The Great Madden Muppet Caper (aka, Madden as Nostradamus)

[JaymeeAwesome]

Quote: Originally Posted by alifeincomplete

Through the halfway point of the season in pick'ems, Madden predictions are 75-41. That's an accurate prediction percentage of just 65.2% with a weekly mean accuracy of 64.6%. Last time I checked, that's a D.

Even though this is a good tactic to gauge how close madden is to real football it would be very difficult unless you are doing this already.

First of all, Injuries would need to be removed...plus each week you would have to restart a franchise and sim to the week you want to check. Make all the injuries and depth charts yourself. You can make the injuries by controlling all 32 teams and releasing the hurt player and making sure no team picks up that player before the game. Plus make sure you have the latest roster update downloaded before the week plays out.

If you are already doing all that then my hats off to you...thats a lot of work and we should definitely expect an accuracy of 90-100%, but if you're not doing all that then I would expect an accuracy of around 65-75%....which is what it is.

If you want I can redo all the weeks for you with the way I laid out....it would take some time but I can go back and check depth charts from those weeks......it would take some time though, time I don't currently have...

But keep on doing it the way you are, I like to see them stay around 65%

[KBLover]

Quote: Originally Posted by alifeincomplete

Through the halfway point of the season in pick'ems, Madden predictions are 75-41. That's an accurate prediction percentage of just 65.2% with a weekly mean accuracy of 64.6%. Last time I checked, that's a D.

Using a school grading scale wouldn't be accurate, imo. Otherwise, every QB in the NFL would get a D or F in accuracy.

What would be the average result of people picking games? 50%? 60%?, 80%? Without that, it's hard to assign a "grade", imo, because there's no baseline for what is average. If Madden was worse than the average game picker - then yeah, D, but perhaps it's about as good as the average picker - that would put it at a C.

I mean, if I look at it in a sports "win percentage" sense, that's a pretty darn good win rate. In NFL terms, .646 win percentage is close to a 10-6 team (.625) and a good week or two from an 11-5 team (.688). I don't think I'd give a 10-6 a "D" for their record.

[alifeincomplete]

I think I agree that it's unfair for me to assert that the game's accuracy percentage is a D or a near-failing grade. I'll have to compare its accuracy to some of the NFL analysts' predictions to see how they stack up. Ultimately, guessing a sport outcome is a little bit strategy, and a lot of luck.

jaymee, I began this franchise by taking control of all 32 teams and turning off injuries. I control who plays each game each week with the depth chart and add/release players. I've made all the major roster moves throughout the year (ie, Edwards to NYJ) and, using KFFL inactives, I "injure" players who need to be injured or taken out of the game each week. My problem here is one of ratings and I'm thinking about utilizing a new franchise to sim ahead to next week to see if things change. I'm using the old update's ratings, so Favre is still an 82 and Kyle Orton is still in the 70s, etc. I think that may have a definite impact on the outcome of some games. If you can think of further adjustments necessary to the franchise and the Madden accuracy, I'd love to hear them!

[alifeincomplete]

Argh, SO close on the MNF prediction! Season standings are 83-46.

To put this in perspective, the following are the through-nine-weeks standings for ESPN's analysts in their weekly pick'ems. I'm hoping that changes to the way the predictions are run (ratings) will provide better results this upcoming week, but we'll see.

Allen 83-46

Golic 87-41

Hoge 89-40

Jaworski 80-40

Mortensen 89-40

Schefter 88-41

Schlereth 88-41

Wickersham 84-45

Accuscore 89-40

Pick 'em 88-41

[KBLover]

Looks like Madden's doing a little worse than most of the experts. They tend to range between 66 and 69% (just ran their percentages).

Though Allen and Madden are tied, so that's a positive that it even matches...or is even close to a human picker.

So Madden might be a D+ if you're judging against experts, though, given it's a program and not capable of making the kinds of analysis humans are, I'd be inclined to give it a curve to a C at least.

Like you said, picking is strategy and analysis and then some luck. Madden mostly has to count on luck since it can't really analyze (or at least it's data is dependent on what it's given, i.e. ratings and rosters)

[JaymeeAwesome]

Quote: Originally Posted by alifeincomplete

I think I agree that it's unfair for me to assert that the game's accuracy percentage is a D or a near-failing grade. I'll have to compare its accuracy to some of the NFL analysts' predictions to see how they stack up. Ultimately, guessing a sport outcome is a little bit strategy, and a lot of luck. jaymee, I began this franchise by taking control of all 32 teams and turning off injuries. I control who plays each game each week with the depth chart and add/release players. I've made all the major roster moves throughout the year (ie, Edwards to NYJ) and, using KFFL inactives, I "injur" players who need to be injured or taken out of the game each week. My problem here is one of ratings and I'm thinking about utilizing a new franchise to sim ahead to next week to see if things change. I'm using the old update's ratings, so Favre is still an 82 and Kyle Orton is still in the 70s, etc. I think that may have a definite impact on the outcome of some games. If you can think of further adjustments necessary to the franchise and the Madden accuracy, I'd love to hear them!

