Response: There is a general framework for making investment decisions called “asset allocation”. This framework is popular with academics because, unlike technical analysis indicators, for example, it has a theoretical basis. However, the framework itself leaves out some details that you need in order to actually implement it. My method is an implementation of asset allocation, in which (I think that) I have worked out these details.

Here is how asset allocation works. Each investment has two characteristics, one called expected return, and the other called risk. Intuitively, here is how they are defined. If you took a large number of unrelated investments with the same expected return, then, on average, they would give you this expected return. Risk is a less clear concept. Usually, risk is taken to mean the standard deviation of logarithmic return, which is also called “volatility”. This is a good definition of risk because it is easy to understand — higher uncertainty in future returns means higher risk. However, the definition has some weaknesses. I have developed another definition of risk which I use in my algorithm.

Both expected return and risk are estimated from historical price data. In other words, mine is a “technical” system — it only uses historical price data. It does not use any “fundamental” information, such as earnings. The argument for technical analysis as opposed to fundamental is well known. Even if you think that fundamental information is relevant, by the time you, the investor, hear of it, a lot of other people have already heard of it and acted on it. Thus, any relevant fundamental information is reflected in the price before you even hear about it. By analyzing price, you can know all the relevant information. Plus, of course, technical systems are easier to test on historical data. Simulated performance is one of several pieces of evidence that your system works. See below for more on this.

(Just to clarify, even though asset allocation is a technical approach, it differs from most other technical analysis systems. Most technical analysis uses indicators, such as moving averages, or chart patterns, or other similar methods. However, there is no theoretical basis for those approaches.)

Since expected return and risk change with time, how to properly estimate them is a very big question. Many asset allocation methods, such as the one used by the Lifecycle funds within the TSP, actually assume that these investment characteristics do not change with time. I think that this is a major mistake.

Just as each investment has an expected return and risk, so does every combination of investments. Asset allocation says to pick the mix of investments that maximizes the expected return while not exceeding a fixed maximum risk. Of course, there is a huge number of investment combinations. Thus, it is not possible to check all of them using a “brute force” approach. However, with a good search algorithm, you can find a mix of investments with a high expected return that does not exceed your fixed maximum risk. That’s all there is to it. In my newsletters I give two allocations. The difference between them is that they have a different fixed maximum risk.

Related posts:

1. Personal Investment Performance (PIP) calculation
2. Bonds continuing to earn money
3. Focus on return and risk, not price; losses are locked in as they occur