huddle ease of schedule flawed?

[dominatorr]

I was checking out the san diego charger's ease of schedule for quarterbacks and it shows that denver is one of the "BAD" match-ups for this year ....this didn't seem right to me.............i checked out nfl.com and the 2010 stats for denver's passing D shows they were ranked 24th in the # of passing touchdowns they gave up and they were ranked 25th in passing yards given up .......

 

so if you go by last years stats, how the heck can this be a "bad" match-up for san diego quaterbacks????????????

 

maybe the scoring is different for quaterbacks in the huddle's system but yards and touchdowns are the biggest determinants for how many points qb's score in my league

[delfamdelfam]

it's a bad matchup for San Diego Chargers because of Champ Bailey.

[dominatorr]

I was checking out the san diego charger's ease of schedule for quarterbacks and it shows that denver is one of the "BAD" match-ups for this year ....this didn't seem right to me.............i checked out nfl.com and the 2010 stats for denver's passing D shows they were ranked 24th in the # of passing touchdowns they gave up and they were ranked 25th in passing yards given up .......

 

so if you go by last years stats, how the heck can this be a "bad" match-up for san diego quaterbacks????????????

 

maybe the scoring is different for quaterbacks in the huddle's system but yards and touchdowns are the biggest determinants for how many points qb's score in my league

 

 

another example is the atanta d which is a "bad" matchup for quarterbacks...... in 2010, the atlanta passing d was 22nd in passing yards allowed and exactly in the middle of the league for passing touchdowns given up

[TitansFan]

Let me explain how the ease of schedule works. It takes into account how teams play on the road compared to how they play at home, not just their overall defense. It splits up the two into two separate matchups, resulting in 64 possible matchups (65 if you include the bye). The 20 worst (least points given up) and 20 best (most points given up) matchups are graded as bad or good respectively, and everything in between is considered neutral.

 

Let's take a look at your example of Denver. In 2010, they gave up an average of 20.8 points per game to QBs when playing in Denver - enough to make the "good" list - and 14.4 ppg when playing on the road - enough to make the "bad" list. This means that when any team faces the Broncos in Denver it will be considered a good matchup because they are on the list of 20 best QB matchups at home, but when they face the Broncos away from Denver it will be considered a bad matchup because they are on the list of 20 worst QB matchups on the road.

 

I hope that helps.

[Brentastic]

What TitansFan said. DMD enhance the EOS analysis a few years back to distinguis between home and away matchups because he recognized that some teams may play better or worse on the road or at home. It was a good observation by DMD and the huddle's EOS is 2nd to none.

 

During the season I find it useful to use the Season Strength of Schedule since it adjusts weekly to the actual current season defensive stats.

[DMD]

It shows it is not only not flawed, but more accurate than what other places do when they ignore venue.

 

The table on the top right of that page shows @DEN was easy at 20.8 and yet playing the visiting DEN was one of the worst at 14.4.

 

Rare that happens but it does. It is purely statistical. It cannot be flawed.

[dominatorr]

It shows it is not only not flawed, but more accurate than what other places do when they ignore venue.

 

The table on the top right of that page shows @DEN was easy at 20.8 and yet playing the visiting DEN was one of the worst at 14.4.

 

Rare that happens but it does. It is purely statistical. It cannot be flawed.

 

first of all, rarely does a team perform better on the road than they do at home......so right there one would think something is fishy....one obvious way this kind of anomaly can happen is if there is a small sample size .....by only using 8 games (home or visiting) instead of using all 16 games, the sample size is cut in half.......playing just a few crappy offenses using that small sample size can cause huge distortions....and this is without mentionsing 2 more games are dropped in the huddle's calculations (the best and the worse game) to make the sample size even less .......

 

bottom line is denver was terrible against the pass last year when they played a decent team (home and away) and in my opinion should not be considered a "BAD" match-up for this year based on last year's stats.... denver played arizona and oakland their last 2 away games last year and these 2 horrible offenses scored 82 combined points againt this "awesome" away denver defense.....

