Valentine Window
In addition to Valentine’s Day, today is the 15th anniversary of the launch of ITA Wealth Management. Investing evolution is evident since my first blog back in 2008. As you may recall, the stock market was in free-fall that year and it extended through the first quarter of 2009. The first blog was written on a different platform as I knew nothing about WordPress in 2008. When I first started, I did not know what a domain name was. The term blog was a foreign concept. My education began with an excellent book, “WordPress for Dummies.”
In the summer of 2013 a Swedish hacker took down the host server where all my blog posts were stored. The host website did not have proper backups, nor did I so I lost over 1000 blog posts. Needless to say, I switched hosting sites and have been with Inmotion ever since.
What are some of the evolutionary investing changes over these past 15 years? The concept of asset allocation remains much the same. This investing strategy has been around for decades and was made particularly popular when William Bernstein wrote his first book, The Intelligent Asset Allocator. An example of the asset allocation investing model is the Schrodinger portfolio. Schrodinger is a classic asset allocation approach with one difference in that it is managed by a computer algorithm rather than a human. Launched as an experiment, the Schrodinger has become one of my favorite portfolios. One does nothing except save. All the work is done by a computer.
Dual Momentum™ came into existence in 2015 when Gary Antonacci published his book by the same name. McClintock and Pauling are the two ITA portfolios still using this investing model. At one time the Franklin also used this approach. I was disappointed in the results so the Franklin was moved to the new and most recent model, Sector BPI.
In 2020 Steven Bavaria wrote a book titled, “The Income Factory.” Portfolios following this investing model are the Huygens, Curie, and Newton. Only the Huygens is reviewed in detail here at ITA. These three portfolios are populated with Closed-End-Funds (CEFs). When possible, use this model with tax deferred accounts so as to reduce taxes.
Before going much further in describing a bit of ITA history, a lot of credit goes to David Faulkner (Hedgehunter) for his development of the Kipling spreadsheet. The Kipling is used to managed the Dual Momentum portfolios as well as the Relative Strength duo, Einstein and Kepler. Screenshots of the Kipling show up in nearly every blog post.
Another important name is Harry Stevens as he bails us out when Yahoo changes their download codes. Harry updates the ITA file, so critical in making the Kipling work as it should.
A few of you may remember the portfolio tracking software program that went by TLH Spreadsheet. The spreadsheet was permanently benched when Yahoo made a major change in granting access to price data. When the TLH Spreadsheet was no longer viable, I moved to the commercial program, Investment Account Manager software and have been using it ever since. I don’t like being at the mercy of programs that could easily vanish in a flash. Were this to happen, I would most likely revert to two of my favorite investing models and they are – the Copernicus and Schrodinger.
The most recent investment evolutionary example is the development of the Sector BPI Model. Four portfolios are currently operating under this model and they are: Carson, Franklin, Gauss, and Millikan. It is much too early to draw any useful conclusions as to how viable this model will be under various market conditions. Stay in touch if interested. To learn more, do a search of Sector BPI or go to the right-hand column and click on the Category. If you go the Category route you will find the oldest blog posts first. If you go the Search route, the newest posts show up first. I hope the difference is clear.
From the very beginning, ITA emphasized diversity of stocks through the use of no-load mutual funds and later Exchanged Traded Funds (ETFs). In the 1990s and early 2000s I went through my individual stock selection phase only to find the additional work did not produce sufficient results and the risk was too high. As iShares, and later Vanguard, launched ETFs, I moved to these investment vehicles and have not looked back.
Those readers who have been with me for a number of years have picked up the diversification of portfolios in addition to diversification of investment vehicles. While it is not necessary to use every model tracked here at ITA, I recommend using two to three different models as no single model works well in all market conditions. I have my own favorites and they tend to be the ones that are simple and easy to maintain.
A special thank you to the many readers who have visited this site over the past 15 years. The number is well over a million hits.
Lowell Herr
Discover more from ITA Wealth Management
Subscribe to get the latest posts sent to your email.
Congratulations, Lowell. Hope to enjoy your guidance and opinions for many years to come.
Cristian
I also congratulate you. I’ve been a long time ITA fan and have used the Kipling and the old TLH Spreadsheet. In fact, I still use it with some modifications to collect price info.
I found this online:
“Niels Bohr, the Nobel laureate in Physics and father of the atomic model, is quoted as saying, “Prediction is very difficult, especially if it’s about the future!” This quote serves as a warning of the importance of testing a forecasting model out-of-sample. It’s often easy to find a model that fits the past data well — perhaps too well! — but quite another matter to find a model that correctly identifies those features of the past data which will be replicated in the future”
Let keep trying. 🙂
Bob Petersen
Bob P.,
Thank you. As for predictions, I’m skeptical of back-testing, although it is the best we can hope for.
Even live models going forward may work for awhile and then vanish. This is known as the Schwert Rule. Here is the definition.
