FBGRatings Seeks to Recalibrate Madden Ratings, Change Game

[DerekHupp]

maybe most of the player audits can wait until the active rosters are done
that is what we would want most

[DCEBB2001]

Quote: Originally Posted by DerekHupp

maybe most of the player audits can wait until the active rosters are done that is what we would want most

I understand that the roster creators out there would like that, but you have to keep in mind that the good people here at OS are not our only subscribers.

It takes a long time to do the edits to get the attributes to match up to the OVRs. It is much easier to do them position-by-position once, than do a group for the guys currently on rosters, then do them again for the guys who are not.

[caballero]

I'll reply here DCEBB, Blazelore's thread (http://www.operationsports.com/forum...rs-ps3-32.html) isn't really appropriate for that.

Quote: Originally Posted by DCEBB2001

I take great offense to be lumped into the same sentence as Donny Moore, who I am sure took no effort to incorporate kinematics and calculus into finding out how fast NFL players really are

huh, did you forget who I am, DCEBB? you DID use SPD=40 time, and only when PGaither and I pointed out to you Donny used that too (at least to a degree), you decided to search other methods, see red part below.

Quote: Originally Posted by DCEBB2001

take a new approach to how players are rated. You can thank caballero and PGaither for getting on my case about that. The whole thing that limited it in the first place was the lack of data to show a true acceleration curve and top velocity. However, once I found out that ALL players in Madden regardless of SPD/ACC stop accelerating at 45 yards I found that I was able to use the split times for each player to measure the change in velocity (ACC) and predict the top instantaneous velocity (SPD) at 45 yards using a cubic function. Now it is just a matter of finding an easy way of doing it for some 20000 players, which should take a few weeks to complete

(post 936 page 94 of this thread)

Quote: Originally Posted by DCEBB2001

I do NOT use the 40 yard dash time to calculate the SPD rating. The SPD and ACC ratings work together to get the desired outcomes. All players have been rated according to the cubic functions of the runs and compared to the entire population regardless of position. I take great offense to be lumped into the same sentence as Donny Moore, who I am sure took no effort to incorporate kinematics and calculus into finding out how fast NFL players really are. From all of the time I spent this offseason trying to find the best way to get these numbers accurate with the available data, the result you find on the website is the best I can do with only a finite number of data points. I am certain that the methodology, which can be easily repeated, is sound and anyone studying kinematics with the same data would yield the same results. I used my connections from grad school to get the best opinions I could, sometimes using my own funds (time for some of these people was not cheap) to get the most logically valid and sound results I could. If someone can do it better, that's great because it will progress the science behind understanding the game. As for now, however, this seems to be the best we can do with what we presently have. I don't recall ever using Bolt's data to use against the population of NFL players as well. I do recall attempting to figure out how fast he would cover his splits, but I never used that as a comparison against the rest of the NFL population in the ratings. Once Bolt becomes and NFL player and runs a 40 with all of the split times accounted for, then we will see how he matches up. Bolt, however, is a different runner than that of an NFL player. Most NFL players reach their maximum velocity around the 30-35 yard mark of the 40 yard dash. Bolt reaches his maximum velocity around the 60-80m mark of his 100m dash. Football players and 100m sprinters require very different running techniques. For football players, you want to go as fast as you can as quickly as you can. For the 100m, it is about maintaining the top velocity for as long as you can (speed endurance). All I ask is that you ask me first about what I did/didn't/forgot to do before simply assuming that I did/didn't/forgot. It comes off as a tad presumptuous and slightly unfair, in my humble opinion.

I'm just going by what you said to me here, man...

a) you never followed up on the SPD thread so I didn't know you choose cubic functions.

b) you talked yourself of a 1 pt regression in SPD each 4 years:

Quote: Originally Posted by DCEBB2001

Using my scale for the 40 and correlating SPD ratings, a player will typically lose 1 SPD point over the course of every 4 seasons after being drafted.

post 66: http://www.operationsports.com/forum...post2043946314

c) As for Bolt, you remember we calculated his splits and approximate 40 time and he timed at 4.10/4.11.
I would have liked (even though he'll probably never play in the NFL) to have the SPD attribute be ranked based on his cubic function.
Now I understand you won't incorporate him, but could you please let me know if he was a 99 SPD, what would the next NFL player be or follow up on this thread: http://www.operationsports.com/forum...-flawed-3.html
?

