Thursday, August 19, 2010

Weekly accuracy assessment

I've got family in town right now so today's post is going to be brief - but awesome. So you all probably know that I do an annual accuracy assessment of experts' rankings, comparing how they've ranked players during to preseason to how those players performed at season's end. Many of you have asked if I'll be looking at the accuracy of weekly rankings, and unfortunately it hasn't been something I've been able to do due to the time involved...but now, my friends, there's a site that has filled that gap and is an incredible complement to the season-long accuracy assessment. Last week released their accuracy awards for weekly rankings and I for one am extremely excited about this. A big kudos to Andy Behrens from Yahoo for being the #1 overall most accurate expert with his weekly rankings from the 2009 season - for the full list of how the experts did, take a look at's Expert Accuracy Rankings - 2009.

Though just as important as the results is the methodology used - to see what they gathered and how they developed each expert's PAY (Prediction Accuracy Yield) and subsequent ranking, see the discussion here.

As a final cherry on top to this incredible project (and immense amount of work!), has also developed a Cheat Sheet Wizard, which is very similar to what I post each August with the average rankings position spreadsheet. I highly recommend you scope it out for updated consensus rankings...and drumroll includes PPR rankings for those who play in PPR leagues.

Basically, spend a bit of time on and you'll get lost in heavenly world of fantasy awesomeness. So have at it, let me know what you think!


dave said...

Hey Sara, I really appreciate the nice post and would love to hear any feedback from your members.

Now if I could just find some time to prep for my own draft, life would be really good. : )

Rogue33 said...

Very cool post. It's always nice to have more information that evaluates the experts. Makes it easier to find the sites to trust!

H00KemH02NS said...

Once again, I would throw an asterisk on all these rankings.

Ranked based on what scoring system? I'm not talking about PPR vs Standard, either. How much a passing TD nets has a whole lot to do with whether Tony Romo is more valuable than Jamaal Charles. How much are you punished for a fumble?

Just saying.

dave said...

Hey H00kem,

I think there are limitations to any study, but to ignore the results completely would suggest that there's zero insight to be had from analyzing this ranking data.

The scoring system and all the checks/balances/asterisks are available on our methodology page, which Sara was nice enough to link to.

At the very least, we're happy that the assessment has created some more discussion on the topic of "accuracy" in fantasy sports.

Randy said...

One comparison between the the weekly and the preseason, if I understand correctly... came in first for preseason rankings (Sara's analysis) but came in 19th (of 21) for weekly accuracy (FantasyPros' analysis).

What does that say? :-)

It's also worth noting that Yahoo! experts took 4 of the top 9 spots, overall. That's pretty significant.

dave said...

Hey Randy,

I won't speak for Sara, but from my perspective, I think you need to think of the two studies as complimentary to each other. And, I think it's important to note that the two studies use very different approaches to assessing accuracy.

I would say Sara's approach is more "pure accuracy" - did the expert hit or miss on the ranking of each player? While our assessment is more about trying to quantify the actual value that an expert's advice provides compared to his peers. There's no right or wrong, just two different ways to look at it.

On CBS specifically, we didn't run numbers for the draft rankings but I'm guessing they would have performed well, given how well they did with pure accuracy. There have been years when I personally have "killed the draft" and stunk it up during the season. Isn't it conceivable that CBS had the same kind of year?

Bottom line is, we need to continue to track everyone so we get a strong history of data. The more poker hands you play, the more likely you'll know whether you're really any good or not.

The Yahoo guys killed it. I'm happy for them because I think they get little respect...because they work for a big media company and not a "hard core" fantasy site.