Welcome to PatsFans.com

The Dan Marino Passing Record Chase Breakdown

Discussion in 'PatsFans.com - Patriots Fan Forum' started by Tyler.Durden, Nov 29, 2011.

Thread Status:
Not open for further replies.
  1. Tyler.Durden

    Tyler.Durden Rookie

    Joined:
    Mar 12, 2007
    Messages:
    586
    Likes Received:
    2
    Ratings:
    +2 / 0 / -0

    I've run a basic projection for TFB, Brees and Rodgers based upon QB passing yards vs opponents defense passing yards per game.

    Obviously quite basic, doesn't take into account away games, playing in a dome, weather, injuries.

    The Marino Record is 5084.
    I project:
    Brady will throw 5331.3
    Brees will throw 5351.3
    Rodgers will throw 5051.3

    * Note: Bold team indicate a home game for the QB
    BRADY
    [​IMG]

    BREES
    [​IMG]

    RODGERS
    [​IMG]

    Any ideas how to sharpen the projection?

    Cheers guys
    Last edited: Nov 29, 2011
  2. Metaphors

    Metaphors Rookie

    Joined:
    Oct 10, 2005
    Messages:
    3,670
    Likes Received:
    8
    Ratings:
    +8 / 0 / -0

    As far as Brady is concerned, staying reasonably close to history would be a good start. Outside of the 2007 Giants game, Brady just doesn't throw for 300 yards late in the year. Combination of weather, opponents, run focus, cruising into playoffs, etc...he just doesn't air it out like earlier in the year. I'm hoping that Brady plays very little against Buffalo, which would make even hitting 5000 yards unlikely.

    Rodgers should throttle back once they clinch. Unlikely to sit as long as they haven't lost (they are more Pats than Colts) but no sense getting him killed throwing 40+ times a game. Also has all outdoor games left.

    Brees is a stat-hound playing 4 of 5 remaining games indoors. Also have Atlanta still on their tail so they are unlikely to ease up.

    Taking these situations into account, Brees should easily take the record if healthy. Rodgers has an outside shot and Brady is unlikely to reach it. Projections that ignore this context are probably not going to be very accurate.
  3. ausbacker

    ausbacker Brady > Manning. PatsFans.com Supporter

    Joined:
    Apr 23, 2008
    Messages:
    13,313
    Likes Received:
    139
    Ratings:
    +369 / 5 / -2

    #51 Jersey

    This is the Patriots remaining schedule and their passing yards conceded per game.

    Code:
    [B]Week  Venue       Opponent     Yards Ave.[/B]
    13    Gillette    Colts        239
    14    FedEx       Redskins     213
    15    Mile High   Broncos      235
    16    Gillette    Dolphins     248
    17    Gillette    Bills        244
    
    If you conclude that Brady throws for just those averages against those pass defenses, then he still ends up with 4,806 yards. If you work in Tom Brady's average performance in 2011 @ 329 yards per game, then he'll pass for 5,272 yards.

    The Patriots won't be easing up. If Brady is playing well, BB will give him a chance to break Marino's record.
  4. MP-Logick

    MP-Logick Rookie

    Joined:
    Nov 29, 2011
    Messages:
    907
    Likes Received:
    0
    Ratings:
    +0 / 0 / -0

    We'll likely need to win our games all the way through Week 17 for the #1 seed because we won't have the tiebreaker over either AFC North contender, so I expect plenty of throwing over the next 5 games.
  5. Metaphors

    Metaphors Rookie

    Joined:
    Oct 10, 2005
    Messages:
    3,670
    Likes Received:
    8
    Ratings:
    +8 / 0 / -0

    Your projection spreadsheet had Brady averaging about 340 yards per game over the next 5. All December games in cold weather sites. All against inferior opponents with the Pats 2 games clear of the Jets/Bills in their division. I just think that is amazingly over-the-top optimistic.

    Look back at the December games for Brady. While he does have the occasional 300 yd game, he is more likely to have a sub-200 yard game (for all the reasons I mentioned earlier). Not saying that will happen this year (the Indy game at least should be an exception based on the weather forecast) but the past results do seem fairly consistent year-to-year.

    I disagree with you on this, but the point is likely moot. Brees is likely going to run away with the record this year. While Belichick is a student of history, I'm thinking he isn't overly concerned about who is in 2nd place for a particular record...and I highly doubt he puts Brady in harm's way just to end up behind Brees and ahead of Marino.
  6. patsfan13

    patsfan13 Hall of Fame Poster PatsFans.com Supporter

    Joined:
    Jan 4, 2005
    Messages:
    24,758
    Likes Received:
    74
    Ratings:
    +161 / 7 / -13

    Well Brady will play all out in the opportunity for the #1 seed is still there. Clearly the weather will be a factor, the wind in particular.

