Absolute Wins Produced (December 8, 2003)
David "BaseRuns" Smyth talks about his Absolute Wins Produced. I'll invite David to write a one-stop-shop article for people to reference, but in the meantime, this fanhome thread is worth reading. I wrote the following:
As for the various win impact methods, you have to decide what perspective you like the most, and use that method:
Perspective 1
I look at things in real-time, like a manager or fan does, and want to attribute the win impact as it happens. This means that I must assume a random distribution of events (centered around expectations of player matchups and managerial tendencies) for all future events.
Perspective 2
I look at things after the game is over, and try to attribute value to the various performances, given that I know all future outcomes. I assume that value was only created by players on the winning team (meaning performances that led directly to runs scoring or not).
Going 4-4 in innings where your team scored no runs, yet won the game had, essentially, very little value. (You can argue it had some because it let your team send an extra batter and prolonged the inning, but now you get into the what-if / probability scenario, and this perspective does NOT like this.)
Perspective 3
I think that all performances are statistical random variations centered around the players/park matchups, and therefore, I don't care whether my team won or lost. I just want to know what would have happened, on average, to a team, if I were able to insert this player into it.
This is the seasonal perspective, where you look at a player's line, and simply use a simple runs-to-win converter to figure out his theoretical win impact on a theoretical team.
I think that about covers it.
I don't really see anyone as being right or wrong here, since this is a question of perspective or opinion.
[David's approach is a] straightforward approach that airs its dirty secrets so that it makes it very easy for us to accept or reject them.
--posted by TangoTiger at 10:36 PM EDT
Posted 10:20 a.m.,
December 9, 2003
(#1) -
ColinM
Here is a brief summary of my problems with Davids method:
- I'm not sure what question this method is trying to answer. What is meant by "value" in the context of this system? I cannot come up with a reasonable definition of value that does not include a relative comparison.
- The only relative comparison that makes sense for the way AWP is constructed is value over 0. If this is the case, AWP does answer the question. However, I propose that 0 value does not exist in the real world. There has never been a player or team in the history of baseball that had 0 chance of making a positive contribution in a given game.
- If you then want to compare a player or team to some replacement level above 0, you absolutley HAVE to consider hitter's performance in losses and pitcher's in wins. Because there is always a chance, however slim, that the use of a replacement(s) could have resulted in a different end result. This is true for both hitters and pitchers regardless of the actual end result.
- Finally, the world is based on probabilities, it really doesn't matter if this perspective likes it or not. You can't really say that if Joe Carter hit a 2 run homer in the first inning of a 2-1 win that the Jays would have lost without it. If he didn't hit the homer the whole sequence of events changes and we don't know what would have happened. We only have probabilities. Not sure how this relates exactly to AWP, I just think it should be kept in mind before handing out absolute credit.
So if value above 0 is what you want, then I suppose this will give that to you, I just don't think it has any relevance in the real world. And you have to leave it as TOTAL value above 0. You CANNOT start introducing AWP over replacement because you would have to consider that a replacement hitter could have produced a win in games which the team has lost. And at this point you must start considering hitters performance in losses to account for that.
Posted 10:55 a.m.,
December 9, 2003
(#2) -
tangotiger
Colin,
David's process assigns "credit" (I shouldn't have used the word value) to players who participate in scoring runs in games in which a team won. So, AWP would be similar to runs scored and RBI. It's a type of counting stat.
In terms of comparison, you can figure out what an average player's AWP / out is, similar as you would figure out an average player's Runs/Game or RBI/AB or something.
And, David's perspective is not based on a "what if would have happened", and therefore, you can't bring in that perspective. It's an after-the-fact accounting system.
Whether this kind of system is relevant to your perspective is another story entirely. I don't have much use for this system (nor do I for RBIs), but it is well-constructed with regards to what it's trying to count.
Posted 11:26 a.m.,
December 9, 2003
(#3) -
Colin
If this is true Tango, then fine, it counts what it wants to count. But in scanning the fanhome thread and previous threads, I see an awful lot of talk about value in relation to AWP. And you can't have it both ways.