 

all i got to say is basing an entire defense on 8 games or less from the season before can give misleading results..........and btw, statistics CAN be flawed or the stats misinterpreted if the sample size is insufficient.......

 

I'm not trying to slam the huddle...I am just throwing out there a potential flaw i see in the analysis solely for discussion and possible consideration when future ease of schedules are made up.......I may be totally off base here

 

btw, denver's passing d may actually be better this year with dumervil coming back but this is beside the point

Edited August 7, 2011 by dominatorr

[Brentastic]

first of all, rarely does a team perform better on the road than they do at home......so right there one would think something is fishy....one obvious way this kind of anomaly can happen is if there is a small sample size .....by only using 8 games (home or visiting) instead of using all 16 games, the sample size is cut in half.......playing just a few crappy offenses using that small sample size can cause huge distortions....and this is without mentionsing 2 more games are dropped in the huddle's calculations (the best and the worse game) to make the sample size even less .......

 

bottom line is denver was terrible against the pass last year when they played a decent team (home and away) and in my opinion should not be considered a "BAD" match-up for this year based on last year's stats.... denver played arizona and oakland their last 2 away games last year and these 2 horrible offenses scored 82 combined points againt this "awesome" away denver defense.....

 

all i got to say is basing an entire defense on 8 games or less from the season before can give misleading results..........and btw, statistics CAN be flawed or the stats misinterpreted if the sample size is insufficient.......

 

I'm not trying to slam the huddle...I am just throwing out there a potential flaw i see in the analysis solely for discussion and possible consideration when future ease of schedules are made up.......I may be totally off base here

 

btw, denver's passing d may actually be better this year with dumervil coming back but this is beside the point

So you want DMD to take a statistical analysis and then inject a whole bunch of subjectivity into it? Gotcha

 

The fact of the matter is this: unlike the cheat sheets or the player projections - both of which are subjective analysis- the EOS is statistical or factual analysis. It's not meant to be bent or skewed to derive the results you think should be there. It is what it is. Very plain and simple. It's up to the user to take that information and make whatever assumptions you wish to make. But to ridicule DMD or anyone else over what is simply just factual information - well, that's just dumb.

[dominatorr]

So you want DMD to take a statistical analysis and then inject a whole bunch of subjectivity into it? Gotcha

 

The fact of the matter is this: unlike the cheat sheets or the player projections - both of which are subjective analysis- the EOS is statistical or factual analysis. It's not meant to be bent or skewed to derive the results you think should be there. It is what it is. Very plain and simple. It's up to the user to take that information and make whatever assumptions you wish to make. But to ridicule DMD or anyone else over what is simply just factual information - well, that's just dumb.

 

funny how simple minds resort to calling names when they disagree with someone .....how, Mr. brentastic, does preferring to have 16 games being used to get a result instead of 8 mean that i must want subjective results and mean i want the results skewed to get the result I had in mind.....ru serious??? I suspect you didn't do well on the reading comprehension part of the SAT's

 

obviously dmd is trying to add more variables to get better results by looking at how teams do at home and on the road.....usually the more variables that are considered, the better the results..........,I just think in this case, it may be the opposite .........and it's all because having twice as many games to put into the calculation (16 instead of 8) trumps the added variable of home and away results in my opinion and i think denver is a good example of the distortion possible using only 8 games.......

Edited August 8, 2011 by dominatorr

[Brentastic]

funny how simple minds resort to calling names when they disagree with someone .....how, Mr. brentastic, does preferring to have 16 games being used to get a result instead of 8 mean that i must want subjective results and mean i want the results skewed to get the result I had in mind.....ru serious??? I suspect you didn't do well on the reading comprehension part of the SAT's

 

obviously dmd is trying to add more variables to get better results by looking at how teams do at home and on the road.....usually the more variables that are considered, the better the results..........,I just think in this case, it may be the opposite .........and it's all because having twice as many games to put into the calculation (16 instead of 8) trumps the added variable of home and away results in my opinion and i think denver is a good example of the distortion possible using only 8 games.......

