LOBBS Roster Rating System-Roster Release

[LOBBS]

Some of you may be familiar with my methodology behind this system that I outlined many moons ago in the M08 Roster section of the Football Idiots forum. I won't rehash too much of that here. Madden 19 inspired me to finally finish it though and push out a first release.

https://www.dropbox.com/s/omb7g1he3a...LOBBS2018?dl=0

It's based on the Super Bowl week EA roster.

Long and short of my system is to adjust players to more closely match their combine/pro day stats. I built a massive database of player results from those drills and developed formulas that are pretty close to what EA uses as a base. From their 40 times with 10 and 20 yard splits, 3-cone and shuttle, vertical and broad jump I could predict their EA rating within a point or so. The problems with EA's rosters occur at the extremes of the rosters. There are countless mid to low round draft picks, undrafted free agents, etc. where it is obvious that EA just attempted to get a name into the game but didn't pay any mind to the real athlete's benchmarks. Or times where they purposely crippled a player's physical ratings to force them to slot in properly on a depth chart. This is fine and all for Play Now and Online matches but those mid to low players matter in franchise play. My method fixes them so that they have a reasonable opportunity to be productive players on a roster or possibly develop into starters. On the other end of the spectrum are the stars and superstars that EA over boosted their physical ratings to drive their OVRs and desired improvements relative to the pack up rather than skill ratings. Admittedly, EA has added many more skill ratings over the years and has improved a bit on those but they'd already set the baseline for an elite player's physical ratings prior to doing so and chose not to revisit them when they added the additional skills. My system also adjusts for that and moves players back closer to their combine/pro day numbers. The "points lost" when downrating physical traits were moved to the appropriate existing and newly added skills ratings.

The other big thing is strength ratings. It annoyed me it M08 and still does to this day. Lineman with strengths in the upper 70s/lower 80s, RBs as strong as lineman, weak TEs, etc. are all gone. I built a new method based on height, weight, BMI (are they built like a brick ****house or long and lean?) and athleticism (based on combine/pro day data). I get way more in depth of my reasoning in the M08 roster thread though I've tweaked a few things since then.

For those that want to dig into the math I've posted my spreadsheet below.

https://www.dropbox.com/s/jpd9y6qnc3...2018.xlsx?dl=0

The bulk of the player's OVRs don't change much but a point or two in this roster compared to the EA roster it's built from. Some of the players that EA just threw together do jump several points up or down on their OVR. I've blended my data to varying proportions with the original EA ratings to simulate how a player looks in pads versus in shorts at their combine/pro day. But in the end, my goal was to fairly rate every player across the board by tying their physical traits back to the most consistently gathered and widely accepted set of data.

[LOBBS]

I should also note that it follows the trait adjustments to work with Sabo's Revival Mod as outlined in the OP of his thread.

I will also be releasing a spreadsheet to tweak EA generated draft classes to follow the same methodology in the very near future.

[ForUntoOblivionSoar∞]

This is pretty cool, but I do have one line questioning: regarding explosion off the line by outside pss rushers, how much in the game does beetical and broad jump affect that (because in real life it is a bug indicator) and do you emcorporate combine jump results in the rating that affects first step for edge rushers? (Which Inassume in the game is acceleration)


If not, for future reference you may want to examine the correlation between elite edge rushers with a great first step and their vertical. I was quite surprised at first when I noticed this, but in hindsight it makes sense: your vertical is about explosive power, and so is the first step off the edge.





Anyway, great work. I am definitely looking forward to this.

[LOBBS]

Quote: Originally Posted by ForUntoOblivionSoar∞

This is pretty cool, but I do have one line questioning: regarding explosion off the line by outside pss rushers, how much in the game does beetical and broad jump affect that (because in real life it is a bug indicator) and do you emcorporate combine jump results in the rating that affects first step for edge rushers? (Which Inassume in the game is acceleration) If not, for future reference you may want to examine the correlation between elite edge rushers with a great first step and their vertical. I was quite surprised at first when I noticed this, but in hindsight it makes sense: your vertical is about explosive power, and so is the first step off the edge. Anyway, great work. I am definitely looking forward to this.

I'd not made that connection but I can certainly add it in the next update. It'll be for all players though not just edge rushers.

[ranta21]

This is awesome stuff. Thank you!

Some recommendations I'd like to suggest.

