Analyzing Fantasy Matchups - Fantasy Football Cafe 2014 Fantasy Football Cafe


Return to Football Talk

Analyzing Fantasy Matchups

Moderator: Football Moderators

Analyzing Fantasy Matchups

Postby mattb47 » Tue Sep 13, 2011 6:06 pm

I know that there are plenty of places that look at how much fantasy points per game a certain teams gives up to a certain position, but that information alone strikes me as being somewhat incomplete when you're a fantasy owner trying to decide between who to start and who to sit in a given week. I think much more useful would be to determine which teams give up more or less to a player than what they do on a regular basis throughout the season.

For example, let's say the you have a couple similarly skilled QBs fantasy wise and want to know which to start this week. Looking at strictly which team has given up the most fantasy points could prove misleading if say one defense has played 4 games and 3 of them were against Aaron Rodgers, Drew Brees, and Philip Rivers while the other defense got to play guys like Andy Dalton, Jason Campbell and Tarvaris Jackson. Defense B could be worse but "look" better by these numbers.

My idea (and I've already started putting together some of the numbers and things) is to analyze fantasy defenses based on how many points they give up relative to a player or team's average offensive output. So if Aaron Rodgers averages 25 FP/G and a defense gives up 22 FP to him, it's actually a plus for that defense rather than a negative and we can get more accurate projections on what players might do against different defenses.

Thoughts? It's probably going to be a bit of work and won't really be accurate at all until a few weeks in but I want to see if you guys think this would be as useful as I do before I do all the work on this one.
Image
mattb47
Hall of Fame Hero
Hall of Fame Hero

User avatar
Fantasy ExpertCafe WriterCafe RankerMock(ing) DrafterEagle EyeCafe MusketeerCafe Blackjack Weekly Winner
Posts: 14238
Joined: 29 Nov 2004
Home Cafe: Football

Re: Analyzing Fantasy Matchups

Postby LMack » Tue Sep 13, 2011 7:15 pm

When reviewing the Christopher Harris article on running a QBBC, I looked in to the fantasy points against statistic with decent depth. The biggest problem I found with it was the inability to take into account football players who have different skill sets but play the same position. If a defense did well against a pocket passer, that doesn't necessarily mean they can contain a quarterback who is dangerous on the ground. If a defense has physical corners and could key in on a slower, physical WR who is a RZ threat, that doesn't mean they can cover a burner out of the slot.
Image
LMack
Head Coach
Head Coach

User avatar
CafeholicFantasy ExpertCafe Musketeer
Posts: 1871
(Past Year: 32)
Joined: 15 May 2009
Home Cafe: Football
Location: Straight Cash Homey

Re: Analyzing Fantasy Matchups

Postby jake_twothousandfive » Tue Sep 13, 2011 7:16 pm

mattb47 wrote:I know that there are plenty of places that look at how much fantasy points per game a certain teams gives up to a certain position, but that information alone strikes me as being somewhat incomplete when you're a fantasy owner trying to decide between who to start and who to sit in a given week. I think much more useful would be to determine which teams give up more or less to a player than what they do on a regular basis throughout the season.

For example, let's say the you have a couple similarly skilled QBs fantasy wise and want to know which to start this week. Looking at strictly which team has given up the most fantasy points could prove misleading if say one defense has played 4 games and 3 of them were against Aaron Rodgers, Drew Brees, and Philip Rivers while the other defense got to play guys like Andy Dalton, Jason Campbell and Tarvaris Jackson. Defense B could be worse but "look" better by these numbers.

Wouldn't you encounter the same problem when looking at individual player averages? A low average could be due to difficult defensive matchups or an uncharacteristically bad day. A single outliner performance would take several weeks to stabilize to a somewhat reliable average. Just as the fantasy points a defense allows to a certain position is misleading when looking at a small sample size, the fantasy points a player scores could be equally misleading, which would consequently lead to misleading results based on your calculations.
Image
jake_twothousandfive
General Manager
General Manager

User avatar
Fantasy ExpertCafe WriterMock(ing) DrafterEagle EyeWeb Supporter
Posts: 2813
Joined: 22 Oct 2005
Home Cafe: Baseball

Re: Analyzing Fantasy Matchups

Postby mattb47 » Tue Sep 13, 2011 7:45 pm

jake_twothousandfive wrote:
mattb47 wrote:I know that there are plenty of places that look at how much fantasy points per game a certain teams gives up to a certain position, but that information alone strikes me as being somewhat incomplete when you're a fantasy owner trying to decide between who to start and who to sit in a given week. I think much more useful would be to determine which teams give up more or less to a player than what they do on a regular basis throughout the season.

