Bargain Shopping: Dollars Per Run
After recently going through some of the younger Athletics salaries and realizing the production the A's are getting for bargain prices, it got me to thinking about the importance of taking a player's salary situation into account.
Is there some way to figure out just how valuable having a player like Huston Street or Bobby Crosby on your team? Talent that is young, cheap and produces at the major league level. I wondered if there was some metric or way to measure how much of a bargain certain players are.
Since I'm not exactly a mathematician or statistician, I enlisted the help of Richard Brian Wade from the SportsBlog Beyond the Boxscore.
He worked out a calculation using VORP from Baseball Prospectus compared to salary. VORP is described as:
Value Over Replacement Player. The number of runs contributed beyond what a replacement-level player at the same position would contribute if given the same percentage of team plate appearances. VORP scores do not consider the quality of a player's defense.
So Richard put this equation together to try and figure out who were the best bargains in baseball this year:
( Games Played / 162 * Salary ) / VORP
An example of how it works:
Brian Roberts has a VORP of 51.6, and a salary of $390,000. The Orioles have played 87 games.
( 87 / 162 * 390000 ) / 51.6 = $4,059.00 Per Run
So then, who tops this list so far in 2005?
Here is the Top 20 position players in Dollars Per Run or DPR.
1 Brian Roberts, BAL
2 Miguel Cabrera, FLO
3 Travis Hafner, CLE
4 Jason Bay, PIT
5 Morgan Ensberg, HOU
6 Chase Utley, PHI
7 Felipe Lopez, CIN
8 Jhonny Peralta, CLE
9 Bill Hall, MIL
10 Joe Mauer, MIN
11 David Wright, NYN
12 Grady Sizemore, CLE
13 Kevin Mench, TEX
14 Clint Barmes, COL
15 Jorge Cantu, TBA
16 Ryan Church, WAS
17 Carl Crawford, TBA
18 Bobby Crosby, OAK
19 Chad Tracy, ARI
20 Coco Crisp, CLE
This list obviously includes some of the brightest young stars in the game, including our own Bobby Crosby.
One of the caveats is that it's impossible to calculate all of the players because at least 51 players have salaries that Richard couldn't locate. The first player to show up making a salary over $1 million is Nick Johnson. At some point, we're going to try and get a more comprehensive list. I would also guess that the A's Dan Johnson is probably relatively high on the list as well.
Of course, this system isn't perfect because it doesn't take defense into account, but it gives you a sense of who is producing at a bargain level. Richard also made the point that VORP is marginal runs, but we couldn't do marginal dollars because it would drastically overrate players making the league minimum.
But what makes these players so very valuable is that it allows you to allocate funds in other areas of weakness. If you have a Bobby Crosby, Nick Swisher and Dan Johnson producing at a ridiculously high DPR level, it allows you to maybe overpay a little to keep a talent like Kotsay (not that the A's overpaid for him).
It's possible there is already some measure or process to figure this out (chances are Beane and company already have it, which is why the draft is the most important day of the year for the A's). But it's why getting those 0-3 service year players who can produce immediately like Harden, Haren, Blanton, Dan Johnson, Crosby and Street are absolutely necessary to keep this team alive.
60 comments
|
0 recs |
Do you like this story?
Comments
Miguel Cabrera
by Oaktown Kid on Jul 13, 2005 11:01 AM PDT up reply actions
one thing to be careful about
it's kind of like $/win. it's nice that the a's are always up there, but what does it really mean if the devil rays are near the top, above teams like the red sox?
It isn't a perfect metric because the
by Tyler Bleszinski on Jul 12, 2005 11:07 PM PDT up reply actions
That's the idea
Here's a suggestion, though - break it up by contractual status (ie, 1-3, arbi 4-6, fa 7+). There's no question that the most efficient team possible would be mad up of the 25 best players in their 1-3 pre arbi years - but that's entirely irrelevent, since no GM could ever assemble a squad even remotely resembling that. (though, if we let Zito walk, our rotation next year might very well achieve historical levels of efficiency) Even a team full of decent, but unspectacular 1-3s would be far and away the most efficient. Every team, though, has room in their payroll for 4-6s and FAs, since 25*$350k=$8.75m and every team is at least $20m higher than that. Since salaries across the three categories are entirely unrelated, it doesn't make sense to compare the salaries of players between them.
Good point...
by Richard Wade on Jul 13, 2005 11:41 AM PDT up reply actions
the most efficient team
Of course
There's also a statistical problem that comes up when comparing teams/players of vastly different payrolls/salaries.
