Quote:
We think we have a nice solution to the low marginal cost of submitting multiple lineups to the same contest.Our lobby lists contests pools. At lock time the actual contests are created through random opponent selection with first entries matched up against first entries, second vs second, and third vs third.
Love it.
Quote:
Originally Posted by DraftboardDan
It’s tough to deal with copying of lineups. We don’t have a large issue with it at this time, partly because all our contests are 10 or 2 man but also because we’re simply not a large enough site.
Quote:
I totally agree with you that DFS in its current state is not overly complex. That said I still think it’s a fun and engaging game that users of all skill levels can enjoy. We’ve toyed with a number of variations on the standard game but haven’t come up with something yet that we're comfortable offering in the current legal climate. Doesn’t mean we’ve given up though. There’s nothing I enjoy more than dreaming up new game formats!
Randomizing some aspect(s) could allow each player to have a unique strategy set available. If you mean something roughly along those lines, then I agree it could be legally dicey.
Regarding complexity, I break DFS into two components:
(A) Generating an accurate set of predictions.
(B) Choosing an optimal strategy based on those predictions.
So far, DFS has been mostly about (A) due to both its nascency and game format while (B) is an easy exercise in solving (mixed) integer programs. While some of the methodology for (A) is incredibly complicated, the marginal gains on predictive accuracy are decreasing; eventually we'll reach a point where room for improvements in accuracy will be negligible. Maybe we're already getting close in some sports.
So I'd focus on adding complexity to (B) by approaching it as a computer science problem. For instance, I try to imagine changes the sites could make that would render my integer programming scripts intractable on modern hardware. The answer may be they can't, but here's an example I saw where I thought they could have done better:
One of the large sites recently introduced a new format where you roster a "captain" worth 1.5x normal fantasy production. It's fairly trivial to solve and also not terribly interesting imo. It's far more complex done this way:
Everything is the same as the standard game format except,
(1) Designate one of your players as the captain.
(2) You're awarded bonus points
if and only if the captain is the highest scoring player on your team.
It's a simple variation, but properly solving it is in another stratosphere in terms of computational complexity. You'd probably either need chance-constrained optimization or a stochastic recourse model in addition to functional forms for the probability distributions for scoring. In other words, even if every contestant had access to the most accurate projections available, it may still be possible to design a game such that no one could generate a provably optimal solution.