5 ms·

Makes sense! I would love to hear more about that method (if it's something you can talk about)

by csjr 9y ago

icefox 9y ago
Couldn't you use the German tank problem to solve these? For Tesla auto sales couldn't you just look at tesla vin numbers? For iPhones, sample mac address's?
- dsacco 9y ago
  > For Tesla auto sales couldn't you just look at tesla vin numbers? You’ve definitely intuited some of the method, yes :). The rest of it entails: 1) how to get all VINs both authoritatively and legally, 2) how to distinguish between valid VINs and assigned VINs, 3) how to reverse the actual revenue projection from the set of all assigned VINs. The first requirement is the hardest. Tracking self-reported VIN delivery from users isn’t rigorous enough. You could use an endpoint and scrape from it, but how would you do it legally and reliably? The second requirement is also difficult. Assuming you’ve found an authoritative source for valid VINs, how do you distinguish which VINs are assigned? Once you have those two, the third requirement is mostly straightforward. You can implement your own VIN decoder using public NHTSA documentation, map each VIN field to options and prices across models, and track sequential VINs using the distinguishing method of requirement 2 on the data you’re getting from requirement 1. Naturally, there are other ways to do this that don’t involve VINs at all.
sumedh 9y ago
Using satellite images to monitor parking lots, cargo ships. Maybe some satellites might be tracking how many trucks leave Apple factories in China. Sometimes it's just good old illegal insider information.
- dsacco 9y ago
  Actually, my friend simply reverse engineered FedEx and UPS tracking numbers, then backtracked sequentially by ordering an iPhone every few weeks. I say “simply” because it doesn’t require any of the infrastructure you mentioned, which is what the public typically associates with “alternative data.” It was actually quite impressive and novel at the time, and inspired other projects of mine. I can talk about this now because (to my knowledge) it no longer works, though it did for a while. He was in charge of approximately $100M when he was working on this.
  - sumedh 9y ago
    Interesting. When so much money is on the table people are going to be creative.

Makes sense! I would love to hear more about that method (if it's something you can talk about)

by csjr 9y ago

Makes sense! I would love to hear more about that method (if it's something you can talk about)

Couldn't you use the German tank problem to solve these? For Tesla auto sales couldn't you just look at tesla vin numbers? For iPhones, sample mac address's?

dsacco 9y ago
> For Tesla auto sales couldn't you just look at tesla vin numbers? You’ve definitely intuited some of the method, yes :). The rest of it entails: 1) how to get all VINs both authoritatively and legally, 2) how to distinguish between valid VINs and assigned VINs, 3) how to reverse the actual revenue projection from the set of all assigned VINs. The first requirement is the hardest. Tracking self-reported VIN delivery from users isn’t rigorous enough. You could use an endpoint and scrape from it, but how would you do it legally and reliably? The second requirement is also difficult. Assuming you’ve found an authoritative source for valid VINs, how do you distinguish which VINs are assigned? Once you have those two, the third requirement is mostly straightforward. You can implement your own VIN decoder using public NHTSA documentation, map each VIN field to options and prices across models, and track sequential VINs using the distinguishing method of requirement 2 on the data you’re getting from requirement 1. Naturally, there are other ways to do this that don’t involve VINs at all.

sumedh 9y ago

Using satellite images to monitor parking lots, cargo ships. Maybe some satellites might be tracking how many trucks leave Apple factories in China. Sometimes it's just good old illegal insider information.

dsacco 9y ago
Actually, my friend simply reverse engineered FedEx and UPS tracking numbers, then backtracked sequentially by ordering an iPhone every few weeks. I say “simply” because it doesn’t require any of the infrastructure you mentioned, which is what the public typically associates with “alternative data.” It was actually quite impressive and novel at the time, and inspired other projects of mine. I can talk about this now because (to my knowledge) it no longer works, though it did for a while. He was in charge of approximately $100M when he was working on this.
- sumedh 9y ago
  Interesting. When so much money is on the table people are going to be creative.