By controlling risk, you outperform over time. It is an odd concept to wrestle with, that taking a defensive role actually creates more alpha.

So how does a trend follower determine when a trend is here? A surfer scans the horizon for an incoming wave and uses his or her knowledge of what a good wave looks like when deciding whether to ride the next one. In the same way, trend following uses historical precedent.

Now, before you get all up in arms, saying, “Just because it happened in the past doesn’t mean it will happen in the future,” remember that our entire thought process as a civilization is based on using historical precedent to predict a future outcome. In fact, our brains are even oriented to think this way. If a baby cries, he gets the attention of his mom, so he cries more often. He has learned by analyzing history that crying gets him what he wants.

Statistics is, “the practice of collecting and analyzing numerical data in large quantities.” It offers a way to study history and see the likelihood that one outcome will happen again in the future. The goal of statisticians is to find out what “the norm” looks like, and through that analysis they can predict the odds something will happen again.

Most of our society is designed around statistics. Traffic laws are based on them. Airplane manufacturers use historical studies to find the best materials to keep planes in the air. College admissions are based on the SATs. Even the IRS uses statistics to analyze income tax returns, looking for those with a high probability of fraudulent information. Historical analysis/statistics are used literally everywhere to gain a valuable insight into what might happen in the future based on the past.

Trend following is no different from fundamental analysis in this regard. All successful strategies use historical analysis. If you think about it, fundamental analysts have developed systems that use specific ratios that, “based on history,” have high success rates. I realize I am oversimplifying, but a low PE company implies that its stock is undervalued, and the odds are good that buying stock at that point will result in a profit.

The reality is that all systems perform ideally in certain environments and are less ideal in others. Trend following is good at avoiding the extremes. When markets are trendless, of course, this strategy does not perform as well as traditional techniques. One style is not necessarily better than the other, just better at specific moments. A combination of both approaches offers the best of both worlds: minimizing loss during declines and maximizing gains in trendless or trending markets.

The Perils Of Data Fitting
The problem with using historical analysis for anything is related to “data fitting.” This occurs when an assumption is made about the future outcome based on past precedent without a large enough sample. If you flip a quarter 10 times and it shows up heads nine times, you could suggest there is a 90% chance the next flip will come back heads. But we all see the problem there.

Because making assumptions using small data sets poses a significant problem, to get around this issue a good analyst will start with a theory and test it. If the theory holds up, he or she will test it further with larger and larger data sets. To me, a strategy, regardless of its technique, is successful only if it is consistent and performs as it is expected to across all time periods.

When marketing any investment strategy to prospective clients, it is imperative that the client get an accurate assessment of how it will perform over time. By showing only specific periods of time (data fitting), advisors might imply better performance than what will actually occur. Picking a handful of very selective charts that omit negative performance only hurts your firm over time.