Just throwing some numbers out there, simple ones, for simple math. We'll pretend that minimum wage is $400k - so a minimum wage team is $10m.
Team A spends $30m and wins 65 games - $20m marginal dollars for 25 marginal games, or 1.2 marginal wins/$million.
Team B spends $80m and wins 95 games - $70m marginal dollars for 55 marginal games, or .8 marginal wins/$million.
Team A is significantly more efficient.
Now watch what happens with Team C, who spends $10.1m and wins 41 games - .1 marginal dollars and 1 marginal win. They post an astounding 10 marginal wins/$million. You see what happened? When one of the numbers is so close to zero, any increase in performance will be seen as an incredible efficiency (or any incredible inefficiency for any drops) when the actual change in performance is barely relevent.
Here's another way to look at it. One could fairly easily determine how many wins, over the last several years a team at any particular payroll (relative to the league) was likely to win. For team A, lets call it 60 games, for team B, lets say that it's 85 games, and, naturally, team C is 40 games.
Under this study, team A would be outperforming their expected record by 5 games, team B by 10 games, and team C by 1 game. By that standard, I think you could fairly conclude that team B is really more efficient than the lower payroll teams.
Well stated...
by Richard Wade on Jul 13, 2005 9:47 PM PDT up reply actions
Well...
by Richard Wade on Jul 13, 2005 12:00 PM PDT up reply actions
If you compare
I already did this...
I guess I'm just ahead of my time. ;-)
Ooops... wrong diary
by FoolshGame22 on Jul 12, 2005 10:49 PM PDT up reply actions
Something else interesting
C: Joe Mauer
1B: Chase Utley (he's played first)
2B: Brian Roberts
3B: Morgan Ensberg
SS: Felipe Lopez
LF: Jason Bay
CF: Miguel Cabrera
RF: Travis Hafner (scary proposition)
DH: Jhonny Peralta
by Tyler Bleszinski on Jul 12, 2005 10:59 PM PDT reply actions
i really like . . .
"jhonny"
Love and kisses,
Jheff
sorta... (OF)
C: Joe Mauer
1B: Chase Utley
2B: Brian Roberts
3B: Morgan Ensberg
SS: Felipe Lopez
LF: Jhonny Peralta
CF: Jason Bay
RF: Miguel Cabrera
DH: Travis Hafner
What about other things like
by WhiteElephantGuy on Jul 13, 2005 12:01 AM PDT reply actions
That's...
by Richard Wade on Jul 13, 2005 11:43 AM PDT up reply actions
Games Played
Doesn't seem like games played by the team would be much of a factor, since it's close to the same for all teams at any given time.
Awesome Calculation by My Opinion
While Crosby, Swisher, Kielty, and Ellis tops our position players and Johnson is worth something, his age and position makes him... well you know.
The only negatives I see in your equation are that the list omitted pitchers, but one thing that really erks me is this:
By using VORP and salaries, you are assuming that every position gets paid the same, but in reality it doesn't, and you don't take into account age which is really important, and what arb year the player is in.
Swisher is obviously worth more than the 5.1 VORP, 316k, because he plays LF, RF, 1B with great defense, he's young, a rookie, and patient for the fastball.
Another statistic could be an anomaly like Dontrell Willis who actually made the Marlins sells out a few games he started in, in his first season, which would make his worth the benchmark.
Of course in order to incorporate all of this, you would need a mathematical model (perhaps a linear model).
Something you might consider adding to your equation would be to divide the players salary to the teams 25 man salary, calculating each players worth to the team, because MLB salaries varies greatly among teams.
But a great calculation none the less, awesome work.
by Instant Replay Umpire on Jul 13, 2005 2:02 AM PDT reply actions
Not surprisingly...
The only real deviations are Thomas and Ginter, which really proves only one thing... no matter how inexpensive you are, you can't hit like a piece of crap and hope to rank higher than even the most expensive players.
by FoolshGame22 on Jul 13, 2005 2:39 AM PDT up reply actions
Hmm...
I think Vorp takes into account ball park effect, but the main weakness I can see would be the injury rate of the player, and age with projected stats. Defense of course is always hard to calculate as well.
But I hope everyone is begining to find out why we traded Hudson and Mulder, because the difference between a Hudson and Mulder, and Haren and Blanton is not only millions of dollars, but slim compared to Chavez and Kendall vs. Teahen and Melhuse
by Instant Replay Umpire on Jul 13, 2005 11:31 AM PDT up reply actions
But of Course a Big Market Club
But the A's would rather have Haren and Blanton, and spend the money saved on other things.
by Instant Replay Umpire on Jul 13, 2005 11:36 AM PDT up reply actions
One thing is certain, however...
by FoolshGame22 on Jul 13, 2005 12:36 PM PDT up reply actions
Is everything you say
by WhiteElephantGuy on Jul 13, 2005 1:46 PM PDT up reply actions
I just thought it was interesting who the #1 DPR
by FoolshGame22 on Jul 13, 2005 2:28 PM PDT up reply actions
To answer a few of your points...