If it is meant to be a stat like RBI then sure, I also think it is well constructed. It adds extra information. I'm very impressed with the way David is able to explain himself and I think he has some real orginal insights within the method. But as soon as you start to talk about it as total value stat, which seems to me to be the case, it does become a "what would have happened" perspective. Because value is relative, and what it is relative to is "what would have happened". And in this case I think the points I've brought up need to be addressed. Unless you think there is a flaw in my reasoning of course, and in that case I'd appreciate knowing what they are.
Posted 11:35 a.m.,
December 9, 2003
(#4) -
tangotiger
No, I agree that we shouldn't talk about it in the sense of value as we define it. We don't hold R, RBI to those standards.
(As a side note, because the word value has morphed into 20 definitions, we should try to avoid it when possible.)
AWP is, by its name, Absolute Wins Produced. Any counting stat would have an "absolute" implied to its name.
How you relate it to "value" is another story. I can construct a measure called (Absolute) Runs Produced, and make it R+RBI-HR. And, I can start to try to make it better for comparison by creating Adjusted Runs Produced as R+RBI-HR - AB/10. Then, I can figure out what an average player has done in ARP, given the same number of outs. So, maybe Pujols is 60 Adjusted Runs Produced above the average player, given the same number of outs. Does this mean that Pujols has a "value" of 60 runs. I don't know. I just know that he was directly involved in 60 runs more than an average player, given the same number of outs.
Absolute Wins Produced is the same thing. Maybe Pujols was involved in 20 Absolute Wins Produced. And an average player, given those outs, was involved in 12 Absolute Wins Produced. That makes Pujols +8. Maybe the reference point would be outs, at that park with that team, I don't know.
The dirtiness of it is rather clean.
Posted 12:01 p.m.,
December 9, 2003
(#5) -
ColinM
Thanks for the quick reply, if we don't want to call it value added, then I suppose I have no problem with it. Like I said, extra information is good, as long as we realize what this information doesn't cover.
You're response is exactly what I was getting at with the first point in my initial objection. Before we start discussing a "value" stat, we need to define exactly what we mean by value. And I think there are certain logical requirements a concept of value needs to meet; this is what I attempted to outline in the rest of the post.
I have to say though that I find your perspectives summary a bit misleading in relation to AWP. Perspectives 1 and 3 obviously can be related to total-value stats (WPA, Lwts), so it makes it seem as if you are offering AWP as a total-value solution for Perspective 2.
Posted 6:12 p.m.,
December 9, 2003
(#6) -
David Smyth
Yes, AWP/ALP is not intrinsically a "value" stat; it is sort of a "counting" stat. The next step is to convert it into a value stat. I prefer to say convert it to a "relative" stat instead of a value stat, because that gives you the freedom to choose what you want to relate it to. My choice is a repl player on an avg team, who produces the same (relative to the team) in the team losses and the wins. That means that he produces the same proportion of the team's runs and outs in every game he appears in.
And as to how well the method "correlates" to the "real world", I gave a couple examples of World Series results, one of which was the 2003 Series. Take a look at that post, and see if those final results don't pretty much reflect a rational MVP analysis.
As far as writing an article, Tango, thanks for the invitation. I am not a very good writer by the standards of the Primer authors, so I find that a bit intimidating, but I'll see what I can come up with.
Posted 11:48 p.m.,
December 9, 2003
(#7) -
Tangotiger
You're right, it is a bit misleading.
Let me clarify:
Perspective 1: marginal win contributions in real-time
Perspective 2: absolute win contributions after the game is over
Perspective 3a): marginal win contributions, theoretical, based on seasonal data
Perspective 3b): absolute win contributions, theoretical, based on seasonal data
So, in all cases, we are simply interested in recording the win contributions, without necessarily worrying about "valuation" or "total valuation". That would be a next step,if we want to.
Win Advancements (WA), Win Probability Added (WPA), Mills' Player Win Averages (PWA), these are all perspective 1. It's a rather simple process to convert that into a "value" stat.
Absolute Wins Produced (AWP) is perspective 2. It's not clear yet if we can even convert that into a value stat and make it mean something.
Linear Weights is perspective 3a. It's a rather simple process to convert into a value stat. It's also rather straightforward to associate this to perspective 1.
Win Shares is perspective 3b. It's not clear how to convert that into a value stat. It's not straightforward (or even necessarily possible) to associate this to perspective 2.
The valuation process is the next step, and shouldn't take away from the perspectives that these metrics try to describe.