As readers know, I have been re-building in a lot of my models and indicators in recent months. I used to have them built in a sellside system, but I have been recreating them in code since that system went down for good (RIP Locus!). I share them often in my weekly notes, and in the chatroom. I have since created summary reports for each of: Positioning, Market Models, Technical Indicators and CTA Trigger Levels. On a going forward basis, whenever possible, I will send a late Friday note with these summary tables once the CFTC positioning report is available. Hopefully this gives you something to chew on over the weekend.
I will explain what each of them are in this note as an introduction. On most weeks, it will be the tables without anything written by me.
I will continue to focus on crypto as a separate publication because those models use other data than market variables.
Finally, to make this fun and most useful for everyone, clients can comment on the note and ask for details on anything in the tables, and I will include them at the end of the next Sunday weekly note.
Positioning Summary. This table is broken down into 3 sections: Large Specs, Small Specs and Commercials. We rank their positioning by standard deviations, and also look at their weekly changes. A notable example is AUDUSD at the moment where large specs are the most long in recent years as a percentage of open interest. We can also see from the breakdown that small specs are also very long.
Models Summary. Here are a number of market models across various asset classes. I call these market models because they are based on market variables, as opposed to fundamental variables. The former can be equity prices, interest rates, mortgage rates, commodities, FX and such. The latter can be P/E ratios, economic data or other dynamics that are not in these models. Some of the models have very high R^2, and others do not. Naturally, we should weight those with the highest statistical significance higher than the others. They are also not timing tools. For example, the models have flagged NKY and EWY (Kospi ETF) as very rich for months, but the dislocations have not closed. Shorting them would have been a big loser, so it requires something changing, such as chart patterns, options flows, positioning or market sentiment for them to turn. As such, we do not trade blindly off the models but they are a valuable input into the process.
Copper Miners stand out as very rich to the model, with a high R^2 of 89% and a 2.6 Z-Score dislocation. For this reason, I have favored Copper itself over the miners. We can see that the model has tracked well for 10 years, and recently spot has shot up violently while the model value has not. Either this is a regime change and the model will eventually catch up, or it is calling BS on this price. The second chart shows the dislocation in % terms over time with horizonal Z-Score levels to quantify it.
CTA Signals. I track each asset relative to their distance from what I call CTA trigger levels. There are two trigger levels. There are lots of ways to track CTAs and by no means is this the only way. I have found these work well for me
Using 10y as an example. It had crossed above the CTA level which meant they were sellers. However, before long we crossed below in spite of better economic data as we were in risk off mode with stocks. You can see that once we triggered here, the rally and move lower in yields was violent despite no real fundamental reason for it. That is why it is so important to track CTA levels even if it is never perfect or comprehensively the only way they trade
Technical Indicators Summary. These are my favorite technical indicators that I have used for 15 years in my trading. Unfortunately, I need to put the paywall here as these indicators are proprietary and I must guard my secrets.