Urban Carmel
@ukarlewitz
Former: UBS Securities, Asia Ex-Pat, McKinsey, Mayor of Mill Valley, Finance Twitter. Fan of causal relationships, behavioral finance and the Quad Dipsea.
This sounds bad. It isn't. Try enough parameters/timeframes, something like this will turn up. When you have a sample of only 8 (two overlap) over the past 54 years and use oddly precise parameters like 11 (why not 10?) then you have an example of post-hoc data mining with low predictive value
Today's bounce was probable (scroll up). I use RSI(5) <30 (top panel) and under the lower Bollinger (20,2) (bottom panel) to determine short-term oversold - see vertical lines in past 2 yrs. This is doesn't say anything about a low being in, or not. It's just the proverbial oversold bounce
When June + July are neg the rest of the year loses an avg of -1.8% and is neg 8/13 (61%). This sounds bad. It's not. The standard error (3.6%) when the standard deviation is so wide means that the avg return for the rest of the year is indistinguishable from zero: no edge. Oddstats knows this
Weekly: 20-wma usually (not always) gets toughed when MACD flips over (vertical lines). Corresponds to the bottom of the blue box in prior post
Motherhood and apple pie: 60' (1st chart): July lows (740) broke today. Lower low + lower high = st downtrend until pattern reverses. Bigger pic: sideways the past 3 months Weekly (2nd chart): 20-wma hasn't been touched since April (arrows). Corresponds with the bottom of blue box
Real retail sales in June +3.1% yoy; ex-gas +2.2%. Sales finally surpassed the high from April 2021
$49.6b inflow into equity MF + ETFs last week preceded by $20b the two prior weeks are banger inflows. I don't recall ever seeing inflows like this (ICI)
Useful context might be that if June ends down it'll be the first time the mid-term pattern has been right this year. It's been wrong all 5 of the first 5 months. A coin toss is more accurate
May 2011 and June 1948 were starts of bear markets (defined as a top to bottom drop >20%)
100% of the outperformance of small caps so far this year happened in the first 14 days of January.
Here's another one. Bottomline, it's still closed. hormuzstraitmonitor.com
Track shipping traffic through the Strait of Hormuz here: datalab.wto.org/Strait-of-Ho...
CMTs never quote A/D stats from the 1960s. Why? Because the SPX more than double while breadth steadily deteriorated, from 1959 through 1974
Last 3 yrs: SPY RSI(5) <20 (vertical lines). n=20. All except one (red) saw at least a one-day relief rally >1%
Right now, some see breadth narrowing and others don't. Leave aside whether that matters soon, a year from now or not all, even the basic facts are not agreed upon. Real-time is always messy like that
Here's one example (of many) of a false positive, from Aug 2013. SPX had not even a single -10% pullback in the next two years, during which it gained more than 30% and the ensuing low was 14% above the signal date.
The A/D line for $SPX peaked in mid-April, 2-months ago. When it diverges from price (like now) this regarded as a bullet proof top indicator. It's not. Here are >20% SPX drops the past 50 yrs and # of mo the A/D peaked before or after it: 2025 -2mo 2022 0mo 2020 0mo 2018 0mo 2011 +2mo cont
Breadth is still either a warning or fine. Many divergences resolve with higher prices, but some don't. So which one is it now? There's no way to know in real-time but in hindsight you will be told it was obvious
Inflation using timely measures of shelter were running under 2% before Iran. Now zeroing in on 3%. From Jeremy Schwartz
We are ten bi-weekly periods into the year. How is 2026 performing relative to Goldman's seasonality chart? Not well. It's batting 50%; a coin toss. A simple bet that the SPX rises in any 2-week period is right 83% of the time. Using seasonality subtracts value from your decision-making
Turnaround: SPX falls >5% in one month and then rallies >10% the following 2 months (like now). Since 1970, n=11. All closed higher by month 6. Risk/reward next 12-months way more than 2:1 positive. Guarantees nothing. From Steve Deppe
SPX is a day away from closing higher 9 weeks in a row. Since 1980, n=4, none a notable top. Please do not run a statistical analysis on a sample of 4 over 45 years
Left: SPX can have positive returns when the AAII spread is +bulls or +bears. Same is true for negative return years. Right: a better way to show this is in a scatter plot. Gemini calculates the r-squared at 0.12, meaning the AAII spread doesn't explain annual returns
More AAII bears than bulls in December 2021. SPX more than 20% lower by June. More AAII bears than bulls in February 2025. SPX almost 20% lower a month later.
June mid-term years have been weak. But in years when SPX is strong through May, June is also strong May mid-term years weak. This one up nearly 5% If it can be great or bad then there's no useful edge
In the last 10 years' bull market, June has been one of the best months of the year, up 9 of last 10. July a perfect 10 of 10 In the 1990's bull market, June was one of the weakest months of the year. July even worse If something can be the best and worst then there's no useful edge