[DCEBB2001]

Quote: Originally Posted by caballero

I'll reply here DCEBB, Blazelore's thread (http://www.operationsports.com/forum...rs-ps3-32.html) isn't really appropriate for that. huh, did you forget who I am, DCEBB? you DID use SPD=40 time, and only when PGaither and I pointed out to you Donny used that too (at least to a degree), you decided to search other methods, see red part below. (post 936 page 94 of this thread) I'm just going by what you said to me here, man... a) you never followed up on the SPD thread so I didn't know you choose cubic functions. b) you talked yourself of a 1 pt regression in SPD each 4 years: post 66: http://www.operationsports.com/forum...post2043946314 c) As for Bolt, you remember we calculated his splits and approximate 40 time and he timed at 4.10/4.11. I would have liked (even though he'll probably never play in the NFL) to have the SPD attribute be ranked based on his cubic function. Now I understand you won't incorporate him, but could you please let me know if he was a 99 SPD, what would the next NFL player be or follow up on this thread: http://www.operationsports.com/forum...-flawed-3.html ?

1. Donny Moore uses 40 times to calculate the SPD and ACC rating. The correlation between both is around 0.97 when you graph it. I used the 40 to determine the overall SPD rating and the 10 yd split to determine ACC. I did this because in the game the SPD and ACC ratings work symbiotically with one another. What that means is that, for instance, if Player A has an ACC of 90 and a SPD of 90, he will reach the 10 yd split in, for example, 1.50s. Now, let's say that Player B has the same ACC of 90 but has a SPD of 50. He will reach the 10 yd split in 1.70s. That means that although both players have the same ACC rating, which you would THINK would accurately determine who has the better 10yd split time, the player with the fastest SPD still gets to the 10yd mark first. That is what I was trying to work around. For some reason the SPD rating is a "modifier" of the ACC rating in the game itself. You can test this out in practice mode if you like, which I referenced several times by stating that the relationship between SPD and ACC is symbiotic within the coding of the game. That means that it is impossible to take the 10 yard split of a player who ran 1.50 and 4.40 and have them reach the 10 yard mark at the same time as a player who ran 1.50 and 4.80. The player with the 4.80, although having the same 10 yard split, would still get to the 10 yard mark SLOWER than his 4.40 counterpart. What I did was use the 40 time to determine the average velocity over the distance and then use the split time as a "segment" to determine the ACC. That way two players with the same 40 time and thus, the same SPD ratings, would reach the splits at the same time compared to one another should their 10 yard splits also be equal.

I did this until just saying "screw it", and actually determining the maximum velocity and acceleration for each player. The problem is that I do not know if this will replicate accurately in the game. Maybe it does, maybe it doesn't. If you want to see, go ahead and mess around with it, but I really don't have the time to beta-test how SPD and ACC works in Madden. Instead, think of every rating on the FBG site as applying Madden ratings to the "real" world outside of Madden. Whether everything is accurate in the game itself is yet to be fully documented due to the fact that the code for the game is so messed up regarding running mechanics, acceleration, speed, ect.

There is no evidence that Moore used the 10 yard splits like I did initially because the correlation between the ACC rating and the 10 yard split for all of the players is only around 0.84...significantly less than the correlation between the EA SPD rating and the 40 time of a player. Ultimately, the method used now is accurate in the real world, but may still be "off" in the game due to the restrictions of the game itself.

2. I did decide on cubic functions back in January.

3. I did state the a 1 point decrease in speed every 4 years after the combine/pro day would be an accurate representation, but using the injury factor to determine a loss of speed is less reliable because injuries vary greatly and their effects on speed vary from player to player.

4. If I get some time, I will do Bolt's SPD and ACC in that thread. Let me work it out mathematically first and then I will post something to see how he compares to the fastest players with the most burst in the NFL. I wanted to keep the population of data the same, so I chose to only use football players who were tested in the 40 as the qualifying measure to determine SPD and ACC ratings. You can also do this yourself by measuring Bolt's ACC and SPD from his splits (cubic function, differentiate, etc).

[bodean]

So Xbox users have to manually enter your ratings/values into the game? No way to download a file to take care of the leg work?

[DCEBB2001]

Quote: Originally Posted by bodean

So Xbox users have to manually enter your ratings/values into the game? No way to download a file to take care of the leg work?

There is not a file to download from the website. We do not make rosters.