    Brees has a big edge because they are playing indoors in domes.
  7. ausbacker

    ausbacker Brady > Manning. PatsFans.com Supporter

    Joined:
    Apr 23, 2008
    Messages:
    13,313
    Likes Received:
    139
    Ratings:
    +369 / 5 / -2

    #51 Jersey

    I didn't offer a chart. You have me confused with the OP.

    What's optimistic about using average passing performance for the season. By all means drop it down but the Patriots have a cupcake schedule and Brady is our best player. It's not out of the realm of possibilities to have him destory these sides then pull him.

    See above.

    I don't care if the other 2 break out, it was more the discussion about Brady breaking it for mine. BB has a track record of showing that he values personal accomplishments.

    I'm not really fussed if he does or doesn't break it.
  8. Tyler.Durden

    Tyler.Durden Rookie

    Joined:
    Mar 12, 2007
    Messages:
    586
    Likes Received:
    2
    Ratings:
    +2 / 0 / -0

    We all know Brady can still get it done in the snow, so we'll see how motivated he is and how it all unfolds.

    It would be a nice record to hold along with his 50td etc
  9. PatsFanSince74

    PatsFanSince74 PatsFans.com Supporter PatsFans.com Supporter

    Joined:
    Dec 22, 2005
    Messages:
    9,953
    Likes Received:
    47
    Ratings:
    +94 / 3 / -1

    Leaving your third point aside, you have instinctively put your finger on the problem with using these statistics.

    The average of Brady's performances to date is 329.7 yards, but the sample is so small as to be meaningless for forecasting.

    To wit, the Standard Deviation of those data points is 93.4, meaning that just going out two Standard Deviations would give a range on possible future performances between 143.0 yards and 516.5 yards per game or, to use your language, would reflect the possibility of "a sub 200 yard game."

    Furthermore, if you look at the variation between Brady's individual game performances and the opponents' average defensive performances, there is no meaningful correlation to be found between these two (very small) sets of data (the RSQ is .16, or virtually no correlation). So, using those data for the purposes of projection is, well, meaningless.

    So, really, all one can do is look at the circumstances of each game, as others have pointed out, including the playing conditions, the stakes, etc.

    If we're still obsessed with the idea of sticking to the "average," but adjust the average for its inherent volatility across just two Standard Deviations, then we would say something like, "Brady will throw for somewhere between 4,342 and 6,209 yards this season with a midpoint projection of 5,275 yards (as another poster pointed out)."

    But, that's a meaningless statement for all practical purposes, so we're really wasting our team using these data to make projections that they are inadequate to support.

    Best scenario: watch and enjoy the rest of the season and let's not try to use a small group of data to show more than they can show.

    (If someone is truly obsessed with this, they could run a Monte Carlo simulation using the available data and then wait until January 2nd to see where the real outcome falls on the curve, but...)
    Last edited: Nov 29, 2011
  10. Rainy-Day Patriot

    Rainy-Day Patriot PatsFans.com Supporter PatsFans.com Supporter

    Joined:
    Dec 21, 2004
    Messages:
    48
    Likes Received:
    2
    Ratings:
    +2 / 0 / -0

    #70 Jersey

    Here's how to sharpen your analysis:

    Create a scatterplot of TB's passing yards vs. opponents' average passing yards allowed.
    Determine the functional form of the relationship. Linear? Quadratic?
    I assume the relationship is negative: Fewer passing yards when opponent is better at pass defending. But, you can see.
    Look for weirdness. Outliers?

    If you've taken a couple of stats courses, you could go on to try this:

    Regress TB's yards on average yards allowed. Include a quadratic term if that will get the functional form right as per your exploration by way of scatterplot.

    You really want to use autoregressive techniques because your independence assumption is shot to hell by chronological relationships, Autoregressive model - Wikipedia, the free encyclopedia. But, I would squint past that "little" problem. Your bigger problems have to do with extrapolating beyond the range of your data into the cold weather etc.