First of all, I'm the first huddler to accuse others from lacking in reading comprehension - so stop stealing my line!

 

2ndly, DMD does use all 16 games. And again, he's simply providing the actual stats. Nothing more, nothing less. It's up to the user to dig deeper if you want to maximize the information for your fantasy benefit.

 

As for name calling, I had to go back and re-read my post to see what you meant. I guess you could argue that I called you dumb but technically I was saying the act of ridiculing the stat compiler (in this case DMD) was dumb. But whatever, semantics. Later, noob.

[dominatorr]

First of all, I'm the first huddler to accuse others from lacking in reading comprehension - so stop stealing my line!

 

2ndly, DMD does use all 16 games. And again, he's simply providing the actual stats. Nothing more, nothing less. It's up to the user to dig deeper if you want to maximize the information for your fantasy benefit.

 

As for name calling, I had to go back and re-read my post to see what you meant. I guess you could argue that I called you dumb but technically I was saying the act of ridiculing the stat compiler (in this case DMD) was dumb. But whatever, semantics. Later, noob.

 

allright, this is my last post on this subject ........this is from the huddle " Taking the eight home games and eight road games, each had their highest and lowest games removed to eliminate uncharacteristic extremes. Then the per game average fantasy points allowed were determined from the six remaining home and six remaining road games"

 

....in other words, only 6 games are used (not 16 ) (i kinda said 8 in the previous posts but this only bolsters my argument) to come up with whether denver's d is a "good" or "bad' match-up for each team they play this year .....

[The Wolf]

allright, this is my last post on this subject ........this is from the huddle " Taking the eight home games and eight road games, each had their highest and lowest games removed to eliminate uncharacteristic extremes. Then the per game average fantasy points allowed were determined from the six remaining home and six remaining road games"

 

....in other words, only 6 games are used (not 16 ) (i kinda said 8 in the previous posts but this only bolsters my argument) to come up with whether denver's d is a "good" or "bad' match-up for each team they play this year .....

 

Why? Nothing wrong with a good debate. Just argue facts and don't make it personal.

[Brentastic]

allright, this is my last post on this subject ........this is from the huddle " Taking the eight home games and eight road games, each had their highest and lowest games removed to eliminate uncharacteristic extremes. Then the per game average fantasy points allowed were determined from the six remaining home and six remaining road games"

 

....in other words, only 6 games are used (not 16 ) (i kinda said 8 in the previous posts but this only bolsters my argument) to come up with whether denver's d is a "good" or "bad' match-up for each team they play this year .....

Ok, dominatorrrrr I kinda see your point. If anything look at it like this: The huddle is giving you too much information by taking the 12 game sample and grouping that sample either by Home or Away. 2 groupings from a sample size of 12. The detail is there to create any groupings the analyst deems necessary. In this case, the Huddle decided to group by Home and Away, presumably because they saw relevance in that grouping. I now see why you believe that grouping is flawed and you might be on to something - I haven't looked into it that much. Maybe a more ambiguous delivery of this data would be for the Huddle to provide just the raw detail? But that's sort of like asking the Fed to open their books

[ajh2]

Sorry, dominatorr but all eight games were used.

 

Just answer this simple question: Which game was the highest for thte Denver Defense and which game is the lowest for the Denver Defense?

 

DMD calculated each game for the Denver Defense and then removed the highest and lowest game to prevent an extremely good game and an extremely bad game for squewing the average for the Defense. This is correct and sound statistical practice. Why DMD remove the high and low is to accurately represent show a team's defense plays at home and on the road. This does towo things to the data provided.

 

1.) Reduces the variance of the data used

2.) Produces a better picture of how the temas Defense played

 

In other words, the EOS is extremely accurate in predicting how a player is going to perform against a Defense with very few notable exceptions. That is the few true studs like Peyton Manning, Tom Brady, etc.

