Successful trading pits two diametrically opposed ideas against each other and forces the trader to assimilate them into a cogent plan of action. First of all, everything we know, we know from history. We constantly relate the market’s present status to something we’ve seen in the past whether it’s fundamental, technical or pattern based. Secondly, we’re asked to project the implications of the current market’s structure into the unseen future. This week we’re going to discuss the implications of applying historical knowledge of the interest rate complex towards the deployment of mechanical models designed to take advantage of rising rates.
There are two primary steps to this process. Our first step is through analogous research. Analogous research begins when a present situation is recognizable because it looks or, behaves similarly to a pattern or collection of fundamental data that the trader has already experienced. The simplest version is the gut instinct, “You know, this market just feels like…….” A more quantifiable version falls along the lines of, “The last time these factors were in place, the market did…..” Analog trading is the primary basis for pattern recognition. The more often a certain pattern is seen and its results are predicted the more easily they are recognized and traded in the moment.
Empirically applying analogous research to the current interest rate sector yields the conclusion that monetary policy is far more likely to tighten than ease further. Sixty years of Prime interest data yields some ideas of what we can expect if history is any precursor to the future. First of all, rates have been abnormally low for an abnormally long period of time. The last time rates were steady for this long was 1960-1967. The Federal Reserve’s Prime rate has been at 3.25% since January of 2009. More importantly, it has been in a downward trend since August of 2007. A trader’s primary concern is always the trend.
The trader’s next objective is determining a target. These require both time and prices. Over the last 60 years once interest rates begin to move, they move by 73% on average before their peak or trough is reached. Furthermore, it takes an average of 30.6 months for the move to be completed. The extremes over this period include a move as small as 14% in as little as six months to more than 200% over the course of three and a half years. Throwing out the largest and smallest moves in both time and distance leaves us with an average interest rate move of 64.5% in 28 months. I think this blows a hole in the, “bonds are less volatile,” investment theory.
Anecdotally, there have been many headlines and news releases that should be forewarning us that a change in monetary policy is coming. These include comments from Fed Chairman Ben Bernanke as well as large investment managers. Perhaps the most noteworthy of which is Pimco’s decision to start offering access to hedge funds as it moves away from the bond business. Bill Gross has arguably been the most successful bond investor of all time. He has ridden the 30-year bull market in bonds up to $2 Trillion in assets under management. When the King leaves the table it’s a good clue that the free meal is over.
The case has now been made for what to expect out of interest rates as well as the time frame in which we expect it to happen once it begins to unfold. This led me to updating previously deployed interest rate models whose effectiveness became limited due to the artificial manipulation of rates following the economic meltdown of 2008. The simplest model is based on the DCB Bond system, which was published in Futures Truth in 2000. This model basically doubled its initial investment between 1995 and September 11th, 2001. The original version of this program traded both sides of the market and has fluctuated in and out of their Top 10 Bond systems since its publication nearly 14 years ago. Meanwhile the long only version posted a new equity high last June and has been flat for the last year.
Trading program technology has advanced substantially since then. We are currently developing programs using a neural network and selective data slices from rising interest rate environments. The main benefit is that we can train it on the type of market movement we expect it to see. The downside is that due to the small windows of time that rates actually move, we only have an average of a 28-month window to use for both in and out of sample testing per episode. The end result will be something traded on a shorter timeframe and will require closer monitoring during the trading day once it is deployed. Fortunately, the tighter timeframe looks like it’s going to help keep risk in check as well. This is how we intend to use history to prepare us for the future. The only certainty in trading is that nothing lasts or, works forever. Define your criteria and trade within your expectations.
