See my latest signals table for details on trader positioning in this and other markets. Other signals taking effect on Monday's open: gold and the 30-year Treasury bond both go to cash.
My new S&P 500 setup is unusual for a couple of reasons. First, unlike all my other setups, it is based on a single group of traders - the commercial hedgers, sometimes known as the "smart money." My other setups use two (and in one case, three) groups of traders to arrive at signals. If all the traders don't agree, those setups remain in cash.
In contrast, my new S&P 500 setup is always either long or short. It turned out that after looking at a few million potential setups, the most statistically robust one was based on the commercial hedgers alone. (My previous SPX setup relied on both the commercial hedgers and the small traders.)
In contrast, my new S&P 500 setup is always either long or short. It turned out that after looking at a few million potential setups, the most statistically robust one was based on the commercial hedgers alone. (My previous SPX setup relied on both the commercial hedgers and the small traders.)
Second, my new S&P 500 setup has no trade delay, meaning the signals based on Friday's COT report get executed on the following week's open of trading. Most of my other setups include at least one component signal with a trade delay of one or more weeks. For example, my previous SPX setup used a three-week trade delay.
But after trading my COTs Timer strategy for nearly five years, I recently discovered that there was a moderate and statistically significant negative correlation between the real-time profits of a setup and how many weeks of trade delay it was based on.
I can't find anything in the data that would explain this. It's actually kind of strange. The fact is setups with longer trade delays are often the most statistically reliable in backtesting. But this is why it's important to take a mechanical trading system out for a spin on the road and see what happens. In my new round of backtesting, I am limiting my setups to those with zero to two weeks of trade delays.
I picked two weeks as my max because of my natural gas trading setup, which has achieved a 106-percent profit since I started trading it in mid-2009 while gas prices exploded and then collapsed in the same period. My gas setup uses signals with one- and two-week trade delays. You can't argue with that kind of success.
I picked two weeks as my max because of my natural gas trading setup, which has achieved a 106-percent profit since I started trading it in mid-2009 while gas prices exploded and then collapsed in the same period. My gas setup uses signals with one- and two-week trade delays. You can't argue with that kind of success.
Special thanks go to Dave, a highly gifted and very generous reader who developed an amazing app that I used to do my first round of analysis of several million potential setups.
I then use a few Excel spreadsheets I've laboriously developed over the years to re-analyze the best of those, check other setups I come up with through other processes and run tests of robustness, such as Monte Carlo testing and one of my favourite testing procedures, walk-around testing, which compares how the setup performs against "neighbouring" setups with slightly varied parameter values.
See my FAQs page for more details on my backtesting process.
I then use a few Excel spreadsheets I've laboriously developed over the years to re-analyze the best of those, check other setups I come up with through other processes and run tests of robustness, such as Monte Carlo testing and one of my favourite testing procedures, walk-around testing, which compares how the setup performs against "neighbouring" setups with slightly varied parameter values.
See my FAQs page for more details on my backtesting process.
I'll post details on my new setup on my backtesting results table soon and a sample spreadsheet for downloading. Stand by for new setups soon for copper, gold and U.S. financials. Good luck this week.
No comments:
Post a Comment