 

[OCSkins]

If I may butt in here.....I think there's a simple misunderstanding. It sound to ME that dominatorrr thinks that the highest & lowest games were thrown out FROM EACH GROUP--home & away. That's why he is saying that only 6 games were used in each bracket, for a total sample size of 12 games.

 

It is my understanding that the highest and lowest games OVERALL are omitted as outliers from the sample size, so the true sample size is 14 games.

 

Right? Or should I start drinking so I can figger it out..... =)

[Brentastic]

If I may butt in here.....I think there's a simple misunderstanding. It sound to ME that dominatorrr thinks that the highest & lowest games were thrown out FROM EACH GROUP--home & away. That's why he is saying that only 6 games were used in each bracket, for a total sample size of 12 games.

 

It is my understanding that the highest and lowest games OVERALL are omitted as outliers from the sample size, so the true sample size is 14 games.

 

Right? Or should I start drinking so I can figger it out..... =)

Dominator is right, the sample size is 12 games. Huddle takes away highest and lowest games from each group.

[tazinib1]

But that's sort of like asking NFL owners to open their books

 

 

fixed

 

 

[dominatorr]

Dominator is right, the sample size is 12 games. Huddle takes away highest and lowest games from each group.

 

saying dmd uses 12 games is like saying he uses 384 games (12 x 32) which is how many games he actually uses if you include ALL the teams.....yes he uses 12 denver games but only uses half of those for a particular matchup.......

 

for example, say I'm wondering if san diego's quarterbacks at home this year against denver's passing d (on the road) is a good or bad matchup, dmd doesnt use the 6 home denver games for this calculation, he only uses the 6 away games denver had last year.....no additional games are used....................thus 6 games are used in total

 

 

so again I think you lose too many games that go into the calculation by breaking them down into home and away games.......a couple strong or weak teams that a particular team plays against can distort the results you want if the sample size is so small (6)............if somehow you could break them down into home and away games and still have a large sample size, then obviously breaking them down would be better

 

 

so the argument in a nutshell is which is a better predictor of how "good" or "bad" a particular team's passing or running d will be in 2011 :

(1) using 12-17 games (using home and away games together) from 2010 thus having a larger sample size but not having the extra variable thrown in

 

(2) using 6 games from 2010 but having the extra variable thrown in

 

Personally, I choose #1 but maybe i'm wrong

.

Edited August 8, 2011 by dominatorr

[The Wolf]

For the sake of this discussion, I wonder how different the rankings would look with the full 16 games factored into the study as opposed to 75% of the season.

[BearBroncos]

Taking out your highest and lowest factors is an analytical norm to ensure your eliminating anomalies as much as possible. Not a pure science, but it's a staple in analytical studies.

[DMD]

saying dmd uses 12 games is like saying he uses 384 games (12 x 32) which s how many games he actually uses if you include ALL the teams.....yes he uses 12 denver games but only uses half of those for a particular matchup.......

 

for example, say I'm wondering if san diego at home this year against denver is a good or bad matchup, dmd doesnt use the 6 home denver games for this calculation, he only uses the 6 away games denver had last year.....no additional games are used....................thus 6 games are used in total

 

 

so again I think you lose too many games that go into the calculation by breaking them down into home and away games.......a couple strong or weak teams that a particular team plays against can distort the results you want if the sample size is so small (6)............if somehow you could break them down into home and away games and still have a large sample size, then obviously breaking them down would be better

 

Using all eight games is done. Taking the highest and lowest out of the equation means they are still represented by virtue of determining what "middle" six are used. .

 

I take umbrage with the word "flawed" since there is nothing wrong or incorrect. It is exactly what it is. ANd it is entirely indicative of what happened.

 

QB's

 

@DEN

 

10 KC 39.5

3 IND 28.3

12 STL 27.4

16 HOU 19.5

7 OAK 18.2

17 SD 15.7

2 SEA 15.7

6 NYJ 13.9

 

average = 22.3 average with hi/lo removed = 20.8

 

DEN

 

11 SD 27.7

1 JAC 20.5

15 OAK 15.9

13 KC 13.8

8 SF 13.8

4 TEN 12.7

5 BAL 9.8

14 ARI 7.3

 

average 15.2 average with hi/loi removed = 14.4

 

Average entire 16 games - 18.7, average with hi/lo removed 18.0

 

So, using all 16 games shows Denver to be an average defense - not good or bad.

