Simulated Losses, Exchange-App Activity, and Bitcoin Buy-and-Hold: A Methodological Re-Examination of BIS Working Paper No. 1049

Bank for International Settlements (BIS) Working Paper No. 1049 (Auer, Cornelli, Doerr, Frost and Gambacorta, 2022; revised July 2023) builds a Sensor Tower panel of retail crypto-exchange app downloads and daily active users (DAUs) for 95 countries from August 2015 to June 2022. Price-led app entry is well documented. The BIS publication page summary states that around three-quarters of users "lost money on their Bitcoin investments", and the PDF introduction (printed p. 3) says "over 75%" and refers to cryptocurrencies. Those figures come from simulations that time Bitcoin purchases to downloads and mark a $100-per-month rule in 2022, after the late-2021 peak. Using the public Group of Twenty (G20) supplement and three public Bitcoin price series, I find buy-and-hold returns of about 69–71× (compound annual growth rate, CAGR, about 84–85%) over the authors' window. Download-weighted dollar-cost averaging (DCA) approximations can reach the paper's 81% loss rate; switching from monthly highs to averages changes that share only slightly. The G20 mean-DAU series has a power-law slope on Bitcoin age of about 7.8 (Bitcoin price-age about 7.3), with China and Japan as regulatory outliers. A slope-consumer price index (CPI) link using World Bank inflation is positive in the full G20 set and disappears without Argentina and Turkey. The entry evidence stands. The three-quarters figure measures a download-timed purchase rule valued in the 2022 drawdown. It is a different quantity from the sample-wide Bitcoin return.