24 Aug 2010

# KUBIAK + Wisdom of crowds

I've used KUBIAK almost exclusively in the past two years for my fantasy leagues and I've gotten a sense of where it is effective: in identifying players to avoid and players who are worthy of a late-round pickup. However, in the past I've ignored the green/yellow/red ratings, and this year I want to incorporate them. I realize that I have invested too much faith in KUBIAK's projections.

The "Wisdom of Crowds" articles intrigue me, although I suspect that the sort of people who offer their ideas on Twitter are the sorts of people who have read FOA and they might inadvertently just parrot the FOA/KUBIAK guesses on how the players will perform.

Nevertheless, I want to incorporate the Wisdom of Crowds information into my KUBIAK spreadsheet - specifically, I want to use the Wisdom of Crowds information in combination with the color rating system in order to moderate KUBIAK's unconvential wisdom. The way I envision this is weighing the Wisdom of Crowds guesses more as the risk rating goes from green to red. For example, WoC will be weighed .25 on Green, .5 on Yellow, and .75 on red.

I'd probably only use this sort of setup as long as it took me to get my starting players, and then rely exclusively on KUBIAK later in the draft.

Does anyone have ideas about how I can incorporate this into the spreadsheet? It'd be great if it was built-in to KUBIAK as an option on the "Scoring" tab!

Posted by: socctty on 24 Aug 2010

1 reply , Last at 24 Aug 2010, 7:25pm by Dan

1
##### Re: KUBIAK + Wisdom of crowds
by Dan :: Tue, 08/24/2010 - 7:25pm

You can do this in Excel with the "if" function. For instance, if column A has the KUBIAK projection, column B has the WoC projection, and column C has the risk rating, then the formula that you'd want (for the player in row 1) is:

=IF(C1="red",0.25*A1+0.75*B1,IF(C1="green",0.75*A1+0.25*B1,0.5*A1+0.5*B1))

In words, that formula says: if the risk is red then use the .25/.75 weighting, otherwise check if it's green and if it is use the .75/.25 weighting, otherwise use the .5/.5 weighting.