I agree that it would be interesting to look at each position individually, but I disagree that it is necessary and also don't believe that age and arbitration year are necessary for determining your best value. Those would simply give you more context. There's nothing inherently wrong with applying the same measurement to all players if you accept the idea that there are genuine underlying reasons for some of the differences. That being said, I intend to produce lists taking into account all of the things you mentioned in the future.
On defense, I've considered using WARP instead of VORP, but am fairly happy with looking only at offensive production, since we have a much better idea of how to measure that than we do defense. Also, we all know that there are other factors outside the measurement that can be taken in to consideration and should be. Dollars/Run tell you part of the story, but not all of it as Blez made clear.
It's difficult to determine what brought people to the ballpark, so I don't think that's a reasonable thing to do, though it would be interesting if you could.
On the model, if you have any thoughts on how to go about that, I'd be interested in hearing it.
I also like the idea of the % of team's payroll, but that would tell you something drastically different than what DPR does.
Thanks for the interest and kind words.
by Richard Wade on Jul 13, 2005 11:56 AM PDT up reply actions
To Calulate what Brought People to the Ballpark
Just take the starting pitcher, and the team against, followed by the day (mon, sat, etc), discounts or promotions that might have effect, and if there are abnormalities, then you have something like Dontrelle Willis.
You don't really have to take into account wins, because a starting pitcher pitches every five days.
Of course locations are different too, it's a known fact that Pettite and Clemens bolstered the Astros $$$, while Jason Giambi boosted the A's $$$ by just leaving.
Hard calculation to construct, but I think it's do-able. However in the end, I would think that only a handful of players really make a difference in $$$, (Bonds, Willis, Clemens, Milestones, possibly Ichiro) come to mind. Nothing more makes a difference than just winning.
As for the age and arbitration year being necessary for determining your best value, I think that might be effective if you want to project future value, and not just present value.
But now that I think about it, the DPR equation you guys constructed is brilliant in terms of being a statistic. From the DPR statistic, it can be a platform for other statistics. You could probably incorporate into the ultimate equation (something I can not give here) but it's basically the winning percentage equation for a 162 game season. Taking into account a stat to predict how many runs you will score, how many runs will be scored on you, and the total amount of money you can spend on your 25 man roster.
by Instant Replay Umpire on Jul 13, 2005 2:13 PM PDT up reply actions
Couple of reasons...
by Richard Wade on Jul 13, 2005 9:51 PM PDT up reply actions
win shares
by brian38 @ Athletics Nation on Jul 13, 2005 10:03 PM PDT up reply actions
If you replace VORP with Win Shares...
- Brian Roberts, BAL
- Jason Bay, PIT
- Travis Hafner, CLE
- David Wright, NYN
- Miguel Cabrera, FLO
- Morgan Ensberg, HOU
- Joe Mauer, MIN
- Grady Sizemore, CLE
- David DeJesus, KCA
- Chase Utley, PHI
- Coco Crisp, CLE
- Kevin Mench, TEX
- Carl Crawford, TBA
- Felipe Lopez, CIN
- Clint Barmes, COL
- Chone Figgins, LAA
- Antonio Perez, LAN
- Adam LaRoche, ATL
- Chad Tracy, ARI
- Bill Hall, MIL
by Richard Wade on Jul 14, 2005 12:56 AM PDT up reply actions
A's in the $/WS Top 100:
- Bobby Crosby
- Marco Scutaro
- Nick Swisher
- Mark Ellis
- Bobby Kielty
by Richard Wade on Jul 14, 2005 12:59 AM PDT up reply actions
Well...
by Richard Wade on Jul 13, 2005 9:52 PM PDT up reply actions
Most important sentence
But what makes these players so very valuable is that it allows you to allocate funds in other areas of weakness.
The double-edged sword of this type of statistic is that this type of individual measurement punishes an elite player. Just how much more should be a player get paid for having a .990 OPS as opposed to one closer to .875? It's kind of an exponential thing after a certain level is it not? But, really, the most important thing is how well a team does with the amount of money it pays for its player's services. And, in that regard, statistical production doesn't mean jack but the amount of wins certainly does. And, even more important than all of that junk is how profitable a team is (long term) when you truly get down to it. Baseball, with its revenue sharing schemes and "fan base" rights bestowed upon cities to fend off competition are things that alter the market in such a way so as to mask what a team would be willing to pay to remain profitable and where they'd prefer to locate in order to seek profits. Beane has shown to be the best GM at maneuvering through baseball's oddities and producing profit for his bosses. Though profit isn't really the sexy thing that fans want to hear about.