    Anyhow, once you fit your model, you can use your fitted model to generate predictions for the rest of the games. With regression you can always add more variables to your model. I would suggest game temperature if we had a few cold weather games, but it's been a very mild fall. You could consider dipping into past seasons, but then you have to deal with that non-independence with a multilevel model of games clustered within seasons, but it's probably worth the added complexity.
  11. Rainy-Day Patriot

    Rainy-Day Patriot PatsFans.com Supporter PatsFans.com Supporter

    Joined:
    Dec 21, 2004
    Messages:
    48
    Likes Received:
    2
    Ratings:
    +2 / 0 / -0

    #70 Jersey

    A correlation of +/- .40 is not "virtually no correlation."
    Last edited: Nov 29, 2011
  12. fnordcircle

    fnordcircle ( o Y o ) PatsFans.com Supporter

    Joined:
    Sep 10, 2006
    Messages:
    3,597
    Likes Received:
    110
    Ratings:
    +326 / 8 / -14

    What happened in 2007 is proof positive of this. Belichick is a football historian and he loves Tom Brady - why wouldn't he want to help Tom rewrite history?
  13. robbomango

    robbomango Rookie

    Joined:
    Jan 6, 2011
    Messages:
    2,915
    Likes Received:
    0
    Ratings:
    +0 / 0 / -0

    Don't worry Dan stats are for losers.
    Last edited: Dec 27, 2011
  14. Patsfan1102

    Patsfan1102 Rookie

    Joined:
    Nov 21, 2011
    Messages:
    121
    Likes Received:
    0
    Ratings:
    +0 / 0 / -0

    Don't care. Stats are for losers.
  15. Brady_to_Moss

    Brady_to_Moss Revis Island is here PatsFans.com Supporter

    Joined:
    Jul 21, 2007
    Messages:
    17,585
    Likes Received:
    204
    Ratings:
    +513 / 5 / -1

    #95 Jersey

    We need Balt to slip for pats to get #1 seed and the pats need to win out if they want #1 seed..throw away
  16. Rainy-Day Patriot

    Rainy-Day Patriot PatsFans.com Supporter PatsFans.com Supporter

    Joined:
    Dec 21, 2004
    Messages:
    48
    Likes Received:
    2
    Ratings:
    +2 / 0 / -0

    #70 Jersey

    If the relationship is linear, then here is the model specification that I am recommending:
    TBPY = Beta0 + Beta1*OAPYA + Error

    Where TB's Passing Yards (TBPY) equals something plus something times Opponents' Average Passing Yards Allowed (OAPYA) plus an acknowledgment that there is plenty of unpredictability involved.

    This may seem fancy (or maybe not!), but it's only one step more complex than the original poster's specification, which was:
    TBPY = Beta0 + 1*OAPYA + Error

    The Beta1 parameter can be thought of as the slope of the best fitting line through the data on a scatterplot. The original poster is forcing that slope to be equal to 1. I am recommending that the slope be not so constrained.

    P.s., Above I guessed that the relationship would be negative, but I should have guessed that it would be positive: The MORE passing yards allowed by the offense, the MORE passing yards that TB will accumulate. Scatterplots are wonderful things since they make direction concrete.
    Last edited: Nov 30, 2011
  17. PatsFanSince74

    PatsFanSince74 PatsFans.com Supporter PatsFans.com Supporter

    Joined:
    Dec 22, 2005
    Messages:
    9,953
    Likes Received:
    47
    Ratings:
    +94 / 3 / -1

    Agreed, but the RSQ is not .40 but .16, which fits that definition.
  18. PatsFanSince74

    PatsFanSince74 PatsFans.com Supporter PatsFans.com Supporter

    Joined:
    Dec 22, 2005
    Messages:
    9,953
    Likes Received:
    47
    Ratings:
    +94 / 3 / -1

    yeah, but the data are far too limited and volatile to make a midpoint or a mean meaningful as predictive in any way other than as part of an observed distribution. that's my only point and is also what's meant by "The Flaw of Averages."
  19. Halifax_Pats_Fan

    Halifax_Pats_Fan Rookie

    Joined:
    Dec 30, 2005
    Messages:
    2,026
    Likes Received:
    6
    Ratings:
    +10 / 0 / -0

    These images rendered another webpage that was NSFW.

    Not cool....
    Last edited: Nov 30, 2011
  20. cavtroop

    cavtroop Rookie

    Joined:
    Sep 7, 2006
    Messages:
    1,913
    Likes Received:
    2
    Ratings:
    +2 / 0 / -0

    The work OK for me, even now. Occasionally for the first one, I'll get a pic of some windmills.

    I think they're doing some kind of 'hotlinking' protection on them or something silly, and will serve up a different pic occasionally. OP should probably have used an image host that anyone has heard of before ;)
Thread Status:
Not open for further replies.

Share This Page

unset ($sidebar_block_show); ?>