Origins of my strategy
The Origins of My System:
This will likely be too in depth for many but I wanted to put my process out there for those who might be interested.
So how did I create my system for taking risk. It actually was an effort that progressed through about 10 years as I learned of new techniques and thought of new ways to implement the information. Below is a high level overview of the process.
I began by collecting many different economic indicators that I felt might give some indication of upcoming performance. I found that many indicators had a cyclical nature to them and so I looked at these relative to the next year of returns on a bunch of different indices. If there was a strong correlation between the level of the economic indicator and the index return then I kept it as valuable information.
I then studied the indicators that gave the strongest signals and noticed that some of the indices went up when the indicator went down and others moved the opposite direction. I also noticed that for some indicators the direction it was heading (up or down) was an important factor as well.
This gave me the idea that one could profit by adjusting their allocation between these indices based on the economic indicator readings. I went back in time and tried a many different ways of implementing the positions. I found that making a monthly bet of 1/12th of the portfolio gave the best results because it allowed for a lot of positions to be taken and allowed for each position to be held for a full 12 months (equivalent to the performance period in my research).
In going back I found that this system picked the best asset class about 50% of the time and the second best about 25% of the time. To make sure this was not time dependent I looked at the first 15 years and second 15 years (both had similar success rates). Those are great results - active managers are happy if they pick a winner 60% of the time.
Results in the first couple of years of actually running this blog have been in line with expectations although it is still early and there are not enough results to give an accurate picture yet. Below are the data points.
With only 26 bets taken since I started the blog 26.92% were the best, 42.31% were second, 26.92% were third and only 3.85% were fourth. Good results that I would expect to move towards the long term averages in my backward looking research as more results come in.
Going back to 1987, prior to starting the blog, the results for 359 periods were 49.6% first, 27.6% second, 14.5% third and 8.4% fourth.
This will likely be too in depth for many but I wanted to put my process out there for those who might be interested.
So how did I create my system for taking risk. It actually was an effort that progressed through about 10 years as I learned of new techniques and thought of new ways to implement the information. Below is a high level overview of the process.
I began by collecting many different economic indicators that I felt might give some indication of upcoming performance. I found that many indicators had a cyclical nature to them and so I looked at these relative to the next year of returns on a bunch of different indices. If there was a strong correlation between the level of the economic indicator and the index return then I kept it as valuable information.
I then studied the indicators that gave the strongest signals and noticed that some of the indices went up when the indicator went down and others moved the opposite direction. I also noticed that for some indicators the direction it was heading (up or down) was an important factor as well.
This gave me the idea that one could profit by adjusting their allocation between these indices based on the economic indicator readings. I went back in time and tried a many different ways of implementing the positions. I found that making a monthly bet of 1/12th of the portfolio gave the best results because it allowed for a lot of positions to be taken and allowed for each position to be held for a full 12 months (equivalent to the performance period in my research).
In going back I found that this system picked the best asset class about 50% of the time and the second best about 25% of the time. To make sure this was not time dependent I looked at the first 15 years and second 15 years (both had similar success rates). Those are great results - active managers are happy if they pick a winner 60% of the time.
Results in the first couple of years of actually running this blog have been in line with expectations although it is still early and there are not enough results to give an accurate picture yet. Below are the data points.
With only 26 bets taken since I started the blog 26.92% were the best, 42.31% were second, 26.92% were third and only 3.85% were fourth. Good results that I would expect to move towards the long term averages in my backward looking research as more results come in.
Going back to 1987, prior to starting the blog, the results for 359 periods were 49.6% first, 27.6% second, 14.5% third and 8.4% fourth.
Thanks for explaining a little bit more of the magic behind the madness! I'm curious what primary economic indicators you're looking at?
ReplyDeleteThis is a good question and should be easy to answer but it is not. The big ones I look at are changes in spreads (difference in yield on a risky bond vs a government bond), changes in Volatility, the steepness of the yield curve (difference between a 10 yr and 2 yr government bonds), the break even inflation (BEI) rate (a gauge of expected inflation), and currency FX levels (US vs rest of world).
ReplyDeleteSpreads tell how much you are getting paid to take on credit risk
Volatility tells how much uncertainty is currently in the market
Steepness of Yield Curve and BEI tells about future expectations
And Currency rates have a significant impact on Intl returns vs US returns
Not sure if that is too much info but there it is.