Well...
by Richard Wade on Jul 13, 2005 11:59 AM PDT up reply actions
Intersecting bell curves
As everyone knows, the distribution of talent at the ML level is the far right end of a bell curve: huge number of RP-level guys, modest number of slightly-above-RP-level guys, modest-to-shrinking number of superior players, tiny number of elite superfreaks.
I haven't looked at the numbers, but I would make a semi-educated guess that the distribution of ML salaries is itself constituted much more like a complete bell curve: few but not insignificant number of ML-minimum guys, huge number of $4-6M/yr guys, and few but not insignificant number of $8-15M/yr guys.
These two graphs simply don't map onto each other. Based on talent/productivity alone, there should be a lot more guys making ML-minimum level money, and a lot fewer guys making $4-6M/yr.
Exploiting this disparity, more than anything else, is the key to Beane's methodology.
very good point
I think what Beane talked about in his interview was that the A's can't afford to pay for past performance, and so they have to find players who will perform above replacement level but be paid the league minimum, which brings us back to Harden, Crosby, Johnson etc.
imho, the beane strategy is pretty clever.
Precisely
wow, interesting
from baseball prospectus, updated today
NAME H/9 BB/9 SO/9 HR/9 BABIP ERA VORP
Barry Zito 7.4 3.5 6.3 1 0.245 3.91 21.1
Dan Haren 8.8 2.6 6.7 1 0.288 4.08 13.6
Joe Blanton 8.4 3.1 4.3 1.2 0.252 4.44 10.4
Kirk Saarloos 9.2 3 2.5 0.4 0.28 3.87 18.2
Rich Harden 6.7 3.2 7.7 0.5 0.253 2.53 19.8
Justin Duchscherer 7.4 1.7 8.6 0.6 0.282 1.49 19.1
Huston Street 6.1 3.3 9.2 0.2 0.267 1.67 17.6
Keichi Yabu 9.1 4.3 7.4 0.7 0.318 4.3 1.2
Kiko Calero 8.3 2.2 7.6 1.4 0.264 4.32 4.1
Juan Cruz 11.2 7.3 10.4 1.5 0.385 8.49 -13
Ricardo Rincon 8.3 3.7 6.2 1.2 0.266 4.15 2.9
Seth Etherton 8.2 2.5 5.1 2 0.222 6.62 -1.6
Ryan Glynn 12.7 3.7 7.9 2.6 0.345 6.88 -5
Octavio Dotel 5.9 6.5 9.4 1.2 0.222 3.52 3.9
Tim Harikkala 11.4 2.8 5 2.1 0.31 6.39 -0.9
Britt Reames 15.9 3.2 6.4 3.2 0.421 7.94 -2.4
Ron Flores 2.7 0 5.4 0 0.111 0 2.1
Jairo Garcia 9 0 0 0 0.25 0 0.6
i don't have the salaries in front of me
And Cruz is really low....
i don't really understand VORP though
replacement level and VORP
Baseball Prospectus defines replacement level for a position to be about 75% of the league average offensive performance at that position, and VORP is a measure of how many extra runs a player has contributed to the team's offense compared to a replacement level player given the same playing time. So, for example, they estimate that the A's have scored about 19 more runs with Crosby in the lineup than they would have if his playing time had been given to German or some other scrub.
Vorp
Next what the formula is supposed to determine is what value a specefic player has compared to the benchmark replacement player.
Thanx guys
Bradford for Payton
http://sports.yahoo.com/mlb/news?slug=ap-athletics-redsoxtrade&prov=ap&type=lgns
Question is ... where will Payton go next and for what?
One might say:
bang for buck...
Baseball teams have done a horrible job paying for future performance, being able to accurately predict future performance, and pay for that vs past performance is what sets aside the good teams from the bad teams.
Here is a graph of a player's VORP and their respective salary. Data is for the 2004 season, in an ideal world, the points on this graph would follow the same line, but the salaries are all over the place...
http://img.photobucket.com/albums/v421/Strng1dah/untitled.jpg
By the way, why the hell do we need Payton?
by TonyArmas on Jul 13, 2005 4:50 PM PDT reply actions

by 























