PDA

View Full Version : Converting 40 times to NCAA 10 SPD



JeffHCross
05-12-2010, 10:12 PM
Over the weekend I started using a much more thorough list of NFL draft eligible players. Ended up recording the 40 times and SPD ratings of over 1100 players. Some of them may not have been accurately recorded, but that's one reason for the increased sample size ... it should all come out in the wash. And, for the most part, it did.

The Results (for the so inclined)

I validated my original equation for All Positions. Even with a five-fold increase in players, there was less than 1% difference in the "All Players" equation for the original list and the new list.
It's obvious that some positions have limits on SPD ratings. This is a huge impact to trying to determine SPD by position. Because the SPDs had some inaccuracy, this meant that about 45% at certain positions were above the Trend Line, 45% were below it, and very few were actually on it. The Trend Line was nearly flat, even though the SPD range for the position was fairly wide. That would result in very inaccurate results for my Per-Position SPD conversion.
Offensive Lineman are horribly slow compared to other positions, in SPD. No Lineman breaks 80 SPD. Very few break 70. Yet the fastest in the draft, Bruce Campbell of Maryland, ran a 4.75 40. Campbell was 55 SPD in NCAA 10. Just to use two examples, Colt McCoy is listed at 4.79 with 80 SPD and T.J. Pitts (Memphis) lists a 4.76 with 89 SPD.
Just to emphasize the last point, the "average" for a DT is a 5.13 40 and 65 SPD. The "average" OL is a 5.25 40 and 57 SPD.


The Equations
If anyone's actually reading this, this is probably what you're really interested in.

I decided to take the breakdown one step further this time, and include categories for the different positions. The reason for this is that some positions in NCAA are designed to have limits in max SPD and min SPD. If I'm trying to convert real 40 times to SPD ratings, I have to take this into account. Combining similar positions into categories allows for more accuracy in the conversion while making sure that the conversion for my QB isn't affected by a DL's SPD rating.

Off Backfield = Quarterbacks and all Running Backs
Receivers = Wide Receivers and Tight Ends
Trenches = Offensive and Defensive Linemen
Back 7 = Linebackers and Defensive Backs

Slope Y-Int equation
All Players -40.23 271.64 SPD = -40.23 * (40-time) + 271.64
QB -21.67 178.97 SPD = -21.67 * (40-time) + 178.97
HB -13.39 150.8 SPD = -13.39 * (40-time) + 150.8
WR -13.42 150.98 SPD = -13.42 * (40-time) + 150.98
TE -15.16 154.51 SPD = -15.16 * (40-time) + 154.51
OL -6.15 89.21 SPD = -6.15 * (40-time) + 89.21
DT -11.79 125.34 SPD = -11.79 * (40-time) + 125.34
DE -8.85 121.11 SPD = -8.85 * (40-time) + 121.11
LB -6.03 111.36 SPD = -6.03 * (40-time) + 111.36
CB -3.43 106.3 SPD = -3.43 * (40-time) + 106.3
S -8.79 129.26 SPD = -8.79 * (40-time) + 129.26

Off Backfield -33.76 241.35 SPD = -33.76 * (40-time) + 241.35
Receivers -24.96 202.7 SPD = -24.96 * (40-time) + 202.7
Trenches -34.21 239.24 SPD = -15.47 * (40-time) + 140.09
Back 7 -22.62 191.51 SPD = -22.62 * (40-time) + 191.51


And now some examples:
Examples QB HB WR TE OL DT DE LB CB S All Players Off Backs Receivers Trenches Back 7
4.29 86 93 93 89 63 75 83 85 92 92 99 97 96 92 94
4.43 83 91 92 87 62 73 82 85 91 90 93 92 92 88 91
4.56 80 90 90 85 61 72 81 84 91 89 88 87 89 83 88
4.71 77 88 88 83 60 70 79 83 90 88 82 82 85 78 85
4.92 72 85 85 80 59 67 78 82 89 86 74 75 80 71 80
5.03 70 83 83 78 58 66 77 81 89 85 69 72 77 67 78
5.27 65 80 80 75 57 63 74 80 88 83 60 63 71 59 72

(I hope you can tell that the categories produced much better results than the individual positions)

Just while I was writing this up, I've already made a change, but I hope that's my last one. Originally, my categories for defensive players had OLs and DTs in one, DEs and LBs in another, and CBs with Safeties in the final category. This didn't produce satisfactory results, especially for the defensive backfield. By grouping the DEs in with the other lineman, and moving the LBs with the defensive backfield (Back 7), the results were much more satisfying.

For example, originally a 4.29 DB was listed at 92 with a 5.27 DB at 83. 94 and 72, respectively, are much better in comparison to the other ratings.

Hope this is of use to, well, anybody besides me. :D

Rudy
05-13-2010, 04:54 AM
I don't think you should have max or mins by position though. I like global attributes to mean the same for every player. Same goes for strength as long as weight is accounted for properly. I'm hoping a 3-4 DE in this year's game will require more weight. I don't want guys to get away with playing someone 225 pounds at that position. Curious to see if the locomotion engine makes this more relevant.

JeffHCross
05-13-2010, 09:54 PM
I don't think there should be max or mins either, and there aren't any maximums or minimums in what I ended up with. There are definitely specific ranges in the test data I used, even though it was 1300 players. But the equations I derived from them don't have any set ranges.

One reason I combined different positions into the Offensive Backfield, Receivers, Trenches and Back 7 categories is that combining fast players with slightly slower players (e.g. HBs with QBs, WRs with TEs, DL with OL, DBs with LBs) allowed me to get two sets of max/mins. The resulting equation stretches much better across both than it originally did across just one. Cornerback was the best example of this. In the original data set, you would have been hard pressed to get a CB lower than 85 SPD. That was based on the data I had, which had very, very few CBs with slow times and low SPD ratings.

In comparison, the LBs were, largely, slower. So combining the two gave me a range that, for a 4.29 to a 5.27, ranges from 94 to 72, rather than 92 to 88 (the original DB range).

SPD does mean the same for each player. Just, according to Donny Moore, certain positions have ranges for SPD that the players won't go outside of. In practice, this isn't completely a bad thing. I mean, are you really going to find a receiver or DB that runs a 5.00+ 40? No way. So there's a reasonable cutoff point that you can use.

The one big problem with that is that the range for linemen appears horribly low.

Rudy
05-13-2010, 10:00 PM
Those tables definitely look better now.

JeffHCross
05-13-2010, 10:59 PM
Hahaha, yeah. The code blocks were just there so I could post them up. That's why I was clamoring CDJ for the 'table' code to be added last night.

cdj
05-13-2010, 11:11 PM
Hahaha, yeah. The code blocks were just there so I could post them up. That's why I was clamoring CDJ for the 'table' code to be added last night.

If there's anything else you need re: tables, let me know. I see they are different than what other sites have, but hopefully they work (for now at least).


Are you anticipating work on this formula based on what you've heard on the changes to SPD & ACC in NCAA 11?

JeffHCross
05-13-2010, 11:43 PM
The ACC formula I didn't develop until after the Locomotion changes were posted. So the ACC work was all done with Locomotion in mind. Overall, I don't think, even with Locomotion, that they're going to be changing the SPD and ACC ratings very much. The effect may change, but the overall mindset and how Donny (and whoever does NCAA ratings) approaches SPD and ACC shouldn't change much.

In other words ... I'm hoping I'm right :D