For example, let's say the you have a couple similarly skilled QBs fantasy wise and want to know which to start this week. Looking at strictly which team has given up the most fantasy points could prove misleading if say one defense has played 4 games and 3 of them were against Aaron Rodgers, Drew Brees, and Philip Rivers while the other defense got to play guys like Andy Dalton, Jason Campbell and Tarvaris Jackson. Defense B could be worse but "look" better by these numbers.


Wouldn't you encounter the same problem when looking at individual player averages? A low average could be due to difficult defensive matchups or an uncharacteristically bad day. A single outliner performance would take several weeks to stabilize to a somewhat reliable average. Just as the fantasy points a defense allows to a certain position is misleading when looking at a small sample size, the fantasy points a player scores could be equally misleading, which would consequently lead to misleading results based on your calculations.


Well, first to answer LMack, I am actually breaking it down to the point where you can see the kind of passing vs rushing production for QBs and the rushing vs receiving production for RBs so after a few weeks some trends are sure to emerge in that respect. Also, I never imagine that something like this could cover every conceivable circumstance as football is extremely hard to predict, but it should be reasonably accurate if I do it right and I am planning on breaking the receiving points against down further than just straight points given up vs. WRs and going to at least looking at WR1 and WR2 production against.

I think you do bring up a good point too jake, I do think that it will take several weeks for the accuracy of the statistics in this type of approach to increase but I think in the end that it will end up balancing out and once we see some patterns emerging, we could easily decide to ignore the outlier performances for an individual player. Are some good players going to have bad performances against bad teams that could mess some things up sometimes? Sure, but patterns in the overall course of the season will emerge for sure and by the time we get into the 2nd half of the season we should have some pretty solid data going.
Image
mattb47
Hall of Fame Hero
Hall of Fame Hero

User avatar
Fantasy ExpertCafe WriterCafe RankerMock(ing) DrafterEagle EyeCafe MusketeerCafe Blackjack Weekly Winner
Posts: 14238
Joined: 29 Nov 2004
Home Cafe: Football

Re: Analyzing Fantasy Matchups

Postby mattb47 » Tue Sep 13, 2011 7:46 pm

Also, are there any suggestions on additional things to add to the equation that you think would make this a more accurate analysis?
Image
mattb47
Hall of Fame Hero
Hall of Fame Hero

User avatar
Fantasy ExpertCafe WriterCafe RankerMock(ing) DrafterEagle EyeCafe MusketeerCafe Blackjack Weekly Winner
Posts: 14238
Joined: 29 Nov 2004
Home Cafe: Football

Re: Analyzing Fantasy Matchups

Postby jake_twothousandfive » Wed Sep 14, 2011 6:30 pm

mattb47 wrote:I think you do bring up a good point too jake, I do think that it will take several weeks for the accuracy of the statistics in this type of approach to increase but I think in the end that it will end up balancing out and once we see some patterns emerging, we could easily decide to ignore the outlier performances for an individual player. Are some good players going to have bad performances against bad teams that could mess some things up sometimes? Sure, but patterns in the overall course of the season will emerge for sure and by the time we get into the 2nd half of the season we should have some pretty solid data going.

My point was that your main criticism of the existing approach (fantasy points given up to a certain position) doesn't take strength of schedule into account and is therefore incomplete. I agree, but you could use the same argument against the method you're suggesting. How do you see your method as an improvement in that regard?
Image
jake_twothousandfive
General Manager
General Manager

User avatar
Fantasy ExpertCafe WriterMock(ing) DrafterEagle EyeWeb Supporter
Posts: 2813
Joined: 22 Oct 2005
Home Cafe: Baseball

Re: Analyzing Fantasy Matchups

Postby mattb47 » Wed Sep 14, 2011 7:02 pm

I think it is an improvement because this system is not an analysis of the players themselves, it is an analysis of the defenses and what they give up. It is also a dynamic system in that it will continually update and change itself over the course of the season as it gains more data and such. For example, it may look like a defense did poorly against a player early on because he had a good day against them, but the more good days that player continues to have against other opponents over the course of the season, the better that first performance looks because the player average is always changing based on how a player continues to perform. So an outlying performance by an individual player will end up mattering very little because they are not the focus of the information, it is the defense's performance over time that gives you an idea of which ones are the better matchups and how they do comparatively with other defenses in the league against the same players.

Example:
Quarterback A plays Team B in week 1.

