4 ms·
Alternate data doesn't even need to be as sexy as satellite photos, hell you almost certainly want the data that isn't sexy, the stuff people haven't thought of
by jamez1 9y ago
Alternate data doesn't even need to be as sexy as satellite photos, hell you almost certainly want the data that isn't sexy, the stuff people haven't thought of because it's too boring. Alternative data vendors above all want sales, and even the funds themselves want things to show off to clients. This gives you great opportunities to look at the alternative data they aren't touching.
Given this is a predominantly a web development community it always surprises me how little creativity there is in the articles on investing. Neural networks and machine learning sound cool but the reality is almost none of the readership would be able to make any money off them.
Simply tracking how many sales or users exist in databases by watching sequential IDs should be the go-to method for any web developer trying to get an edge. I would have expected HN to have articles where people are getting creative on that, ie trying to use measures of entropy on usernames to get rough subscriber numbers etc.
Even plain scraping of prices etc, is often full of great insight that is ignored. If a grocery store drops their prices in profitable categories against their competitors, that could be the signal about an incoming price war for an entire sector. There's a lot more information in that than social media feeds and all the other sorts of sexy data that get coverage in the media.
- geomark 9y ago"...tracking how many sales or users exist in databases by watching sequential IDs..." Seems like a great application of the German tank problem [1] that was mentioned on HN the other day. [1] https://en.wikipedia.org/wiki/German_tank_problem https://en.wikipedia.org/wiki/German_tank_problem
- bookmarkacc 9y agoThe reason hedgefunds look at satellite images and oil tankers is because everyone looks at sequential ids and price changes so that doesn't give an edge.
- jamez1 9y agoThat's simply not true - equity analysts can cover anywhere between 5 and 500 stocks, do you really think they have the time or skill set to track all of that? It really is laborious, grueling work. If you look at the possible returns the equity market is going to make from a stock in a dollar value, and how much research spend is as a percentage of that, you'll quickly see it doesn't pay for much. You can tell simply by looking at the broker research - that's probably the extent that analysts take things. The big stocks obviously have a lot of it happening. (eBay listings, airline pricing etc is obviously touted a lot) But once you start to go down to the mid caps, you enter a void where there isn't much heavy data focused research done, and it's very possible you can have a better gauge of the business than any other investor on the planet once you pull out this data out.