<<I created this post last year for Big Data and I don't want to lose it.>>

I've been thinking about the changes to speed to reduce the gap between 0 and 100. Hopefully, everyone reading this knows that speed is distributed along a parabolic curve. Until the 4.6 change, I really didn't put much deep thought in this because I fundamentally know how parabolas work, and through observational data, I noted that after 80-85 display speed most players are about as fast as one another.

Read that last sentence again, and think about it for a minute. We like to complain about players dropping 95 SP WRs to RB and picking up big chunk yardage, but those big chunk plays are executed almost as effectively with a 92 SP RB or an 88 SP RB or an 84 SP FB (in the case of the Slot Out).

Let's get to the maths, I know that there are maths

So here's my question, and this is pure assumption based in what I would do if I were converting a linear attribute to a parabolic one: what happens if we treat the value on the player card as the X-axis on the parabola and not the Y axis?

This makes sense from a pure math and coding standpoint. You generate a random value (Speed - here the X value) and then plug that value into your speed algorithm to get the real game value (here the Y value).

We already know that a display speed of 0 does not mean that the player will not move. They are just very slow. So let's assume that the Y intercept for the parabola in 4.5 was (0, 5), so that a player with a display speed of 0 would have a real speed of 5.

Which looks like this:



Now let's plot the speed values for positions and what the real speed would be using the same parabolic function:

Position Max Speed - Display Speed - Real Speed

DT - 80 - 96.2
DE - 84 - 97.6
LB - 89 - 98.9
RB - 92 - 99.3
WR - 95 - 99.8
DB - 96 - 99.8

This shows why your fastest LBs are able to track most players, including your fastest WRs, and only technique separates the best RBs from the fastest LBs.

Now let's assume that JDB raised the 0 speed to 25, which sets the Y intercept to (0.25). Here's what display speed the players need to hit to see the same speeds.

Which looks like this:



Position Max Speed - Display Speed - Real Speed (New functional max)

DT - 77 - 96.2 (DT - 80 - 97)
DE - 82 - 97.6 (DE - 84 - 98)
LB - 88 - 98.9 (LB - 89 - 99)
RB - 90 - 99.3 (RB - 92 - 99.5)
WR - 95 - 99.8 (WR - 95 - 99.8)
DB - 96 - 99.8 (DB - 96 - 99.9)

Now let's get real crazy and assume that JDB raised the 0 speed to 50, which sets the Y intercept to (0, 50), which is closer to what I think happened.

It looks like this:



Position Max Speed - Display Speed - Real Speed (New functional max)

DT - 73 - 96.2 (DT - 80 - 98)
DE - 78 - 97.6 (DE - 84 - 98.7)
LB - 85 - 98.9 (LB - 89 - 99.4)
RB - 88 - 99.3 (RB - 92 - 99.7)
WR - 94 - 99.8 (WR - 95 - 99.9)
DB - 94 - 99.8 (DB - 96 - 99.9)

If my line of reasoning is correct, this explains why a 70-80 SP DE breaks down the line faster than a WR can break coverage because they are functionally moving as fast as most WRs.

What does this mean for speed minimums?

Here's a quick example without a lot more numbers. If you were setting a speed minimum of 80 for certain positions, like I do for running backs and linebackers, then you just need a speed of between 72-77 to get the same results.

Here's a table that goes with the three graphs above. I created this when I was thinking through what to set minimum attribute ratings at.

It looks like all players will hit a max speed somewhere between 86-89 display speed points, which means higher speeds are only fractionally faster than 86. There are some things that impact speed that we're not talking about in this thread yet that we should start making a list of.

I think the observational data is interesting. I tend to think in terms of absolutes when thinking through what attribute is ideal to have.

Here are the comments that IoT dug up:



The bolded bit above is where things get the most interesting to me. So from like make a somewhat serious assumption that you can plug in acceleration display numbers for force to determine how much Acceleration (Force) a heavier weighted player would need to equal the same Acceleration (Force) of a lighter weight player.

For this example, let's keep some things constant and assume both players have the same Max Speed and have the same amount of time to achieve that speed. If we treat those as constants, we get an equation that looks like this:

F(1) / m(1) = F(2) / m(2)

F(1) = (F(2)/m(2)) * m(1)

Now let's keep on this somewhat serious assumption that we can just plug player weight in for mass, and you get a table that compares the acceleration a CB or LB needs to match the same "burst" as a WR:

WR CB LB
50 48.5 60.2
55 53.3 66.2
60 58.2 72.2
65 63.0 78.2
70 67.9 84.2
75 72.7 90.2
80 77.6 96.2
85 82.4 102.3
90 87.3 108.3
95 92.1 114.3
100 97.0 120.3

CB AC = ([WR AC] / 197) * 191
LB AC = ([WR AC] / 197) * 237

Because LBs have a max display acceleration of 89 (if that's accurate), using our quick model, we should note that a WR with 75 or more acceleration will always have faster access their full speed than an LB will have faster access to theirs.

This matches kind of close to what I've seen minus all the code around fatigue and ball carry sustaining max speed over time.