45 dips, not 289: what a 15% week has and has not told youBitcoin / U.S. dollarBITSTAMP:BTCUSDharrell.jordonBitcoin closed Saturday 15.4% below its close seven days earlier (Bitstamp BTCUSD daily closes: 75,747 on Nov 7, 64,114 on Nov 14). The two loudest readings of that are "deep value" and "get out". Before picking one, I counted what a week like this has been followed by. The rule: a daily close 14% or more below the close seven days earlier. The data: Bitstamp BTCUSD daily bars, 18 Aug 2011 to 14 Nov 2026. No costs are applied, because nothing here is a trade. Count 1: 289 earlier days met the rule. Count 2: those days sit in 45 separate episodes. An episode starts on a day that meets the rule when no day in the prior 30 did. Saturday started the 46th. What followed, measured from the first day of each of the 45 earlier episodes: 30 days later: higher 24 times, lower 21. Median +3.6%. Range -51.4% to +183.6%. 90 days later: higher 26 times, lower 19. Median +10.1%. The 95% Wilson interval on 24 of 45 is 39% to 67%. That is a coin flip that cannot be told apart from a mild edge or a mild disadvantage, with a spread wide enough to hold both a halving and a near tripling. Two things I take from it. 1. The drop by itself is not the information. It has come before both outcomes about equally often. 2. Sample size. A study of "buy after a 14% week" that treats each of the 289 days as its own observation is reporting about 6.4 times the sample it has, because those days overlap and move together. 45 is the honest count, and it spans very different market eras. How to check it yourself: on the daily chart compute close / close - 1, mark the days at or below -0.14, group marks that fall within 30 days of each other, and compare the close 30 bars after the first day of each group with that day's close. This describes past data. It is not a forecast and it is not advice.