[DCEBB2001]

Caballero

OK, here is what I have come up with after my analysis of BOlt's World Record run of 9.58 in 2009. I will compare these numbers to the 40 times and subsequent SPD and ACC ratings of NFL players as well.

Here are the splits from Bolt's run:
Reaction Time: 0.146s
10m: 1.89s
20m: 2.88s
30m: 3.78s
40m: 4.64s
50m: 5.47s
60m: 6.29s
70m: 7.10s
80m: 7.92s
90m: 8.75s
100m: 9.58s

Here are the times for each segment:
Reaction Time: 0.146s
0-10m: 1.89s
10-20m: 0.99s
20-30m: 0.90s
30-40m: 0.86s
40-50m: 0.83s
50-60m: 0.82s
60-70m: 0.81s
70-80m: 0.82s
80-90m: 0.83s
90-100m: 0.83s


From these splits we can make a few assumptions:
1. The total time it took for Bolt to run the 100m, minus the reaction time, was 9.434 seconds.
2. Bolt's fastest segment was from 60m-70m which he ran in only 0.81 seconds.
3. Bolt continued to accelerate until some point before the 70-80m segment.

Using this data, we can graph Bolt's run, create a function, and then determine his maximum velocity and acceleration. We can also determine when these were achieved during the run and interpolate them into an hypothetical 40 time and subsequent splits. This can then be compared to data show by NFL players.


When you negate the reaction time, the splits are shown below. At this point I will convert meters to yards. 1 meter is equal to 1.0936 yards.

0.000y: 0.000s
10.936y: 1.744s
21.872y: 2.734s
32.808y: 3.634s
43.744y: 4.494s
54.680y: 5.324s
65.616y: 6.144s
76.552y: 6.954s
87.488y: 7.774s
98.424y: 8.604s
109.360y: 9.434s



From this, we can estimate the split times at the 10yd, 20yd, and 40yd marks match the data collected for NFL players at the NFL Combine and various Pro Days.

0y: 0.00s
10y: 1.60s
20y: 2.50s
30y: 3.32s
40y: 4.11s

When you look at this data, Bolt reaches the 40 yard mark in 4.11 seconds. The best verifiable time by any NFL player since 1998 is 4.21s. The biggest difference is in the first 10 yards for NFL players. The best verifiable 10 yard split time is 1.40 seconds, whereas Bolt reached the 10 yard mark in 1.60s. This seems to make sense considering that Bolt is a taller runner who relies on his stride to make up ground later in the run, while sacrificing acceleration in the first 10 yards. After all, Bolt doesn't need to accelerate, stop, and accelerate again like many NFL players need to do on any given play.

Let's compare Bolt's splits to two other players; Marquise Goodwin and Reggie Dunn. Goodwin, along with Trindon Holliday, holds the record for the fastest 40 time since 1998 at 4.21 seconds. Dunn holds the record for the fastest final 30 yards of any player at 2.70 seconds (4.22-1.52=2.70).

Goodwin:
0y: 0.00s
10y: 1.50s
20y: 2.41s
40y: 4.21s

Dunn:
0y: 0.00s
10y: 1.52s
20y: 2.49s
40y: 4.22s

As you can see, Goodwin and Dunn accelerate much quicker than Bolt through the first 10 yards. However, Bolt destroys them by a tenth of a second after the 40 yard mark. Bolt, who is actually still behind Goodwin and Dunn at the 20 yard mark, turns a defecit as large as .09s into a surplus of .10s over the course of the final 20 yards. That is a .19 turnaround over the course of the final 20 yards. Bolt's split over the final 30 yards is an amazing 2.51 seconds!

When we create cubic functions and differentiate to determine the maximum speed and acceleration for all three players, we find that Bolt, during the first 40 yards (when he is not even at top velocity as shown by the segment times above) reaches a top velocity of 25.85 MPH. Based on this, and the way that FBG ratings scales SPD, Bolt would have a SPD of 99 while the next closest player would have a SPD of 90. For acceleration, the fasest acceleration would be 99, while Bolt would have an ACC of 74.

As you can see, if we incorporated the world's fasest man into our equations, Bolt would be way ahead of every NFL player in the SPD rating, but would be only above average in ACC.

[jfb33]

Do you know any formula that we can apply to current player abilities to obtain the correct overall? With that we could at least make the change by ourself in our game.