Player rankings - a statistical analysis

[fightonusc]

So I've started to do some statistical analysis on the best way to rank the values of RBI Baseball batters. There's a key question that I'm trying to finalize before settling on anythnig:

Looking at the three measurable player attributes (power, speed and contact), how important is each attribute to a player's overall worth. Meaning, is the offensive value of a player 50 percent power, 30 percent speed and 20 percent contact? Or maybe 40 percent power, 25 percent speed and 35 percent contact? You get the idea. Of course, this is overall worth - if you're looking at leadoff men versus clean-up guys, you might weight it differently, but I'm trying to compare everyone together here.

I tend to think of it about 60 percent power, 25 percent speed and 15 percent contact. Does this seem about right to you guys? If not, let me know why it should be different.

[Stock]

If p = Power Factor, c = Contact Factor, and s = Speed Factor, and S = the average speed factor on the game (excluding pitchers):

Player Statistic = [(p - (c/2)) * (1 + ((s/S - 1)/2))]

If you were going to use YOUR equation, you need to rebase each Factor by dividing it by the average on the game.  This is because, Speed is only from around 117 to  148.  But power numbers range from the 700s to the 950s or so.  Contact numbers range from around 4 to 40. (ie. they are not on the same scale)
If you were going to weight average the factors, your contact and speed factors would over dilute the power factors, even if you did give them a super small weghting.

[Ryno23]

e=mc^2  bitches

[Stock]

Nice...
I was bored one day in a stats class in college about 6 years ago.  I did the same thing that Mr. Fightonusc is trying.

I don't remember the EXACT formula I used, but is was something like I showed.... at least the idea about rebasing all the variables.
No I get to create models that are not as exciting as creating RBI Statistical models (Insurance) .

[Ryno23]

Would this all not be relative to each player?

Power isn't important to a player like vince coleman.

Speed is not big deal to DaEvans.

Of course, speed would be more important to Evans than power would be to Coleman.

[Stock]

I think in RBI, power is almost everything.  I think speed should be used to give the edge to players with similar power stats.  I don't think many would argue that Evans is much more valuable than Coleman.
However, if you took players w/ similar power such as Mitchel, Hernandez, Guerrero, and Raines; Rains definitely gets the nod for his speed.

Also, depends on how you play.  For example, if you are playing straight ball/no knucklers (or as some people would call it-batting practice), contact means less since you should be able to pull every ball and hit the sweet spot.  In anything goes, contact means more since you will be hitting several balls to the opposite field and off the end of the bat, etc..

[fknmclane]

Numbers scare me.  I probably shouldn't even bother posting on this topic.

usc, I guess you're trying to come up with a Bill James like equation to rate RBI players?

I think power should obviously carry the most weight, probably 60 or 65%.  I think a completely player should be fairly quick, so 30% for speed.  That leaves just 10% for contact, but it really isn't all that important. 

I'd have to agree that style of play probably comes into play (little repetitive there, sorry.)  If you're a hardcore curve player, contact is more important because a guy can do more with a shittier pitch.

[fightonusc]

Thanks for the feedback.

Stock is right that the power ratings, contact ratings and speed ratings are all on different ranges. So, what I did was take percentages of the maximum rating in each category. Meaning, for example, that Andre Dawson's 954 is considered to be the maximum power ranking, or 100 percent. Pedro Guerrero's power number is 849: That's 89 percent of Dawson's power. The bottom number is Rafael Santana's 708, which is 74.2 percent.

You can do the same thing for the speed ratings, with Vince Coleman's 148 being the maximum, and the four players at 118 (Armas, Spilman, Buckner and Sullivan) being at 79.7 percent.

The big problem comes from the contact rating, because in terms of raw numbers, the the ratio betweem the best contact number (Boggs with 4) and the worst (three guys at 38) is so vastly different, that you can't just use a straight percentage. (If you did, the three players each right after Boggs with a 6 contact rating would still only be at a 66.7 percent - way too much of a difference.)

So what I did was this: I set a baseline that the worst contact hitters (the 38s) would be a 75 percent of Boggs' 4 (a balance of the percentages between best and worst of the power and speed rankings). From there,it reasons that each contact point would be worth 0.73 percent (.250/34 - the difference between the worst and best numbers).  By doing that, I could assign a more sensible percentage of maximum contact to each player.

