Archive - Mar 19, 2012
A while ago, I wrote about programs that I had created using the Empire Avenue API to extract portfolio information. Originally, I wrote about using it for my own portfolio, but you can also get limited information about other people's portfolios.
So, I started pulling down the portfolios of the people that I own stock in to do some expanded analysis. Given the limits of the Empire Ave API, I can only pull down around half a dozen large portfolios per hour. Nonetheless, I kept it running for an extended period until I had pulled down around 450 different portfolios. I then loaded the data in MySQL and started crunching some numbers.
From these 450 different portfolios, I came up with approximate 30,000 different tickers. 38 different tickers appeared in over two thirds of all the portfolios I analyzed. 134 showed up in half of the portfolios, and 360 showed up in at least a third of the portfolios. Nearly 10,000 showed up in only one portfolio.
Not surprisingly, as a general rule, the 38 tickers were highly active players that are on various leader boards. It seems as if by using this data, it should be possible to recognize people that cluster into certain groups and acts as connectors between these groups. However, I have yet to find nice ways of coming up with these groupings.