The more people think there is no small cap premium the better, more zing for me
I can confirm the previous posts that the small cap factor positively interacts with other factors, esp. with profitability and other quality metrics and that it is hard to capture the small cap factor without any other factors. It is well described in the book “Your Complete Guide to Factor-Based Investing: The Way Smart Money Invests Today”, which is quite cheap and which I can recommend.
When I analyse a factor in a ranking system, I usually add a small cap factor in order to see, if there is a positive influence. Example:
You can also use a universe with positive earnings or positive cashflows to can find the difference:
You can find similar differences, if you use a small cap universe and applying a ranking there (but I personally do not like to cut away large caps - to remove non-profitable stocks seems more logical to me and this leads also to a more “static” universe, which has less turnover/slippage in a Strategy).
Other approaches to find the small caps are more sophisticated. For example it is well known that small caps outperform during bull runs: So you could use sth. like a market timing formula like (I don’t recommend to use this formula by its own, because most of the time the ranking is “inactive”):
eval(close(0,getSeries(“RSP:USA”)) > highVal(60, 0, getSeries(“RSP:USA”))*0.99, MktCap, NA)
(you can use also other treshholds)
Resulting in a ranking backtest, where the smallest stocks outperform all other:
In addition you could also experiment with EV, the volume, the price, the number of Analysts etc to find a similar thing like the Small Cap premium.
Great post.
I think this related to what @yuvaltaylor calls augmenting factors and what you described here are examples of procedures of finding these augmenting factors.
I’ve had some succes using a similar factor using the high yield spread as pointed out here: Crisis Investing - #4 by RTNL. Basically it is the same but instead of RSP:USA, ##CORPBBOAS and a minimum treshhold for the high yield spread (>5). Gives similar resuts.