Schwert Rule of Anomalies: “After they are documented and analyzed in the academic literature, anomalies often seem to disappear, reverse, or attenuate.”
I’m hoping the Sector BPI model does not follow any of these three anomalies. At least I know the Copernicus and Schrodinger will not fall into one of these crevasses.
Lowell
Lowell,
Congrats for hitting this milestone! I have been reading your results and commentary for a few years now. Your work is a testament to seeking new investment systems. All of us here are obviously in that mix too, but you keep not 2 or 3 systems going, but 15! Just amazing.
By the way, there’s book by Valeriy Zakamulin called “Market Timing with Moving Averages” that goes through a systematic exploration of moving average methods. The real keeper, though, is the part on in-sample (backtesting) and out-sample (forward testing) performance testing. Not cheap, but well worth the money just for those sections.
Many fruitful returns!
Craig
Craig,
Thank you for taking the time to follow ITA and for the book recommendation. I was not aware of Zakamulin’s work. Yes, the book is not inexpensive.
Lowell
I will give you the $2 overview.
The testing/validation part of the book is what anyone who is reasonably well-versed in statistics or data science would do. The problem we face is that time series data presents its own challenges with regard to partitioning the data into test and validation. Unlike standard partitioning methods which use either bootstrapping or stratified random selection, you have to ‘choose’ the time windows in such a way that they include a balanced set of up- and down-markets–which may, if you are not careful, introduce a form of data cherry-picking. The windows also have to be of sufficient duration for one to observe the requisite trend following behaviors.
I have simplified this a bit, but the way I check for such things is to use Portfolio Visualizer’s backtesting capabilities for specific time windows, and then I play these forward and average the results of the forward tests. It is pretty close in spirit to what he executes in his book. Unfortunately, Portfolio Visualizer cannot do moving average envelopes, which is one of the most robust trend following systems he surveys.
If you know a little about R/RStudio, he also has published a set of R packages on this web site (https://vzakamulin.weebly.com/the-book.html), and he has summarized the results on the Alpha Architect site (links are in the book webpage I just referenced).
I know this is a bit off the blog topic, but I wanted to relay to you that there are ways to avoid data snooping and data mining in the optimization procedure. Not easy, but do-able.
Again, congrats on your 15th anniversary…here’s to 15 more!
Craig
Craig,
Back in the mid-1980s a former student and I wrote a monthly mutual fund newsletter titled, ITA Trend Analysts. If I recall correctly we use either a 190- or 200-day Exponential Moving Average. Knowing that many investors were following the 200-day simple moving average, we wanted our trading signals to come a few days before the crowd moved.
We tracked the NYSE and if I recall correctly, the DJIA and DJTA. If two of the three moved from below to above, be made our purchases. If two of the three moved from above to below the 200-day EMA, we sold. Our model worked sufficiently well to where we eventually moved to newsletter #11 our of the 87 that were tracked by Hulbert Financial Digest at that time.
I’ve held on to the ITA as it stands for Investment Trend Analysts.
Several years ago a toyed with R software. Not to the point where I could code a back-testing model. There are a few readers of this blog who are or were quite proficient in writing R code.
Lowell
Interesting history, Lowell!
I still use monthly moving averages on my ETFs as a foundational system, as they perform well (still) and have superior downside protection. Like you, though, I have another method based upon relative strength (with an exit based based upon moving averages), and I have also used dual momentum in models. I have found the same problems with DM as you have recently. Nonetheless, I still have it in my quiver, as different models work better in different regimes (markets are somewhat mean-reverting right now, and DM does poorly in these environments).
Another item of note: one of the points determined by this book is that there is a formal relationship between moving averages and momentum methods. Such info is also summarized on the link given above.
As you surmised, the reason why I put the Zakamulin link in my note was because it would seem logical that some of the readers were R-savvy. With that in mind, I am contemplating some R code which will do the back- and forward-testing I desire for other systems which cannot be performed on Portfolio Visualizer.
Best,
Craig
Craig,
The one model that cannot be back-tested is the Sector BPI as Point and Figure (PnF) graphs are subject to the turn around signals determined by the end user. PnF graphs therefore differ from user to user.
Another variable is the frequency of checking the BPI graphs. Results will differ if one uses daily vs. weekly data.
I think it difficult to control all the variables despite the simplicity of the model.
Lowell
PS Yes, this is a little off topic.
Hi Lowell,
Congratulations on 15 years! I want to express my appreciation to you and David as you both have been so generous with your patience, time and knowledge. I have found your site to be a wealth of information that has given me many ideas. It has been a source of steady advice during these past turbulent years. Thanks so much!
Phil
Phil,
Thank you for your kind remarks. I too owe many thanks to David (Hedgehunter) and Harry Stephens for their many contributions.
I always viewed this site as an educational effort to help individuals manage their own accounts. One can save so much money if they will learn to manage their own accounts. This is why we try to show “live” examples that range from the very simple to a few examples that are a tad more complex.
As readers see, complexity does not always turn into superior results.(g)
Lowell