-I would use (40 Time) - (20 Split) for SPD. Now you are double counting the first 20 yards (ACC and SPD). SPD in Madden is the maximum velocity. The last 20 yards of the 40 yard dash is the best measure we have for that. It would change the ratings quite a bit, but it would be more authentic, and no more double counting.

*-I would also not count the 40 yard time in ACC for same reasons as above.

-I would argue that Height shouldn't be a negative for STR. It is the one of the more important traits Coaches have for Offensive Tackles and Pass Rushers. I don't think it should be a negative.

-Continuing with STR: Why not add Bench Reps and Arm Lenght to the formula? Arm Lenght is all about controlling opponents during blocking situations. If not available data, you could estimate the Arm Lenght based on Height. Bench Reps could be estimated with Weight.

-We don't have Arm Lenght as an attribute but it is also very important to WR's and DB's. We could maybe add arm lenght attribute to JMP rating. Lets say we have two players with same Height, Vertical and Broad, but the other has few inches more arm lenght. I'd argue we should give him an edge in JMP to simulate the edge he has in jump balls.

Like I said. This is excellent and will use this myself. Thank you again.

[LOBBS]

Speed, by definition, is simply a distance in a matter of time. In theory, a player should still be accelerating (or at least have the opportunity to) thru the end of 40 yards. Without some Next Gen type stats, it's impossible to know if he hit/would have hit his max velocity at say 32 yards or 42 yards. I'd love to have that velocity number for all 2900+ players. However, the standard for years has been the time it takes to cover 40 yards.

As far as acceleration using the splits goes, what you really have are three separate events in one attempt. A measure of 0-10, 0-20 and 0-40 yards. To attempt to capture that explosion in the acceleration curve I weighted the 0-10 more than the 0-20 which was more heavily weighted than the 0-40. You will notice that I did not select the values for maximum and minimum arbitrarily. Every calculation in my spreadsheet is a range between the minimum and maximum observed values in the dataset for that test. A player's rating is relative to his performance versus his peers. A solid score in the 0-10 and 0-20 relative to his peers is rewarded and a deceleration at some point in the 0-40 is penalized. I just don't break them out where you can see the score at each split. Again, not knowing exactly when and where a player hit his maximum velocity during the dash leaves us with making the best of the data we do have. I may play around with the weighting of the splits to put even further emphasis on the 0-10 and slightly more on the 0-20 but it may actually hurt the overall ratings. If I recall correctly, there are a handful of players with crazy fast bursts so the rest would all suffer a bit.

Height, in and of itself, is not a penalty for strength. The scale for height runs from 5'8 to 6'8" with 6'8" being a 100 rating. Height without mass is penalized if it drops their "BMI". I chose "BMI" as it is a widely accepted formula for one's composition. Are they tall and skinny or tall with a filled out frame? The one that's fleshed out should be stronger than the beanpole. To help differentiate between being athletic and just being fat according to their BMI I did cross reference their 40 time with a constant. I am seriously considering breaking out the scores for all the combine/pro day drills in addition the 40 to give a better approximation of athleticism/strength in that regard when referenced against their BMI.

I don't agree with bench press as a end all/be all measure of strength. Biomechanically, it's a pretty terrible exercise. It's pretty common in most serious strength programs to move away from it to more effective things like flies. I've heard a lot of draft scouts refer to it simply as a measure of whether the guy has a history of hitting the gym or not. I'd consider adding it as an additional data point but not carrying any significant weight relative to the other factors.

I agree that arm length is important for some positions but not so much as a measurement of strength. For players on the edge it really is a measure of the radius a player can occupy without letting someone get into their body. Lineman with relatively short arms tend to get bumped down to an interior position where they're not expected to cover as much ground on an island, etc. I'll do some more research to see if there's any value into adding it.

[Mattryan720]

This is awesome

[NINJAK2]

LOBBS- I wanted to pose this question to you and any other roster editors. Please respond privately if you guys don't want to clog up the thread but I thought you or someone else may have an answer.

Have any of you guys ever experienced a crash to the desktop when trying to use a specific team in a user created roster? I've been using a user created roster(Not your roster LOBBS) and for some reason when ever I try to play a game involving the Chiefs no matter if it's with them, against them, or just cpu vs cpu I get a hard crash back to desktop in the middle of loading the game. I have done a cpu vs cpu game with over 20 different teams thus far in this roster and the Chiefs are the only team that seems to cause this crash. Any idea what might cause this? I've looked through the roster with Amp and checked depth charts and can't seem to find anything off. Could having too many players on the team roster be the cause?