Blog Entry

More Thoughts on Predictive Content

Posted on: October 26, 2009 4:37 pm
I had no idea that the topic about the Guru was going to be such a polarizing subject.  I absolutely love the fact that you (the collective fantasy community) can put so much thought into not only whether you preferred one way over another way, but also coming up with totally different ways to derive a data point that would be relevant in making your lineup decisions.  I even enjoyed the fact that people took the time to criticize my spelling. For the record I don't spell, grammar, or even fact-check these blogs.  My main goal is to make it through each blog without swearing (a challenge for me on a sentence to sentence basis when I speak). 

A few main themes I have gotten either through the feedback on this blog or feedback from other forums:

1.  The Guru (or any predictive number for that matter) being on the Set Lineup page sways people's decisions.

This may sound strange but I don't think I understood the effect that putting a predictive number on the lineup page has on people until I read this column.  We've had it as a part of our products for so long that it's easy to forget that people actually use this as the sole basis for setting their lineups.  I have always looked at this number sort of like an 'over/under'.  It's a number that's out there, determined by some "expert" and it's on me as the team owner to figure out whether each player will exceed that number or fail to reach that number.  After all that is supposed the be the allure of Fantasy Sports...that I, as the team owner ultimately decide what each player will do each week.  If you are like me, you do not want to rely on someone else, even someone who gets paid to analyze this stuff, to make the final decision on who goes into your lineup.  But I am now realizing how tempting it can be to see a big number next to a player and want to put them in because of that. 

I saw the idea tossed around that maybe there should be an option to have the Guru turned off, or at least not show it on the Set Lineup page.  I even saw a comment that the lineup page is cluttered enough already and the Guru doesn't need to be on there.  We already have plans to tweak the Set Lineup page, and all of this feedback will be considered. 

2.  Both ways of doing the Guru, either by formula or by human, are inadequate.  Users seem to want additional data points. 

This one doesn't really suprise me at all.  And I want to give you all of the data points you can handle.  I've mentioned before the idea of a 'Consistency Factor', which would come in handy when trying to decide between between two players who average 10 points a week, but one guy is consistently in the 8-12 range while the other guy is either scoring 20 points or 0 points.  For some, they may want the guaranteed 8-12 points, but for others, they need the risk/reward play of a guy capable of blowing up for 20+ points. Regardless, it's relevant data. 

So, what else would you use as a data point for making your lineup decisions? 

3.  Users seem to want to see a "rating" of the predictions.  Meaning that you'd like to see how close/far the predictions have been overall.

I love this idea.  I've always believed that there isn't nearly enough accountability in the Fantasy prediction field.  Plus, Dave Richard and Jamey Eisenberg are running out of reasons to dislike me, so I should provide them with a few more.  Seriously though, I think we should do this not only for our writer predictions, but for the formula as well.  I'll have to see if I can get buy-in from our editorial team. 

I really appreciated all the feedback I got from that last post and would love to see everyone keep it coming.  There are some big changes on the horizon not only in Fantasy but for all of and this has proved to be a great source for great ideas.   


Since: Dec 29, 2009
Posted on: April 19, 2010 12:15 am

More Thoughts on Predictive Content

Huh, okay then... cool!

Since: Jan 2, 2010
Posted on: January 2, 2010 8:44 am

More Thoughts on Predictive Content

A piece of data that would be great to have would be home v. away statistics. As we all know, the home team generally has a clear advantage in the NFL, winning around 70% of all games, but that isn't always as true for individual players and defenses/ST. I noticed that rankings about the defenses a player was facing were available in a couple more places this year, but more data regarding home/away splits would be a big help.

Also, not entirely germane to this topic, but since I'm here - our league always is playing with the waiver process as we are always looking for what's most fair, most competitive, etc. This season our league used the FAAB process for the first time. It seemed most of our league liked it, but one critcism across the board was that it didn't afford much game day felixibility, unlike our prior waiver process.

Example: Brian Westbrook is a 'game-time' decision and you find out that he's a scratch 15 minutes before gametime. Under the FAAB there's no way to pick a quick sub from waivers, while under the old waiver process it was possible. While there are arguments that you should have subs on your bench anyway, that's a choice each league should be able to make for themselves. Maybe there is way to allow each league greater freedom to tailor the free agent/waiver process to their wants, the same way roster sizes and shapes seem to have a good bit of flexibility from league to league?



