Madden ratings and Simulation football.

[ForUntoOblivionSoar∞]

Quote: Originally Posted by CM Hooe

You are correct that the rating system has 100 discrete points of delineation (0-99; 101 points if you count 100). In that sense, it is accurate, sure. However, those numbers don't guarantee how players will perform on the field given that weighted dice rolls, player traits, user inputs are factors in the outcomes of plays. For example, if you have a Terrell Owens clone with 95 CTH and "Drops Open Passes", he could drop that key 3rd down pass you need in the playoffs despite his high CTH rating. Heck a player with CTH 99 could still conceivably drop that pass, however unlikely. On the very next play, though, you could get a heroic in-traffic catch out of some 4th-string reserve with 75 CTH / 75 CIT. Is that scenario "infinitely accurate" the player's ratings? I'd argue absolutely not. Another example - for whatever reason, I could never play well with Brett Favre in any Madden game I ever played. I don't know what it was, but I sucked whenever I used Favre far more than with any other quarterback, even some backups. I was better with Quincy Carter than I was with Brett Favre. Favre was consistently rated 90+ in Madden, but I could never get good performance out of him specifically for whatever reason; is that "infinitely accurate"? Nope. The game is already dynamic enough to produce scenarios which don't conform to player ratings; how many countless stories have you seen on this message board about low-rated players who end up being great in people's franchises? How many low-rated players have you played with who have become feel-good stories in your franchises? That's exactly what's happening here; the game is behaving dynamically. Sure, the ratings represent the peak performance level of a given player - so a player with 90 CTH will certainly be more likely to catch more passes than a player with 75 CTH over a long enough timeline - but just like real life quantitative scouting reports the ratings are not themselves guarantees of that level of performance. I see no reason to take information away from the player. The 0-100 scale is easy to understand, real-life scouting reports grade players quantitatively with similar scales (so it's realistic), and those ratings given do not absolutely define a player's in-game performance. Other than offering the option to a small group of players who want less information for a greater sense of challenge, I see no benefit to changing anything.

Where an exact rating system misses the mark entirely is in draft busts (and gems) and free agents and veterans who suddenly decline. No one saw Peyton Manning falling off a cliff his last season, for example, but in Madden everyone would have known.

As for the draft, the only important trait permentantly hidden until the draft is development. But that isn't the only place where Maddens system misses on the draft. There are sometimes players who produce at a very high level early on but then dramatically regress. No one ever sees the regression coming. For example, Kaepernick, who became extremely tentative and indecisive in 2014 immediately after a horrid performance against the Cardinals. He remained in the massive funk for two seasons. (Note: contrary to popular belief it wasn't teams "figuring out the read option" that caused his decline. It was a plummeting of confidence in what his eyes were telling him. The Cardinals jumped his passes because Jim Tomsula's offensive staff were previously unemployed, bottom of the barrel castoffs and had too predictable a scheme, and from that point on until 2016 Kaepernick played scared).

[PanthersFan89]

Yeah, we need to not have everyone 99 overall at all

Sent from my KIW-L24 using Operation Sports mobile app

[adembroski]

It's cool when usability and realism collide. It happens a lot more than anyone thinks, because everyone tends to go "that's too complex."

A system that translates ratings into realistic information would actually be more usable for the casual and whatnot because it would boil everything down to easy-to-digest chunks.

For instance, you might translate all of the QB ratings into a series of categories: Size-Speed, Pocket Presence, Short Passing, Long Passing, and Vision.

Each of these categories is an amalgamation of several of the core ratings, but it is not a weighted average like the current OVR rating is. Instead, it first evaluates against a baseline by asking yes or no questions for each rating. For instance, a player with a THP of less than 75 would immediately get the worst possible grade in deep passing and no other rating need be evaluated. It doesn't matter if he has 99 DAC because he can't get the ball that far quickly enough for that accuracy to be of any use.

The design of these yes/no questions and the follow up checks to determine a final grade would be a trial and error design process, but it wouldn't be terribly difficult. One could easily figure it out in excel in a couple of afternoons.

I envision a ratings screen for a player that shows 4 sections; evaluation from each level of coach (position coach, coordinator, head coach) and measurables (combine numbers). The combine numbers would be static once a player's combine is finished, but there would be notes from the coaches ("Has lost a step with age," "That knee injury really hurt his vertical," "His time with the training staff has really improved his upper body strength.")

Above that, across a season the player receives performance grades (similar to FIFA's).

The above ratings existing in 4 stages, based on how much scouting has been done (we can consider that a player in the league is being constantly scouted, and so once he's with a team for a time, the only variable will be the coach's ability to evaluate).

So, for instance, we you begin scouting everyone in the class with have only 1 bit of information; his strongest trait. The evaluation will simply say, "has a huge arm," or "fantastic mobility." You have that, just that, his strongest trait, and you don't even know *how* strong it is.

Getting to stage 2, you get the 2 strongest and 2 weakest. Same deal, no precision, just what areas of his game are good, and what are bad.

Stage 3 gives you all of the traits, and a bit more information; +, 0, -. + means he's in the top 25% of prospects in an area, 0 means middle 50%, - bottom 25%. These are not exact, there are variables that mean someone might be displayed as one category off (in owner mode, I'd use a scouting department budget to determine how often this fuzzy logic is applied).

