Periodically I write an article that reviews the past few months of articles. Why on Earth would I do this? Primarily for two reasons. One is that many new readers are involved and often they do not go back and look at the past articles. Two is that my articles are rarely tied to anything that is happening in the markets. Generally, they are about experiences I have had as a technical analyst for 45 years; the good, the bad, and the ugly. Hence, they have shelf life (well, certainly in my mind they do). You can click on the headers for a link to the article.
FOR NEW READERS of THIS BLOG: Below are the links to the past Article Summaries. A suggestion was made by a wise reader that many new readers do not go back and read the old blog articles. I can’t blame them; however, this will provide you with an index and short summary for each of them. Currently, about 139 articles.
I wanted to start a series on this subject for quite some time and figured now was the time. This series will step you through all the details of model creation; or at least how I see it and have done it. This introductory article gives a short overview of the process and a detailed outline that will be followed for the series, or at least will try to follow. While I understand not every indicator or measure is available on StockCharts.com, I will try to provide enough information that you can create your own or offer indicators on StockCharts.com that are similar.
This is a review of the performance of various indices during 2017. 2017 was an amazing year for trend followers, buy and hold types, and almost all others except market timers. Market timers probably had a bad year. However, looking at asset classes other than equities, it was quite a volatile and difficult year. We probably won’t see a year like 2017 in a long time.
Okay, so this is number 2 in the Building a Rules-Based Trend Following Model. I tried to come up with a better article naming convention, but nothing seemed better than this boring one. In this is article I introduce you to a very important charting concept called digital and analog measures. Sometimes I refer to the digital ones as binary if they only involve two results such as on/off, up/down, etc. If they involve more than two, I try to call them digital measures. And sometimes I don’t do that consistently. Digital measures are a great way to see just the indicator’s signal and ignore all the analog noise. Analog indicators are the typical ones you see all the time, they show the signals but also everything in between.
Another great concept is introduced in this article, compound measures. While is sounds simple, I show a couple of examples that probably make it look more complex than it is. A compound measure is a digital signal that summarizes what 3 or more other digital measures are saying. In my example I show 3 digitals and the rule is that if any two of them are on, then the compound measure is on. And vice versa for off. This is a great way to combine indicators and offer a little democracy in the process.
Here I introduce the concept of “weight of the evidence.” This term was used by Stan Weinstein in his “The Professional Tape Reader” newsletter in the 1980s. And I stole it from him. I also show a graphic of my weight of the evidence used in the 1980s; all by hand, on graph paper. My weight of the evidence involves price-based measure, breath-based measures, and relative strength measures. These are all weighted based upon their historical contribution to trend following performance with a resulting digital measure that shows their combined values from zero to 100.
I rarely discuss the current markets as StockCharts.com has all the technical analysis experts who do a fantastic job of it. However, I took this opportunity to show where my weight of the evidence was relative to the current market. I show the nine weights of the evidence indicator binaries along with the total weight of the evidence digital measure. If the market warrants, I’ll do this again, but only if it brings value to this series of articles.
One cannot begin to use a technical indicator without understanding its performance over various periods of history. I introduce the performance measures I use along with my evaluation periods for testing these indicators. The goal of indicator evaluation periods is to run it through different market periods and time frames and see if it holds up decently through most of them. You do not want an indicator that only seems to work well through all of your historical data. I also take the opportunity to discuss optimization. This is a very dangerous area for most. If one does not understand the downsides to optimization I recommend you avoid it.
When I wrote the article on the current market a couple of weeks ago, I found that tying what my model was saying as it would be appropriate for this series. Yes, my model experienced a couple of whipsaw trades. This is when you sell based upon your stops, the market bottoms, rises, and when the model says to buy again, you are buying at a price higher than when you sold. Let me state this clearly and boldly. Trend following will have whipsaws – period. You cannot avoid them; learn to live with them. As I like to say, whipsaws are frustration, bear markets are devastation.
I had planned for some time to write about pulling the trigger as a trader. This is one of the biggest failures of many; they love to do the analysis, but when it comes to actually making a trade, they cannot pull the trigger. Sadly, I had this article titled “pulling the trigger.” Unfortunately, or fortunately, the Florida school shooting occurred, and I changed the title to Execution. I’m usually not one for political correctness, but in this case, it was a must.
I take one of my price-based indicators I call Price – Long term and give a full analysis of it. Long term for me isn’t all that long, generally in the 4-8-week time frame. I show the binary for this indicator over a 17-year period, so you can see that it does a reasonable job of picking out the uptrends and downtrends of the market. This is one of the measures that is not available on StockCharts.com, but I show you an indicator and its parameters that is on StockCharts.com that is quite close to it.
A performance table of my price long term indicator is shown along with the following parameters: winning years, trades per year, average return per trade, total return, compounded annual return, compounded annual return while invested, percent of time invested, ulcer index, largest trade loss, maximum drawdown, and ulcer performance index. This table shows my current signal parameter with other parameters above and below the current one. This is to allow me to see if the current parameter used is still the best, or at least as good. I also discuss these performance measures and show which are more important than the others.
Occasionally, I like to take pot shots at some of the Wall Street experts on television. I state that I strongly believe most of them are just analysts and could not produce a profit/loss statement if their life depended on it. The problem with listening to these types is most don’t deal with the hard-right edge of the charts; you know, the area where you have to actually make a decision. They will show you what happened in the past and how it played out. This might serve well as education but is certainly isn’t trading or investing advice that you can act upon. Aftcasting is the opposite of forecasting. I didn’t really have to say that, did I?
Over the years I have stated that I have been a trend follower most of my 45+ years in the market with a few painful deviations into market timing, shooting from the hip, guessing, and a few others I’m too embarrassed to mention. I try to explain the true value of being a trend follower with rules and stops. A methodology that will let you ignore the ramblings of the experts, the newsletter writers, the wire house brochures, and any other sources of market-oriented marketing disguised as advice. Trend following with rules and stop losses will let you sleep, play golf, and relax.
I introduce another of my price-based measure that I call Adaptive Trend. This was a concept developed by the Bloomberg service and they called it ‘Trender.’ Adaptive Trend involves average true range, standard deviation, and exponential smoothing. I offer the details of the formula in text and show a chart of the various ways it can be displayed and used. I explain that it was added to my weight of the evidence when another measure began to cease being of value – a long term interest rate indicator.
Dance with the Trend,