I'm just a homer gamer non-institutional trader. Has anyone here dabbled around with "Alternative Data"? It seems to be a hot buzz word right now. The first generation of this stuff was centered around getting sentiment from social media and picking up on tweets before official news releases. I read a story about how a group of traders got long RBOB when a picture of a refinery on fire was tweeted and their special software picked it up before it became news and before the price spike occurred.
Stuff like that is cool but what I am really interested in is this next generation of Alternative Data. You have people using realtime satellite imagery to measure shadows above oil refining holding tanks to estimate inventory before the official inventory numbers are released by the government. You have people looking at satellite photos of brick and mortar retailer's parking lots across the country to estimate success before earnings.
The sources are endless, Railroad activity, anonyomized retail receipt data, IOT device reporting, Natural Resource cargo numbers, satellite photos ...it is just endless.
Aside from thinking out loud in this post, I wonder how hard it would be to gather same stores sales data for every single location and then try to isolate which stores performance correlates with the quarterly earnings report? You could then just focus keeping an eye on a few locations to get a true leading indicator of the stock performance. This has already been done by flying satellites over every J.C. Penny location to count the number of cars in the parking lot. This type of thing just ain't accessible to the home gamers but it is a crystal ball in some instances.
Looks like quandl has an offering:
https://www.quandl.com/alternative-data
Stuff like that is cool but what I am really interested in is this next generation of Alternative Data. You have people using realtime satellite imagery to measure shadows above oil refining holding tanks to estimate inventory before the official inventory numbers are released by the government. You have people looking at satellite photos of brick and mortar retailer's parking lots across the country to estimate success before earnings.
The sources are endless, Railroad activity, anonyomized retail receipt data, IOT device reporting, Natural Resource cargo numbers, satellite photos ...it is just endless.
Aside from thinking out loud in this post, I wonder how hard it would be to gather same stores sales data for every single location and then try to isolate which stores performance correlates with the quarterly earnings report? You could then just focus keeping an eye on a few locations to get a true leading indicator of the stock performance. This has already been done by flying satellites over every J.C. Penny location to count the number of cars in the parking lot. This type of thing just ain't accessible to the home gamers but it is a crystal ball in some instances.
Looks like quandl has an offering:
https://www.quandl.com/alternative-data