Stage 4 is letter grades or the 0.0-5.0 scale or the 9 point scale or the 7 point scale, honestly, you could let the user decide because it's all a calculation anyways and can be converted in any way you want. Again, fuzzy logic applies here based on the "ability" of your scouts.

(AN ASIDE: My owner mode idea was to have a series of departments to which the owner allocates funds for the season. Marketing to fill the stadium, PR to minimize the damage of disappointing seasons or perceived bad signings/hires, College Personnel to evaluate for the draft, Pro Personnel to evaluate NFL talent, Coaching to build your staff, Personnel Relations to handle contract negotiations and signings. This budget would dictate just how good your college scouts are, and thus, how often they miss on a given evaluation).

Anyways, this is a high level look at it. I always want to put this stuff together in excel just to show off how it'd work, but it's hard to actually do design projects now that I don't actually get paid for it... but I also like putting ideas out there and hoping maybe the team get's wind of it and thinks of something along the lines.

[ForUntoOblivionSoar∞]

Quote: Originally Posted by adembroski

It's cool when usability and realism collide. It happens a lot more than anyone thinks, because everyone tends to go "that's too complex." A system that translates ratings into realistic information would actually be more usable for the casual and whatnot because it would boil everything down to easy-to-digest chunks. For instance, you might translate all of the QB ratings into a series of categories: Size-Speed, Pocket Presence, Short Passing, Long Passing, and Vision. Each of these categories is an amalgamation of several of the core ratings, but it is not a weighted average like the current OVR rating is. Instead, it first evaluates against a baseline by asking yes or no questions for each rating. For instance, a player with a THP of less than 75 would immediately get the worst possible grade in deep passing and no other rating need be evaluated. It doesn't matter if he has 99 DAC because he can't get the ball that far quickly enough for that accuracy to be of any use. The design of these yes/no questions and the follow up checks to determine a final grade would be a trial and error design process, but it wouldn't be terribly difficult. One could easily figure it out in excel in a couple of afternoons. I envision a ratings screen for a player that shows 4 sections; evaluation from each level of coach (position coach, coordinator, head coach) and measurables (combine numbers). The combine numbers would be static once a player's combine is finished, but there would be notes from the coaches ("Has lost a step with age," "That knee injury really hurt his vertical," "His time with the training staff has really improved his upper body strength.") Above that, across a season the player receives performance grades (similar to FIFA's). The above ratings existing in 4 stages, based on how much scouting has been done (we can consider that a player in the league is being constantly scouted, and so once he's with a team for a time, the only variable will be the coach's ability to evaluate). So, for instance, we you begin scouting everyone in the class with have only 1 bit of information; his strongest trait. The evaluation will simply say, "has a huge arm," or "fantastic mobility." You have that, just that, his strongest trait, and you don't even know *how* strong it is. Getting to stage 2, you get the 2 strongest and 2 weakest. Same deal, no precision, just what areas of his game are good, and what are bad. Stage 3 gives you all of the traits, and a bit more information; +, 0, -. + means he's in the top 25% of prospects in an area, 0 means middle 50%, - bottom 25%. These are not exact, there are variables that mean someone might be displayed as one category off (in owner mode, I'd use a scouting department budget to determine how often this fuzzy logic is applied). Stage 4 is letter grades or the 0.0-5.0 scale or the 9 point scale or the 7 point scale, honestly, you could let the user decide because it's all a calculation anyways and can be converted in any way you want. Again, fuzzy logic applies here based on the "ability" of your scouts. (AN ASIDE: My owner mode idea was to have a series of departments to which the owner allocates funds for the season. Marketing to fill the stadium, PR to minimize the damage of disappointing seasons or perceived bad signings/hires, College Personnel to evaluate for the draft, Pro Personnel to evaluate NFL talent, Coaching to build your staff, Personnel Relations to handle contract negotiations and signings. This budget would dictate just how good your college scouts are, and thus, how often they miss on a given evaluation). Anyways, this is a high level look at it. I always want to put this stuff together in excel just to show off how it'd work, but it's hard to actually do design projects now that I don't actually get paid for it... but I also like putting ideas out there and hoping maybe the team get's wind of it and thinks of something along the lines.

I think a good compromise here is to have a GENERAL MANAGER MODE, rather than "owner mode," and in this mode part of the fun is scouting and team building, so HERE the ratings will be hidden in this way you suggest (which is similar to my own suggestion). Obviously things like speed and strength will be known form combine results, although the whole "he's lost a step" thing can be an added dimension.

For those who want the classical Madden rating system, they could just keep that as is for COACH mode.

And of course, GM Mode would have all the features from Coach mode. It'd just be a more in depth mode for those who have the time and inclination to have a GM simulation, with all the scouting uncertainty that goes with it.

[Toupal]

Quote: Originally Posted by PanthersFan89

Yeah, we need to not have everyone 99 overall at all Sent from my KIW-L24 using Operation Sports mobile app

I wrote this thread awhile back that piggy backs off this idea.

http://www.operationsports.com/forum...tory-fans.html

Essentially separate the greats from the good and the good from the bad by expanding the rating system so a majority of the player ratings don't run from 70-90 overall.

[cusefan74]

I don't think ratings should go away, just the overall rating. People pay way to much attention to the overall rating and they miss out on some really good players because of it. That would make the preseason more important then ever.