What is back-testing?
Back-testing is an analytical process in which historical data is used to determine the success or failure of a strategy. If a tested strategy generated good past results, one assumes it will do well in the future.
For example, you might believe dividend paying stocks perform better than stocks with no dividends. With a back-testing application, you can pretend you built a portfolio of dividend paying stocks ten years ago and then analyze how it performed in the ensuing decade. With good results, you might start shifting your assets into stocks with dividends or funds containing same. If the results are bad or neutral, you test other strategies until you find one with satisfactory returns.
Why you should avoid back testing
Past performance is irrelevant
Looking backward, you can “discover” all sorts of successful strategies. Unfortunately, you are not really finding anything of relevance. Very few stocks and sectors perform exactly in line with the market. By definition, some will be above average, others below. There are an infinite number of ways to group stocks. Some of those groupings will have performed better than others. Because they have done so in the past does not mean they will do so in the future. Consider the relative performance of the asset categories shown in Figure 36-1.
Each major asset category has its believers. Some investors favor growth, others value. Some like the ‘action’ in small-caps; others prefer the solid nature of large-caps. Some believe the best opportunities are found abroad. As Figure 36-1 demonstrates, no one major asset category has been consistently atop or beneath all the others. This lack of consistency among categories applies as well to groupings based on other factors such as yield, key ratios and industry.
Computers are back-testing 24/7
While you are reading this, computers throughout the world are running programs seeking patterns in a quest for excess profits. They analyze every publicly available statistic such as stock prices, corporate earnings and economic data. In the time it takes you to create a single strategy and run a back-test on a popular website, hundreds of computers have run millions of simulations. If there is value to any discernible statistical trait, multiple firms have already executed the trades and removed the opportunity before you’ve read your own test results and had your ‘aha’ moment.
Active management is of questionable merit
Even before computers came onto the scene in a big way, research revealed the futility of active equity management. Burton Malkiel’s “A Random Walk down Wall Street (W.W. Norton & Company, Inc., 1973, reprinted many times since) popularized this notion while mutual fund data before and since have basically proven it.
With so many profit-seeking investors having access to the same public information, foreseeable excess gains are few, fleeting and rapidly eliminated. With current technology, reaction time to new information is measured in milliseconds. Over twenty-year periods, about 90% of actively managed mutual funds are outpaced by market indexes. When you back-test to make investment decisions, you are actively managing your money with vastly fewer tools than the pros who themselves consistently lag those same indices.
What you should do instead
Back-testing takes time and adds no value. If you have used it and profited, count yourself lucky and quit the game while you’re ahead.
Look forward with your investments, not backward. Monday morning quarterbacking is as futile in investing as it is in football. Save time, avoid frustration and ignore back-testing tools. Invest with a logical, forward looking regimen.
* Data sources:
Small-cap value: Russell 2000 Value Index
Small-cap growth: Russell 2000 Growth Index
Large-cap value: Standard & Poor’s / Barra 500 Value Index
Large-cap growth: Standard & Poor’s / Barra 500 Growth Index
Foreign: Morgan Stanley Capital International EAFE Index
Bonds: Lehman Brothers Aggregate Bond Index