I was going to spend some time looking for more rose-colored swans to write about today when I got a call from an old friend, Henry Carstens, who is one of the best stat researchers and traders I know. Some of you may recall that in 2008 Carstens collaborated with me on a project to test the performance of S&P 500 stocks that traded below book value and had low debt levels. The results were fantastic, and this simple strategy has remained part of my investing arsenal ever since.
In addition to being a good guy, Carstens is incredibly smart. He has degrees in mathematics and economics and, along the way, picked up a minor in physics. Besides actively trading the markets based on statistical patterns and interpretations, his company, Vertical Solutions, offers forecasts on different markets, as well as in-depth market research. After we talked sports, chess, politics and kids, he offered insights that might be valuable to traders and gave me permission to share them with Real Money readers.
One of the first questions I asked was about structural changes in the market. Many of my statistically oriented friends have reported that their models don't work anymore. Carstens told me that he began noticing this a few years ago. The performance and reliability of traditional pattern analysis peaked in 2007 and has continued to deteriorate ever since. I asked if he had any idea why -- perhaps the bond market's relationship with other markets had changed, or the rise of black boxes caused the changes? He said something that I think is critically important: We can speculate all we want about why it happened, but why it happened doesn't matter; it only matters that it happened, and you have to adjust to the new environment.
My friend then broke out his data and went to work looking for a solution. Carstens found one that is simple and elegant, as most solutions usually are. He changed his clocks. He now runs his business on European time. He finds that all the price action in markets like the S&P 500, bonds, bunds and gold occurs between 4 a.m. and 10:30 a.m. Central time. The big moves occur during the European market's hours of operations, and then nothing significant happens for the rest of the day from a trading point of view. By the time most U.S.-based traders have finished their coffee and settled in, the trading is already over.
We also talked about the Volatility Index (VIX) as a trading tool. Like many, Carstens doesn't find any absolutes about the VIX. The key lies more in trends in the index and certain threshold levels. In general, he finds that when the VIX is above 28, the stock market tends to be more mean-reverting and traders should focus on trades that fade the current direction. When the index moves below 28, the market tends to become more of a trending market and traders probably want to look for trades that go with the current trend. When the index moves below 14 (though not a definitive trading signal), the stock market begins to be "toppy."
Carstens shared something else that can make all of us better traders and investors. He has found that adding a simple statistical test to your research can dramatically improve results. We often look at a trading or investing ideas, like buying stocks below book value or buying the index when stocks are at a multiple-day high or low, and initially find positive results. But it doesn't work as well as we think it should in real life. So Carstens charted all his ideas for the past few years and found that testing for a conclusion's T score will remove all the random results and allow you to focus only on those patterns or parameters that are significant. He found that the conclusions that scored above 1.6 worked, and those below that level were simply random market noise. On his website, VerticalSolutions.com, he describes how to run this simple test.
This is all valuable information for traders and investors alike and, as always, I'm grateful to my friend Henry for sharing his wealth of knowledge with us.