Quarterback A Stats:
Week 1: 25
Week 2: 17
Week 3: 29
Week 4: 35
Week 5: 34

So if you looked at things at Week 2, it would look like Team B was pretty weak against QBs because it gave up 25 points to a guy averaging 21 points per game. But as the season progresses and QB A puts up more and more good games, his average increases to 28 points per game in week 5 which points to Team B being better than average against QBs in this one player's case. Now when you take into account ALL the players that the team has faced, it no longer matters as much when a single player has a single really good or bad game because it will end up averaging out in the end.
Image
mattb47
Hall of Fame Hero
Hall of Fame Hero

User avatar
Fantasy ExpertCafe WriterCafe RankerMock(ing) DrafterEagle EyeCafe MusketeerCafe Blackjack Weekly Winner
Posts: 14238
Joined: 29 Nov 2004
Home Cafe: Football

Re: Analyzing Fantasy Matchups

Postby Cuffs » Wed Sep 14, 2011 9:34 pm

mattb47 wrote:Also, are there any suggestions on additional things to add to the equation that you think would make this a more accurate analysis?


That's an ambitious undertaking, but the more info you get in there is bound to make it more helpful from a fantasy standpoint than the simple "pts given up to rbs" numbers.

Among significant variables, though, are (1) were there injuries to key defensive players when, e.g., Grossman racked up 300 yards and 2 TDs; (2) were there key offensive players missing from...well, either team (if Manning played and Foster played, Schaub's stats likely would've looked different). Mendenhall's game was awful against Baltimore, and Baltimore is a great run D, but it never would've held Mendy that low but for the enormous hole Pitt was in, so they had to throw a bunch. How were Hasselbeck's stats/TDs against Jacksonville skewed by CJ's more limited role?

The list goes on -- pretty much any detail you can imagine that could play a role could warrant an asterisk for a given week. Still, like I said, even if you simply cover 8-10 variables, you will be compiling a more comprehensive set of stats than is available on the typical fantasy football sites. It's bound to be more accurate than what is commonly available now.

Good luck.
Cuffs
Defensive Assistant
Defensive Assistant


Posts: 354
(Past Year: 35)
Joined: 31 Aug 2010
Home Cafe: Football

Re: Analyzing Fantasy Matchups

Postby DraftDodger » Fri Sep 16, 2011 12:25 am

I once tracked defenses against the pass vs. WR1, WR2, WR3, TE, and RB and the system did a pretty good job of highlighting players who had a high potential to exploit a pass defense weakness in a given week. It was a lot of work each week, and in some cases I was kind of guessing which spot to put a receiver in, but it was eye opening too. In some cases, teams that were "horrible against the pass" were actually great at shutting down the #1 receiver, and teams that were "great defending the pass" consistently gave up scores to a TE or RB. The system was a little flawed though, as leopards don't change their spots. A TE who never catches passes doesn't suddenly get a 100 yds and 2 TDs vs a team that gives up lots of fantasy points to TEs.

No matter what system you apply, don't forget a smidgen of common sense.
DraftDodger
Head Coach
Head Coach

User avatar
Cafe Writer
Posts: 1400
Joined: 24 Aug 2002
Home Cafe: Football
Location: The Cafe

Re: Analyzing Fantasy Matchups

Postby WickedSmaat » Fri Sep 16, 2011 12:45 am

What if say for comparison's sake you have a rolling average, say the last 16 games for the team. So you'd come out with a 2011 column if someone wanted to see then you could go on to something like a rolling average of the last 4, 8, 16 games right next to it. Just gives some more numbers to analyze but you could also view something like a 16 game average to be more stabilized. Obviously there's more to it but I know I take in last week, last 2 weeks and last month when I look at guys for fantasy baseball when it comes to picking up players.
-Wicked
WickedSmaat
Hall of Fame Hero
Hall of Fame Hero

User avatar
Cafe RankerMock(ing) DrafterEagle EyeWeb SupporterPick 3 Weekly WinnerMatchup Meltdown SurvivorCafe Blackjack Weekly Winner
Posts: 8165
Joined: 1 Oct 2005
Home Cafe: Football
Location: Singing songs of satisfaction toooooo the wooooooorld..... (the world).

Next

Return to Football Talk

Who is online

Users browsing this forum: No registered users and 6 guests

Forums Articles & Tips Start & Sit Sleepers Rankings Leagues


Get Ready...
The 2014 NFL season kicks off in 4:07 hours
(and 37 days)
2014 NFL Schedule


  • Fantasy Football
  • Article Submissions
  • Privacy Statement
  • Site Survey 
  • Contact