[Inspirited]

How would you or do you go about choosing your team in a one week salary cap contest such as those held at fanduel.com or draftstreet.com?

I'm thinking you need to estimate each player's expected stats and then use these stats and the points per particular type of stat to get the estimated total points for each player. From here it reminds me of the 0-1 knapsack problem I've read about recently where you use dynamic optimization to solve it.

salary cap -> weight of the knapsack
cost of each player -> weight of each item
each player's expected points -> value of each item

?

[Inspirited]

I guess this would be a little different because you also have constraints on positions.

[Justin7]

Start with the cheapest player in each position.

For all players in each position, look at the "marginal points per dollar" added based on whatever projections you are using. Then pick the players that have the highest marginal points per dollar.

for example, if the worst WR cost $5 and you thought he was worth 2 points... A $15 WR that produces 7 points is really adding 5 points for $10... you're paying $2 per point. If you found any other position that was cheaper in terms of points per $, you would spend money on that position first.