For nearly two decades, top value investor and columnist John Dorfman has been tracking a purely quantitative “robot portfolio” that has beaten the S&P 500 by about 12 percentage points per year. His advice: Don’t use it.

In an interview with Wealth Track’s Consuelo Mack, Dorfman talks about why his robot approach isn’t suitable for the vast majority of investors. The strategy takes all stocks with market capitalizations of at least $500 million, eliminates those whose debt is more than their equity, and then selects the ten with the lowest price/earnings ratios. While its track record is impressive, Dorfman says the strategy can be very volatile from year-to-year, and can become very concentrated in a particular sector or region. Because of the volatility and lack of diversification, he says he uses the strategy as more of a jumping off point. He and his team will cherry pick stocks that the strategy identifies, he says, and also select other stocks that have manageable debt and reasonable valuations. Doing so does not always enhance the returns of his robot portfolio, he notes, but it allows him to build a more diversified, less risky portfolio.

Dorfman talks about expectations and valuations, explaining that companies with low P/E ratios are not expected to do much, while those with high P/Es are priced for perfection.  He prefers investing in those low-expectation firms, provided they are financially sound, because even modest improvement in their performance can push shares higher. Dorfman also talks about why he is high on auto parts firms, and why he thinks that a “titanic” opportunity is on the horizon in the oil industry.