I went with the equation that power is 60 percent, speed is 25 percent and contact is 15 percent, and simply multiplied each relevant number by the necessary amount to get a weighted number. For example, here's Reggie Jackson's numbers:

Power: .991
Speed: .865
Contact: .861

So, the equation to get his weighted score would be:

(.991*60)+(.865*25)+(.861*15)

I've attached a Excel sheet with all of the hitters ranked with this weighting applied. (It's the column on the far right.) I've also included, for reference, where each player ranks in the three categories, and their ranking if you simply add those numbers together without weighting.

The weighted numbers seem to conform to conventional RBI Baseball wisdom about the best and worst players:

Top Ten

1.  Andre Dawson, 95.10
2.  Jose Canseco, 94.14
3.  Mark McGwire, 94.00
4.  Reggie Jackson, 93.98
5.  Eric Davis, 93.84
6.  George Bell, 93.72
7.  Don Baylor, 93.35
8.  Darryl Strawberry, 93.29
9.  Dale Murphy, 93.04
10.  Kirk Gibson, 92.79

Bottom Ten

111. Mark Davidson, 82.12
112. Willie Randolph, 81.72
113. Roy Smalley, 81.47
114. Lee Mazilli, 81.29
115. Steve Lake, 81.18
116. Al Pedrique, 81.17
117. Marc Sullivan, 80.83
118. Wally Backman, 80.38
119. Spike Owen, 80.24
120. Rafael Santana, 76.75

Thoughts?

[Stock]

I think it looks pretty good.  From your list it looks like you may have put too high of a rating on speed.  This may be why players lie Armas and DaEvns are left off of the top 10.

I would probably only use 2 variables.  Power and Speed.  I would alter the power number by subtacting a percentage of the contact number from it.  Then weight accoringly.

For example if a player has 900 power and 20 contact.  Maybe take 900 - (0.5 * 20) = 890.  Use the 890 for your power/contact number.  Then adjust this how you adjusted YOUR power number. 
Now you just have two variables to weight average.  I probably would only weight speed at 15% max.

All in all, your method looks good.  No matter how you do it, there are going to be some players that don't look right in the rankings.

[fightonusc]

I see where you are coming from, Stock. I do think that it's a difference between straight pitch and Anything Goes about the importance of the three variables. Speed can be the difference between a double play being turned or not, and in anything goes, being able to advance an extra base is much more important (small ball vs.waiting for a three-run homer).

DaEvans was actually No. 11 in the list, so he's right there. As for Armas, he's No. 26. Although he's No. 8 in power, he's at the bottom in speed and No. 96 in contact - basically, he's a home run or an out, and a big, big double play risk.

I feel like if someone values more speed and less contact, or more power and less speed, they can simply adjust the percentages in the weighted numbers and get personalized results.

[BDawk]

HRs are where it's at.
Singles are for pussies.

[Stock]

Nice job 'usc.
For shits&giggles, I did my rankings.  Mine are actaully very close to yours.  We are only 1 player different in the top 10.  I had Murphy 11, and DaEvns 6.

For the bottom 10, I had Schroeder, Barrett, and Herndn worse than Sullivan, Lake, and Mazilli.

I attached the file.

[fightonusc]

Yeah, pretty similar. Both of our rankings place about equal emphasis on power first, then speed and finally contact. I tend to value power slightly less than you, but it's still far and away the most important thing for both of us.

[fightonusc]

HRs are where it's at.
Singles are for pussies.

I would love to play you at the Cousin to the COTUT. With that attitude, I will kill you. I'd be willing to even play you for $1,000 of JoeDirt's money. What do you say?

[BDawk]

Do you promise to hit only singles?

[fightonusc]

Do you promise to hit only singles?

No. But I promise that you will...

[Ryno23]

HRs are where it's at.
Singles are for pussies.

Holy shit!  We have a celeb DeeNee'er.

BDawk is Dusty Baker.

PS:  I hope you die!  (if you're actually dusty)

[BeeJay]

Dusty Baker = the worst thing that's ever happened to the Cubs franchise.  Not only is he a racist, idiotic asshole, but he also ushered in an era in which the lovable losers became a bunch of headhunting jerks.  Fuck Dusty Baker.

[Stock]

No kidding!!  People dislike the cubs more now that they are actually contenders than they did when they absolutely sucked. 
How does that happen?

[ultimate7]

I not a huge fan of Baker's managerial ability, but I don't beleve he's a racist, and I don't think the Cubs have become headhunting jerk.