Friday, December 30, 2011

Mechanical Trading Strategies & Backtesting and how they Create Bias

Mechanical-based trading strategies produce unintentional style or sector bias in portfolio construction. Backtesting is not the key to generating returns.

The argument behind this article is that purely mechanical-based trading strategies have a tendency to produce unintentional style or sector biases. This is problematic because the investor will find himself manifesting a viewpoint or weighting that he may have purposefully set out to avoid.

Indeed, many favour a mechanical-based strategy because it obviates the necessity for a discretionary viewpoint. This is important because an investor might want to avoid the pitfalls of selecting sector/stock weightings. Similarly, from a behaviourial finance prospective, he might want to avoid the psychological pressures of having to adjust his positions by constantly having to make discretionary decisions. In particular, value investing is seen as being susceptible to these pitfalls.

Back Testing of Trading Strategies and Why it Doesn’t Work

There is another danger here that may prove to be more pervasive via its insidiousness. Mechanical trading strategies always involve substantial back testing and, investors try to test conditions across a range of market conditions. However, what if the strategy results in a sector/style weighting that is favourable across the period of back testing, but not likely to repeat itself going forward?

The likelihood is that he will reject the system outright, or else try to refine, or add new factors to, the system in order to incorporate the new results. If the argument of this article is correct, than this process of curve fitting is likely to break down again, and the investor will find himself disappointed again.

All that this refinement is likely to achieve, is to produce a new sector/style bias -which works as
 a ‘best fit’ solution to the updated results-that will may fail as future conditions change.

Mechanical Trading Strategies Produce Unintentional Style or Sector Bias?

In a sense this is an epistemological question. In producing a mechanical strategy, traders will be trying to define a set of universal and consistent parameters upon a series of results that is actually ever-changing and evolving. Markets don’t have memories and they are not obliged to follow rules. Similarly, there is little regard for sentiment in market based strategies. This is a pity because, many believe that sentiment- think Geenspan’s ‘irrational exuberance’ and Keynes ‘animal spirits’- is what guides markets.

Furthermore, in creating the parameters for mechanical evaluation of, say stocks, they will struggle to avoid defining a factor, that doesn’t intrinsically favour a style or sector.

The latter argument is best expressed by some examples of the type of investment ratio, or filters, that are used by traders. Namely, dividend yield, price to book, price to earnings or PE ratio, price to cash flow.

Using Fundamental Filters

This section is best expressed by a series of snapshot examples within bullet points:

*Dividend Yield. Stocks that pay dividends tend to be more mature and generating cash flow. Therefore, selecting on the basis of yield tends to result in slower growth, mature companies that have previously doing well. In other words, yield would have guided investors in buying banking and housing stocks before the crash of 2008!

Similarly, favouring stocks without a dividend, would have resulted in favouring technology and internet stocks right up until the dot com bust. Furthermore, high yield companies tend to have less free cash to invest for growth. These types of stocks maybe favoured in a recession but not in a recovery.

*Price to Book. This is a favoured measure for value investors. The problem here is that different industry sectors require different amounts of tangible assets in order to generate profits. For example, a plant hire firm must hold substantive amounts of capital machinery, whilst an intellectual property company (biotech, semiconductor design, software etc) does not.

Similarly, mechanically evaluating the assets of a company does not tell investors about the quality of those assets and, its ability to generate returns in future.

*Price to Earnings. PE ratio will tell investors little about cash flow, or the capital expenditure requirements of the firm. Furthermore, they will differ in relation to where the company is in its growth cycle. However, the most damaging aspect of solely focusing on PE ratio is that it encourages a focus on companies that already have earnings. Whilst superficially attractive, this strategy means investors will rarely expose themselves to sectors like biotech or oil exploration.

*Price to Cash Flow. Focusing on this metric has many of the pitfalls of the PE ratio. In addition, it can encourage investors into low growth companies. Similarly, the market could be pricing these sectors cheaply because, they are about to be structurally challenged. For example, large cap healthcare companies currently generates huge cash flows, but how sustainable is it with the onset of an ongoing Obama health care reform process? What threats do generics pose? What drugs are in the pipeline? None of these questions are answered by solely focusing on the mechanical approach.

Portfolio Evaluation

The examples above are just a few considerations of the complexity or even impossibility of not manifesting a sector or style bias by taking an mechanical approach. Many believe that the correct approach to portfolio construction is to understand that portfolio construction is as much an art as it is a science. Furthermore, for a rational investor, taking an outright style or sector ‘view’ is a necessity, because he will serendipitously do it via mechanical strategies investing in any case!

Keynes, John Maynard "The General Theory of Employment,Interest and Money" Palgrave, 1936

1 comment:

  1. Finding the right software package to help with financial management was quite a hassle, until I came across Trading Blox Equity System Pack. After speaking with them, I learnt of this great new software. It helps you to manage your own money in today’s fast paced society by making better decisions. They are a small team so the support I received was amazing. I would recommend anyone to check out this new product