Kit Winder, CFA

The bytetrend.io platform offers an MCP that allows you to interrogate 20 years of trend data on 4,000 stocks, plus ETFs, commodities, and crypto. This allows you to plug it into an LLM of your choice, and interrogate the data with natural language. Given the incredible moves today (Nasdaq $QQQ most notably), I asked it to investigate similar episodes from the past and share the strongest conclusions. What history says: These days are rare and come in clusters. There have been 50 days of -4% or worse since 2005, about 1 in 110 sessions, and 34 of them fell in 2008, 2020 and 2022. After any -4% day, another followed within a month 64% of the time; on an ordinary day that chance is 10%. There’s no reliable bounce. The next day was up 29 times out of 50, with a median of +0.3%, much like a normal day (55% up). On the first -4% day of a selloff, the next day was up only 5 times in 11. The headline 12-month return is flattered by 2008 and 2020. Across all 50 days the median 12-month return is +33.5%, against +19.4% for all days. Drops that started within 10% of the high, as today’s did, give +20.8%. That’s an ordinary year, not a buy-the-crash payoff. Only two near-high drops failed. Of 14 with a full year of data, 12 were higher a year later. The exceptions were 4 January 2008 (−36%) and 3 February 2022 (−14%), both the start of bear markets. In 2022 Weekly Local was already 3 going in. Weekly Local separated the good cases from the bad. On all 9 days the Nasdaq fell 4% with Weekly Local at 5 the day before, it was higher 12 months later (median +44%). Those 9 days make up six separate episodes, so it’s a small sample. The low usually wasn’t in yet: four of the six episodes closed lower within three months, by up to 16% (October 2018) and 18% (February 2020). In the recoveries Weekly Local bottomed at 2. In 2022 it fell to 1, and in 2008 to 0. What’s new this time is weaker leadership against the world. In all nine of those cases Weekly CAPR was also 5. Today it’s 3, and the Nasdaq’s CAPR is 5.4% below its 3 June peak, two days before the June drop. No -4% day in the dataset has had Weekly Local 5 with Weekly CAPR at 3 or less. The nearest cases are Brexit day 2016 (+36% a year later), February 2022 (−14%) and April 2025 (+31%). High beta works both ways. After 48 of 49 -4% days, the Nasdaq beat MSCI World over the next 12 months; the exception was February 2022. On an ordinary day that happens 84% of the time. Breadth was already soft. Across the 4,093 ByteTrend stocks, the share of Weekly Local 5s yesterday was 27.5%. That’s in the bottom fifth of readings since 2010, where the median is 39%. Before the harmless shocks of 2016 and February 2018 it was 41–44%. Today’s level is closer to August 2011, February 2022 and April 2025, which had mixed outcomes. Hedges have worked less well since 2021. On these days $GLD (SPDR Gold Shares) rose half the time overall but only 3 times in 12 since 2021, and it fell 3.7% on 5 June. $TLT (iShares 20+ Year Treasury Bond) rose 71% of the time overall and half the time since 2021. $XLP (Consumer Staples Select Sector SPDR) rose 6 times in 45, and bitcoin 10 times in 30. @ByteTree @AtlasPulse @WinderKit