Tuesday, September 06, 2011
US MVP Year to Date
I have my own set of MVP and low volatility indices over at betaarbitrage.com. Here's the MVP drawn from the S&P500, compared to the S&P500 this year. It had a very good August, primarily because it has only 50 stocks from within the S&P and one of them was Motorola (that generated 1.5% to the index in August). So, it's up 13% more than the S&P for this year, whereas prior to that, from 1998 through 2010, it outperformed only by 4.8% annually. An it has about a 0.5 beta, and about 18% less volatility. Basically, it outperformed because of standard tracking error, being a subset of 50 stocks, so I don't read too much into one month.
I'd like to find the other MVPs and see how they compare, as I bet over short periods like 8 months they vary a lot. One distinction of mine is that I totally ignore industry concentration, and just take the 50 stocks that generate the lowest portfolio variance (estimated on 3 latent factors using the prior 252 business days). A lot of people add industry limits, but I find that double counting. It is not as if any industries have an obvious 'size' or 'value' effect, and to the extent they are correlated that should be addressed in my algorithm. So, que sera sera, industry-wise. I think it might be my special sauce relative to all these newcomers.