Anyone here have success with P123’s Insider Buying functions, i.e. do they add excess return?
I’ve been testing a strategy that according to the literature should boost returns for those stocks where there is significant insider buying in the prior 30 days. I’ve seen no measurable impact in my tests (unlike most of my other screens), nor, for that matter, do I that these functions do much in the way of filtering stocks.
Any suggestions would be welcome, thanks.
fwiw, I don’t presently utilize insider buying data in any of my models. When I was working with a year or so ago I remember thinking there might be something there, but compared to other factors it seemed hard to discern. I remember wondering if the insider sell data might be skewing results also because of options grants, and that high insider selling might actually muddy the waters and be a sign of high bonuses due to good company execution. I need to look at it again though - I think I recall marco updated one of the insider ownership fields since then.
I am not an expert on the literature, but have read a lot of it. I believe excess returns from insider trading is shown, primarily (or almost totally), to come from when either the CEO or CFO buys a more-than-trivial amount of share-value. (Actually, the CFO has resulted in higher excess returns than CEOs.) I am new here, but I think, unfortunately, the database used here does not break out that level of insider info.
As a side note, I have tried incorporating the insider info available here into my screens and they had no positive impact (of course, other models very well could.)
I think the market may move faster than you in terms of how quickly you can see the data through P123.
For instance for Ins#ShrPurch:
Timeliness
Ownership data is very fluid. A single insider transaction can occur on any day. Since we only expose aggregate data, though, an arbitrary period is used to snapshot the data.
For insider transaction data we use S&P monthly snapshots. These are done at the end of each month.
There are two notable exceptions: Insider#Own and Insider%Own factors. These factors are calculated daily by us and we then snapshot it once a week for historical simulations.
Exclude pre-2004:Coverage
Note that complete ownership data for these series begins around December 2004. Some data exists prior to that time in the database, but it has more holes and is not as consistently available across companies.
Thanks for the replies everyone, sounds like my experience is not an outlier.
I currently use it, but it’s rather low on the list.
If I remember correctly, the literature suggests that the types of purchases/sales are more informative than the top level numbers of insider ownership and the amounts bought or sold.
In any case, I suspect that historical out-performance associated with insider ownership patterns is highly susceptible to information diffusion.
Building an xml parser to pick out insider patterns from SEC forms 3/4/5 and 13-D/E/F was one of my first projects as a junior quant.