, front-tenor gamma and open interest clustered right on top of spot around $64K–$65K, and a Kalshi year-end implied median of $62,500 — below today's price — describe a market with no upward structure and a public consensus that sees no year-end gain from here. A typical systematic or ML approach would tend to net out close to neutral in this configuration, with trend components finding nothing to trade and risk controls deferring to the drawdown. We share the caution but think a flat stance slightly under-weights the asymmetry: cheap valuation versus cost basis, uncrowded leverage, easing financial conditions and outright fear in sentiment are the conditions under which a small long is paid, even when the chart is not yet cooperating. Hence a modest tilt rather than a full-size expression.