Since: Sep 14, 2008
Posted on: December 29, 2009 11:09 pm

More Thoughts on Predictive Content

The Guru rating is not a good way of selecting your players. How often have you seen a player put up video game numbers only to get a high guru rating the next week. But wait. The player is facing the toughest run defense in the NFL, how are they going to rack up 21 points ?

Since: Aug 28, 2008
Posted on: December 10, 2009 5:17 pm

More Thoughts on Predictive Content

Theme number 3 sounds great.  But rather than a percentage for accuracy given a certain player, I think it would be nice to see a prediction history like this:  Player X was underestimated for the past 3 weeks OR Player X was overestimated for the past 2 weeks.  It would also be nice to include predictions on the player page under Fantasy Trend.

Since: Nov 18, 2009
Posted on: November 18, 2009 1:53 pm

More Thoughts on Predictive Content

" So, what else would you use as a data point for making your lineup decisions?  "

Good question. When making lineup decisions, one of the main factors I take into account is matchups. I want to know how the opposing team has done against this position. For example, Kurt Warner is playing against the Jets and Matt Schaub is playing against the Browns. I want to know the following:

1) avg # of fantasy points that the Jets and Browns give up to QBs. Ideally it will use my league's scoring system.
2) avg # of yards that the Jets and Browns give up to QBs.
3) total # of touchdowns that opposing QB's have thrown against them.
4) Their overall rank in the NFL. For example, the Jets are ranked 3rd against the pass and the Browns are ranked 27th.

I currently get #2, #3, and #4 by going to

eam/_/stat/total/position/defense and clicking on the relevant headings, "Passing, Rushing, Receiving, etc".

For kickers, I want to know:
1) avg # of fantasy points the opposing defense gives up to kickers. Ideally it will use my league's scoring system.
2) total # of XP's scored on that defense
3) total # of FG's scroed on that defense

For Defense/Special Teams:

1) avg # of fantasy points the opposing offense has scored. Ideally, use my league's scoring system.
2) avg # of total yards the opposing offense racks up each week.
3) avg # of total points the opposing offense scores each week.
4) avg # of sacks, interceptions, fumbles the opposing offense takes each week.

Thanks for asking.

Since: Aug 9, 2009
Posted on: November 8, 2009 4:39 am

More Thoughts on Predictive Content

Don't underestimate the power of the guru!  In our league the guru could potentially hold a lot of power.  Our rules say:

"If an owner leaves anytime during the season, the Commissioner will designate the team a “league team” and the lineups will be set each week by the Commissioner solely by following the website’s “guru” predictions for best lineup."

This was the only way we could come up with to eliminate any potential conflicts of interest if this situation ever occurred.

Since: Aug 10, 2006
Posted on: November 6, 2009 1:11 pm

More Thoughts on Predictive Content

I appreciate all of the really well thought out commentary.  This is exactly the type of feedback I was hoping to get when I started doing this blog.  It's taken me a while to find the content that really spurs this level of thinking, but now that I've found it, expect more of the same to come.  Great stuff here. 

