High Variance Strategies in Multiplayer Snapdraft Contests (Part 1)

I've talked about this in posts and articles on other sites, but nobody ever seems to really understand what I'm getting at. High variance strategies are the way to go in multiplayer Snapdraft contests. The larger the contest, the more emphasis you should place on variance relative to your opponents, and the less you should focus on slight improvements in the expected score of your players. Most people approach multiplayer contests the same as they would approach a heads up contest - they try to pick the best team they can. This is the wrong way to go about things!

Let's use this week as an example. Of the players who have yet to play, my rough ranking of running backs would be as follows:

1. Frank Gore

2a. Adrian Peterson
2b. Cedric Benson

3a. Jonathan Stewart
3b. Chris Wells
3c. Maurice Jones-Drew

That's based on the ability of the player, their matchup, expected playing time, and expectations about whether their team will be running the ball a lot.

I need to select two running backs. Let's assume for now that I'll use a QB at my flex position.

Given that Benson is a third of the price of Peterson, obviously I should select Gore and Benson, right? Wrong! If this was a heads up contest, that would be the correct answer - Gore and Benson will provide the best performance for the price. But let's say I'm playing in one of the 101 player contests. My expectation (partly based on what I saw in the contests that locked yesterday) is that about 70% of my opponents will use Gore in their lineups, and around 40% will use Benson. Assuming that those picks aren't correlated in any way, that means 28% will use Gore+Benson. If I use those two players, even if they go off for 30 points each, the rest of my team is going to have to outperform 28 other people to win the prize!

Let's say instead that I use Stewart and Wells. I'd guess each of them will be on less than 10% of my opponents teams, and I'll probably be the only person out of the 101 to use both of them. Now let's assume they each go off for 30 points. I'm going to have a very good chance of winning. Are Stewart and Wells as likely to score 30+ each as Gore and Benson. No, of course not. But even if they've got half the chance of doing so, that gives me a better chance of getting the top overall score. Remember, with Snapdraft's winner take all prize structure, you're not being rewarded for high scores...you're only being rewarded for EXTREMELY high scores relative to your opposition. And the best way to have that happen is to make picks that differ from your opponents.