I do not usually write about investment styles, but I suffered through two winters at the University of Chicago, which means that  I am entitled to occasionally touch on the subject.

I have been listening to Barry Ritholtz’s Masters in Business podcast on Bloomberg. Lately, the podcast has been interviews with quant investors. One of their big themes is that ‘value’ stocks tend to outperform the market over time. Here, value is defined as something with a cheaper multiple (P/E, Price to Book, pick a metric) than its peers. There is a ton of data exploring this phenomenon, and everyone seems to agree that it works across all sectors.

When I heard this again, years after business school, it sparked a discordant note in my head. The University of Chicago side of my brain says believe the data. But the Valley side of my brain says this rule cannot possibly apply to tech.

When I was a research analyst, we used to say that “There are no Value stocks in tech.” The idea being that tech stocks with low multiples are cheap for a reason, and that they rarely recover. The landscape is littered with “Value Traps”, cheap stocks of companies that had fallen by the wayside. The idea being that in technology if a company mis-executes or misses a product cycle, it then faces oblivion, displaced by whatever the new technology is. The textbook example of this is Nortel, which was incredibly cheap all the way until it declared bankruptcy, having largely missed the 3G upgrade cycle (among many other problems).

Ritholtz, the podcast host, once made a similar concession noting that the value strategy of buying the worst performers of last year worked well “Unless it’s steam power or buggy whips.”, implying that technologies do go obsolete.

Now, I do not have access to all the stock databases anymore, so I can not directly research this. (If you have the database and want to share it, drop me a line. Happy to do a research project here.) But my guess is that, if we go back and look at “Technology” stocks for the past thirty years (or twenty or ten or five), we will find that Value methodologies work for technology as well as they do for any other sector.

So why the discrepancy between the hard data and empirical wisdom?

Obviously, this so-called ‘wisdom’ could just be an emotional bias. The stock market is littered with these, and there is a reason that quant funds have grown so big. However, I think there is something else going on here. I think we are mis-pricing technology in some way, and we need a better way to measure the quality of earnings of technology stocks. In this discrepancy, I see an arbitrage opportunity of some sort. The conventional wisdom and the quant data can be reconciled if we measure things better. In short, how do we quantify the risk of technological obsolescence, and then factor that into our stock-picking?

One of the themes of quant investing holds that while value stocks outperform, “quality” stocks outperform as well. And if we can find a way to measure quality, and apply that to value investing, we may have a way to meaningfully outperform. When the quants talk about quality they use metrics like revenue growth and margins or share buy backs and dividends. In technology, I think we can find a better measure of quality. The trouble is, what is that measure?

I do not have the answer just yet. I know buy backs and dividends are flawed since so many tech companies are growing so rapidly that deploying capital in this way is probably not a good investment. Also, I know that margins only tell part of the story. There are structural reasons for some parts of ‘technology’ to have better margins. Semiconductors are lucky to get 50% gross margins, but software companies should have at least 80% gross margins. But buying software over semiconductors is only a strategy that works during certain points in the cycle, and probably would have been a losing bet over the last three years.

On the other hand, our goal should be to remove as much subjectivity from this as possible. I am looking for quantifiable rules that remove any potential emotional bias.

My guess is the answer lies in some combination of growth metrics and changes in margins. I also think it needs to be applied on a sector basis. Buy a bundle of semis stocks and short a bundle of SaaS names, for example. My guess is that if we apply these new rules to technology sectors as a whole, there is a solid strategy in there, somewhere. I recognize that these are already widely used quant strategies. So at this level, I think there is room for a more qualitative metric. Just assigning companies to different tech ‘buckets’ alone will introduce some of this.

I do not have a complete answer yet, but I do think it merits further exploration.