Problem: How to combine information across companies when the information arrives asynchronously due to “non-informative” reasons. For example, company A presents in a (industry) conference on Day x NOT because that is the day they announce any news, but because that is the day the conference happened to be schedule. Company B happens not to attend this conference, but attend another one in a later time (again decided for random reasons like “that was the best day the event organisers and journalists could find”). We therefore find out about the firms in “random” times. How can we use such “randomly asynchronous info flow” to decide how to allocate $ between companies A and B? (and in general across N companies with flows of random-dates events)?
Note: this is not the same as, say, earning announcements. These are well-planned in specific periods. It is also not the same as analyst up/downgrades: the analysts will communicate the info AS SOON as it arrives. On the other hand, this problem becomes even more “central” when one uses “alternative data” as in the attached document: these are data, e.g. From review of consumers online or whatever, that are totally “random + asynchronous”. This is a major problems when using “alternative data” - which is one of the “hot things in the market” these days.
tevgeniou
commented
6 years ago
Problem: How to combine information across companies when the information arrives asynchronously due to “non-informative” reasons. For example, company A presents in a (industry) conference on Day x NOT because that is the day they announce any news, but because that is the day the conference happened to be schedule. Company B happens not to attend this conference, but attend another one in a later time (again decided for random reasons like “that was the best day the event organisers and journalists could find”). We therefore find out about the firms in “random” times. How can we use such “randomly asynchronous info flow” to decide how to allocate $ between companies A and B? (and in general across N companies with flows of random-dates events)?
Note: this is not the same as, say, earning announcements. These are well-planned in specific periods. It is also not the same as analyst up/downgrades: the analysts will communicate the info AS SOON as it arrives. On the other hand, this problem becomes even more “central” when one uses “alternative data” as in the attached document: these are data, e.g. From review of consumers online or whatever, that are totally “random + asynchronous”. This is a major problems when using “alternative data” - which is one of the “hot things in the market” these days.