I’m loving the tell all, folks.
I really cannot add a whole lot to the conversation regarding individuals factors. Most systems’ core factors are often very similar; the main differences in them seem to be how one measures the variables, and the level of sophistication in normalizing those variables. For example, most comprehensive systems will look at a profit to price ratio. On the simplest level, this is price to GAAP EPS. More sophisticated implementations will attempt to normalize earnings, normalize the comparison with peers, and/or create historical precedents for “normal”. Given my inability to add much to that conversation, I thought it might be more worth your while if I discussed a few findings which I think are value added to asset price research:
A. Value Convergence vs. Price Divergence Theorems
A bottom up approach to valuation incorporates all known (public and private) information in order to identify a thing’s true value. The motivation to conduct this intricate approach is based on an idea which I call “Value Convergence Theorem” (VCT), which supposes that a thing converges with its underlying intrinsic value as markets digest known information. Implicit in this idea is the slow churn of the price discovery process; it takes time for markets to properly interpret what is known. In this way, VCT is consistent with the weak form of the EMH which does not contradict the idea that one may identify profitable investment opportunities through fundamental research. A DCF is the standard method by which to evaluate intrinsic value. VCT is typically the realm of investors.
A top down approach to valuation is based on the “Price Deviation Theorem” (PDT), or the idea that price temporarily differs from value when new information has not been incorporated or due to market disequilbria. That value already equals price presupposes that markets incorporate known (public) information in a fairly rapid and efficient manner; this is most consistent with the semi-strong form of the EMH. Thus, profitable investment opportunities are possible to those who can most rapidly digest and respond to new public information (or those who have access to privileged private information). Traders, who tend to be proponents of PDT, will seek out unexpected changes in the baseline through news developments, gathering privileged or “whisper” information, by responding to emergent correlations, front running sentiment (i.e., “buy the rumor, sell the news”), or by acting as counter-party or liquidity provider to unusually large volumes of supply or demand (i.e., “fade the rally”). Due to the more rapid decision making cycle herein, the consequences of being wrong regarding the interpretation of new information for a single case are far less severe than for in-depth fundamental analyses.
Note that proponents of PDT are not necessarily opposed to VCT, yet realize that – since price usually equals value – it requires far less effort to simply react to new developments. Given that markets are even weak-form efficient, the level of effort required to conduct a bottom up valuation usually far exceeds the potential rewards since the most likely outcome (e.g., for reading 200 pages of the latest 10-k, 50 pages of the latest 10-Q, tens of thousands more words from news releases and analyst reports, and countless hours of modeling) is a failure to find a robust discrepancy between price and value. As a fundamentalist, I can attest that it takes a lot of research to gain some intuition about the information which is versus is not incorporated into a stock’s price currently and over time.
I think both camps have merit. In reality, the price discovery process incorporates market participants who are motivated by any and all combinations of the value convergence and price divergence theorems. Thus it makes sense for investors to look at stock prices from both perspectives in order to address the potential for a price-value disconnect. Originally, I fell strongly into the VCT camp after reading Graham and Graham-Dodd. However, I have learned from experience that deep dive value investing is a painful and usually unrewarding process.
B. Individual stock price momentum is a result of industry pressures.
I.e., normalized for peer co-movement, stock price momentum does not exist (i.e., has no ability to anticipate future equity returns). That documented momentum anomalies are actually due to external factors should be much more palatable to proponents of EMH, who since the 1990s have been baffled by its mere existence. It is not nearly as problematic to have a peer group move together for an extended period of time due to cyclical or secular pressures.
Note, however, that EMH theorists never seemed to have a problem with mean reversion.
C. Inefficiency is a prime candidate for exploring and exploiting factor interaction
Inefficiency amplifies the magnitude of factor signals. Low normalized analyst coverage, volume/liquidity, institutional ownership, and short interest are indicative that the market under-appreciates known factors. Normalized in this sense adjusts the amount of coverage for the size of the company, the number of share outstanding, and amount of liquidity.
For example, if a stock which appears to be “cheap” also has lots of analyst coverage, lots of institutional ownership, and heavy short interest then said stock is most likely a value trap. Likewise, if those same factor which lent the appearance of cheapness were to be present in an under-followed company, then the likelihood of inefficient pricing is much higher.
Note the interaction between these variables; weighted factorization (as is the norm) ignores the interaction since the inefficiency itself is neither bullish or bearish on its own.
Obviously, the EMH plays a big part in my thinking. I know it’s anathema to some of you (erhem… Yuval), but for me it’s the gorilla in the corner of the room… 1000 lbs and still growing. I once thought that some day the move to passive investing will allow more inefficiency to creep back in. But now that I think about it again, I definitely believe that that efficiency gains in active investing due to data proliferation and AI will continue to erode the amount of alpha available to peons such as myself.