 

But instead of 18.0 points per game overall, it is 2.8 points more at Denver where the Broncos play better and opponents throw more and it is 3.6 points per game less than average when Denver is on the road. 6.4 points difference in all.

 

18 point overall average. Games in Denver had 5 games over that mark. Games on the road had 6 games less than that mark. I would contend that is more accurate than using all 16 games which washes out a statistically significant difference. And using 6 games instead of 8 games made it more accurate as well since it erased the skewing of the KC game (39.5) and the ARI game (7.3).

 

In the end, this is just a tool and your own wisdom how to apply is what matters. The beauty of it though is that it is purely statistical and objective from exactly what happened. Add your own subjectivity what it means.

[SEC=UGA]

I think we should expand the regular season to 64 games, 32 home, 32 away, for each team... By doing this dominatorr could have more info from a larger sample size and not accuse DMD of not using a large enough sample. Really, it's a win-win situation. We get more football and dominatorr will have more statistics available so that he can form an opinion.

[DMD]

I think we should expand the regular season to 64 games, 32 home, 32 away, for each team... By doing this dominatorr could have more info from a larger sample size and not accuse DMD of not using a large enough sample. Really, it's a win-win situation. We get more football and dominatorr will have more statistics available so that he can form an opinion.

 

Problem with that is that you would have too small a sample since by week 11, some teams will have only played 4 home and six away and one bye. So your sample on home games would be open to the impact of one really big or small game. Trust me - we considered it and looked into it but by the time there is reasonable accuracy to the tool, it is later in the season and it could be inaccurate early in the year.

[The Wolf]

I think we should expand the regular season to 64 games, 32 home, 32 away, for each team... By doing this dominatorr could have more info from a larger sample size and not accuse DMD of not using a large enough sample. Really, it's a win-win situation. We get more football and dominatorr will have more statistics available so that he can form an opinion.

 

Get it on the agenda for the next round of labor negotiations in 2021!

[dominatorr]

Using all eight games is done. Taking the highest and lowest out of the equation means they are still represented by virtue of determining what "middle" six are used. .

 

I take umbrage with the word "flawed" since there is nothing wrong or incorrect. It is exactly what it is. ANd it is entirely indicative of what happened.

 

QB's

 

@DEN

 

10 KC 39.5

3 IND 28.3

12 STL 27.4

16 HOU 19.5

7 OAK 18.2

17 SD 15.7

2 SEA 15.7

6 NYJ 13.9

 

average = 22.3 average with hi/lo removed = 20.8

 

DEN

 

11 SD 27.7

1 JAC 20.5

15 OAK 15.9

13 KC 13.8

8 SF 13.8

4 TEN 12.7

5 BAL 9.8

14 ARI 7.3

 

average 15.2 average with hi/loi removed = 14.4

 

Average entire 16 games - 18.7, average with hi/lo removed 18.0

 

So, using all 16 games shows Denver to be an average defense - not good or bad.

 

But instead of 18.0 points per game overall, it is 2.8 points more at Denver where the Broncos play better and opponents throw more and it is 3.6 points per game less than average when Denver is on the road. 6.4 points difference in all.

 

18 point overall average. Games in Denver had 5 games over that mark. Games on the road had 6 games less than that mark. I would contend that is more accurate than using all 16 games which washes out a statistically significant difference. And using 6 games instead of 8 games made it more accurate as well since it erased the skewing of the KC game (39.5) and the ARI game (7.3).

 

In the end, this is just a tool and your own wisdom how to apply is what matters. The beauty of it though is that it is purely statistical and objective from exactly what happened. Add your own subjectivity what it means.

[/quote

 

i agree...flawed was the wrong word