Great I'm glad to hear that you've been doing all that, for next week just as a test run the week with two different franchises....one with the EA current ratings for players and one like the way you're doing it...See what the outcome will be.

Seeing as though you're doing all the grunt work except ratings you may see a difference of +/- 2....but great work....I'm so glad that you were doing all that work with the players...I sooooooooo didn't want to put time into doing it to see how it compared to yours. I would rather spend in on my Dynasty, because that ends up being like 3 hours of time per game to get all the stats into html format...

[alifeincomplete]

Okay, we've got some things going on here that are different than before. For starters, I apologize about not being able to post this until now. I was at my November battle assembly all weekend and didn't get a chance to post the results of the sims until now. What follows are two sets of predictions, one using the 'old' style of predictions using the original roster update and a 'new' style using the latest roster update. I want to see if it makes a significant difference; the more accurate style will become the new method for the rest of the season and post-season.

"Old" style:

Week 10
Season results: 83-46
Week results: 11-5

Madden: 49ers 17, Bears 14 (OT)
Reality: 49ers 10, Bears 6
Correct!

Madden: Jaguars 10, Jets 7
Reality: Jaguars 24, Jets 22
Correct!

Madden: Broncos 27, Redskins 18
Reality: Redskins 27, Broncos 17
Incorrect!

Madden: Dolphins 17, Buccaneers 14
Reality: Dolphins 25, Buccaneers 23
Correct!

Madden: Saints 27, Rams 18
Reality: Saints 28, Rams 23
Correct!

Madden: Vikings 24, Lions 17
Reality: Vikings 27, Lions 10
Correct!

Madden: Titans 23, Bills 14
Reality: Titans 41, Bills 17
Correct!

Madden: Steelers 31, Bengals 16
Reality: Bengals 18, Steelers 12
Incorrect!

Madden: Chiefs 21, Raiders 16
Reality: Chiefs 16, Raiders 10
Correct!

Madden: Packers 27, Cowboys 24
Reality: Packers 17, Cowboys 7
Correct!

Madden: Eagles 17, Chargers 10
Reality: Chargers 31, Eagles 23
Incorrect!

Madden: Cardinals 35, Seahawks 21
Reality: Cardinals 31, Seahawks 20
Correct!

Madden: Falcons 29, Panthers 25
Reality: Panthers 28, Falcons 19
Incorrect!

Madden: Patriots 20, Colts 6
Reality: Colts 35, Patriots 34
Incorrect!

Madden: Ravens 20, Browns 10
Reality: Ravens 16, Browns 0
Correct!

_____

"New" style...

Week 10
Season results: 83-46
Week results: 12-3

Madden: 49ers 33, Bears 30 (OT)
Reality: 49ers 10, Bears 6
Correct!

Madden: Jaguars 28, Jets 23
Reality: Jaguars 24, Jets 22
Correct!

Madden: Redskins 27, Broncos 16
Reality: Redskins 27, Broncos 17
Correct!

Madden: Dolphins 20, Buccaneers 6
Reality: Dolphins 25, Buccaneers 23
Correct!

Madden: Saints 31, Rams 17
Reality: Saints 28, Rams 23
Correct!

Madden: Vikings 27, Lions 17
Reality: Vikings 27, Lions 10
Correct!

Madden: Titans 18, Bills 13
Reality: Titans 41, Bills 17
Correct!

Madden: Bengals 27, Steelers 13
Reality: Bengals 18, Steelers 12
Correct!

Madden: Chiefs 27, Raiders 10
Reality: Chiefs 16, Raiders 10
Correct!

Madden: Cowboys 30, Packers 20
Reality: Packers 17, Cowboys 7
Incorrect!

Madden: Chargers 41, Eagles 23
Reality: Chargers 31, Eagles 23
Correct!

Madden: Cardinals 34, Seahawks 13
Reality: Cardinals 31, Seahawks 20
Correct!

Madden: Falcons 20, Panthers 10
Reality: Panthers 28, Falcons 19
Incorrect!

Madden: Patriots 35, Colts 21
Reality: Colts 35, Patriots 34
Incorrect!

Madden: Ravens 24, Browns 13
Reality: Ravens 16, Browns 0
Correct!
____

Altogether, I'm blown away. A great week for Madden's accuracy at pick'ems either way and the updated rosters shows a slight edge.

[alifeincomplete]

Okay, so the 'new' style with the updated rosters has won out, going 12-3 over the old rosters. I will be sticking with this new method for the remainder of the season and postseason.

The season stats are now... 95-49 on the year which is a 66.0% (rounded up from 65.97%) accurate prediction percentage with a weekly mean of 65.9%.