How valuable are various datasets?

[pmaymin]

Thought you might be interested -- my new paper suggesting a deep-learning market based approach to answering this question generally, and applied to some NBA datasets in particular.

paper: http://ssrn.com/abstract=2703636

or video with slides here: https://youtu.be/vsaAOAgja7g

[Crow]

If you are a super serious big bettor, you should have a tool as good as the Vantage database or better or several. If you are a team, it would be hard to justify passing on any reliable / well-organized / non-redundant database with a reasonable claim to knowledge and some unique knowledge that isn't insanely priced (for a $100 plus million / yr enterprise) unless someone else or you can build better or until your experience with it gives you a really strong case for skrimping on that cost. Not using it for much analysis or well or having a GM, Coach and / or players who won't listen or adapt to advice generated from it might justify cost cutting, but only if your case is hopeless.

[Nate]

So, will you be back in a year with a new and different crowning achievement in basketball analytics? I guess we can only hope so.

It seems very much like Vantage is trying to produce the sort of data that will make the various techniques used for baseball analytics more viable for basketball analytics. I do find myself wondering whether Vantage uses broadcast footage, or something that's a bit more raw.

[mscbuck]

So, will you be back in a year with a new and different crowning achievement in basketball analytics? I guess we can only hope so.

That's Philip's MO!!!!

[pmaymin]

Thank you so much for the kind words!!

just fyi here's another piece about the approach, this one more descriptive than quantitative:
http://www1.vantagesports.com/Articles/ ... 8AACwqCgYv

[sethypooh21]

I'd still very much like to take a spin through some of the Vantage data, but that has proven challenging.

[pmaymin]

I thought you'd been setup with access quite a while ago? Send me an email if you're having difficulties, let's get you sorted.

[sbs]

So, @newdatasports Vantage's tout service that was attached to this paper started publishing fully transparent picks after the games began. It lasted 4 days.

They went 11-19 in their published plays and now the Twitter account is deleted.

https://twitter.com/vantagesports/statu ... 9526905856

[Crow]

That is a noteworthy or at least curious development, I think.

Any official explanation made anywhere? How about a followup here? I assume the service is continuing, just not that form of advertising at the moment?

sbs, if you are inclined to say, are you in a related business, a related business user, or...?

Do you have any public or proprietary datasets to recommend for sports analytics for team performance understanding and improvement (my main interest) or application to betting (a big interest to many, a curiosity at this point for me)?

Any awareness of the value of datasets at sportradar.com (a big company- a 1000 employees- I hear that Ted Leonsis, Mark Cuban and Michael Jordan invested in) for either purpose or anything beyond the becoming typical media products? Or perhaps Sports Data Systems?

[Nate]

...
They went 11-19 in their published plays and now the Twitter account is deleted.
...

Maybe they quit when the two-tailed P value on the hypothesis that their win rate is 0.55 went below 0.05.

[pmaymin]

Hate to be mysterious, but lack of success was definitely not the problem or the reason the public service was taken down. FYI, the paper is currently under review at a technical journal. (And by the way, the "p value" calculation/hypothesis above is not correct.)