Since: Nov 5, 2009
Posted on: November 5, 2009 4:42 pm

More Thoughts on Predictive Content

Mike, thanks for updating this topic, great to know the CBSonline Product team is actively working to improve the product design.  
One thing you might consider is the gap between Advanced Users and regular users.....those of us who want additional data points and resulting product complexity vs. players who may want less.  The ability to customize pages like Set Line Up with radio-buttons/boxes to select data points each individual might like to include is one common way to accomplish that goal.  I may want to see 3 week trend (FPS), History vs. Opponent, Guru fps prediction (however you ultimately decide to do that), defensive match up strength, weather, etc. all as elements of deciding whether or not to start a player.  
Actually, CBS Fantasy Football (particularly with the Accuscore 3rd-Party Product offering through Fantasy Scout) houses pretty much all of the data points most users want to use to make their start/sit decisions, they just aren't all in one place.  Right now I spend a lot of time searching around to find all the data points I want to help derive my decisions.  I do check the Guru forecast, but I also navigate over to the Pundit (Jamey, Dave) player rankings, I check out Accuscore/Fantasy Scout, I navigate to the Defensive Matchup Tracker, player/team depth charts (in case injuries to other players impacts my player's potential touches), all of the stuff is in there one way or the other, I just have to dig around a bit to find it all.
If the Set Line Up screen had buttons that allow'd me include all of (or at least MORE of) these elements as columns in addition to the Guru projection, I'd suspect many players would adopt and use them.  And I know this sort of thing goes over VERY poorly in communities like this, but I for one would be willing to PAY to have product offerings like that....I already pay for Fantasy Scout/Accuscore, and I'd pay for any product like that which provided data estimates.  Fantasy Scout is often 'at odds' with Guru or Jamey/Dave, but of course Accuscore doesn't include the 'human element' gut feel that Jamey might have....Fantasy Scout predicted Favre would struggle against Green Bay last week based on the paper-match-up, but Jamey predicted him as the #1 Fantasy Start based on the fact Favre would want to perform well at Lambeau and against his old team.  In that case I trusted Jamey and it served me very well.  
The point is, I had to hunt around to find all of those data points and get to my decision.....I know, Mike, you guys want eyeballs and time-spent-on-site metrics to be high so you can serve more Ford F-150 banner ads and all, so in that sense you WANT me hunting around to find all the data points....but giving us all a very useful tool will ensure user loyalty and customer satisfaction is higher long term.  Invest in us, we'll stay loyal to you.

It's great to know how Guru works.  And Fantasy Scout/Accuscore products are cool, but aren't 'integrated'.  Bring it all together, give users more flexibility on including it in their page layouts, you'll have very happy me, happy enough to pony-up some cash to pay for the improved, advanced product.

Cheers and thanks for all the hard work.

Since: Sep 30, 2009
Posted on: November 3, 2009 12:37 am

More Thoughts on Predictive Content

On Point 1:

Part of the problem may be the big bold font in Live Scoring > Preview, which seems so decisive. Maybe it should be in little skinny italics with ellipses and a question mark after it, just to look a little more uncertain.





That way people might be a little less swayed.

Though I suspect many of them are simply looking for assurance. They can't be looking for accuracy; within just a few weeks it's obvious they won't get it.


Since: Oct 29, 2009
Posted on: November 1, 2009 5:51 pm

More Thoughts on Predictive Content

I love the fact that there is a Guru or prediction of the what each play might score from week to week.  It honestly is the biggest reason i like cbs better than most other fantasy sites, including the sports GOD ESPN, who seems is more interested in how it looks (players pictures and pretty pictures lol) than useful info.  My only mistake is that i was using the guru's number rating too much for who to start and who not to play instead just playing my best players.  Ive come to realize that great players usually play great no matter what.  This being said the other thing i think comes into play, obviously, is the players oppenent.  That is one thing i wish cbs had easier access to: meaning that i want to see how well the defenses are doing aganist the run, pass, and that speafic player.  I am able to find this info on your pg but it a little harder than id like.  i dont kno maybe i am nitpicking but i do love the guru number bc im sure or at least i hope they/you have a little bit more inside info. than being that im not paid to find out this info. and you.  So while im at work doing which has nothing to do with finding out how well Ronnie Brown is going to do against the Jets defense in WK 8, i kind of hope that the GURU is gathering all the info. it can get and make a EDUCATED decision so when i get home from work i have some sort of basis on who to start on sunday. Dont get me wrong like i said before i love the guru it makes my week leading up to sunday very intersting by seeing who im gonna start, sit, release, possibly even trade because of how you think they are goin to do.  Another thing i wish was changed is the ranking of the players i dont think i updated after the season starts.  Also the predictions for season arent changed after the season starts and the players season is going worse/better than expected.  I really use these numbers when i decided to drop or add a player because i would like to know where he ranks and how well someone thinks he might REALLY do the rest of the season.  But please dont rid of the guru maybe, like you said "tweak it a little" by possibly getting more people involved and pooling their info and opinions.

The views expressed in this blog are solely those of the author and do not reflect the views of CBS Sports or