Comment by hendzen

Citadel and Two Sigma both have had large teams working on this for years. They are making a lot of money doing it but barriers to entry are very high. You need to collect all the data in a way such that you can reconstruct how it looked at any point in time. The vendors can’t be trusted to not make retroactive corrections, so you have to collect it for years before it becomes useful. Doing that takes a lot of time and money.