View Full Version : Measuring Player Importance in Statistics as far as W-L

08-06-2009, 08:06 PM
One of the statistics that has always interested me is how players perform when their team wins or loses. This is in part to my own personal experience, especially with my team during this past season. To me, it seemed like in games that we won, I would have multiple hits and/or get on base a few times (a game that would satisfy me). In losses, it always seemed like I had an off night, with maybe a few K's or just an overall bad game. It didn't seem (to me) that there were many games where the opposite happened (I had a bad game but won/I had a good game but lost).

Unfortunately, I never had the advanced statistics of my own team to test this theory. But today, as I was browsing the Reds' batting splits at BaseballReference.com, I noticed the "by game outcome" section. I then looked up the OPS of Reds regulars during this season in wins and in losses to get a look at who is "important" to the Reds.

Here are the stats (I excluded guys like Rolen for sample size):

Wins Losses
Willy Taveras .748 .419
Alex Gonzalez .548 .519
Joey Votto 1.145 .871
Brandon Phillips .908 .699
Jonny Gomes 1.125 .792
Laynce Nix .887 .610
Ryan Hanigan .875 .651
Jay Bruce .871 .595
Ramon Hernandez .759 .621
Chris Dickerson .755 .682
The data backs up what most of know. When Taveras is simply average (~.750 OPS), the Reds are very very successful. As was seen last night, Willy can make things happen when he gets on base. If only he could get hot (I realize that's a BIG if), I really do believe that the Reds could put a decent streak of wins together.

Other Observations:

1) Votto and Gomes have consistently hit the ball. Even in losses they have still hit well.
2) Gonzalez has been really really really bad.
3) I am a little surprised that Phillips' OPS is not higher in wins.


08-06-2009, 08:43 PM
You've drawn the causality line in the wrong direction. You said:

When Taveras is simply average (~.750 OPS), the Reds are very very successful.

The implication here is that if Taveras is simply average, the Reds will be very, very successful. This is cleary wrong. If we replaced Taveras with a .750 OPS bat, the REds would not go undefeated. In fact, the would win only a few games more than they have already -- if that. The correct way to read this data is: When the Reds are very, very successful, Willy Taveras is still simply average. By contrast, when the Reds are very, very unsuccessful, Joey Votto is still very good. In other words, they are who we thought they were. It's not like the Reds lose when Votto plays poorly and it's not like they win because Taveras plays well.

The general conclusion is this: the performance of an individual player at the plate has very little predictive value about the outcome of a given game. The player is such a small part of the entire system, that no one player has much influence over the outcome.

However, every player does have some influence -- in fact, on a per PA basis, they have the same amount of influence. The differences in their production from what they normally produce in those PA has an equal impact on our chances of winning. So at the start of a game, we can expect a given average level of production from each player. And the sum of those averages gives us the average level of production from our team. And that level of total production results in a certain likelihood of us winning. Any change in production from a given player above or below that expectation increases or decreases our chances of winning accordingly. An extra hit from Taveras helps us just as much as an extra hit from Votto. And vice versa.

For example, let's say Taveras produces an average of .2 runs per game and Joey Votto produces .8. You do that for every player and figure out that the Reds produce an average of 4.0 runs per game. On a given night, Votto might be off and produce only .6 runs. We would then expect the Reds to produce just 3.8 runs that night, slightly lessening our chances of winning. Thus, when we look at Reds losses, we find that Joey Votto tends to be worse than he normally is. This holds true for every single player. So when we look at wins and losses, we find that everybody tends to do better in wins and everybody tends to do worse in losses. But no single good or bad night from a given player has very much effect compared to the collective value of the night's his teammates are having.

Thus, this analysis doesn't really tell us anything about individual players as having different amounts of impact on our wins and looses. Votto isn't more influential on our likelihood to win or lose than Taveras is. It's just that he tends to the things that lead to victory more often than Taveras. So even when the lineup as a whole performs poorly and we lose, Votto is still likely to have performed pretty well. Where it is interesting to me is in looking at the team or a player and getting a sense of what a good day looks like and what a bad day looks like. So my observations are this:

- Votto is likely to be better on a day when the Reds lost than Taveras was on a day when the Reds won.
- When the Reds have a collective OPS over .850, they tend to win. When it's below .650, they tend to lose. That's a handful of baserunners per game and really highlights the value of getting on base and the horrible cost of making outs on the basepaths.

08-06-2009, 09:07 PM
If we replaced Taveras with a .750 OPS bat, the REds would not go undefeated. In fact, the would win only a few games more than they have already -- if that.
I understand your logic and the faults in mine. Guess I didn't necessarily use the right words to get my thoughts across.

I know that the Reds would not go undefeated if Taveras were replaced with a .750 OPS bat. What I do know is that if Taveras could OPS around .750, his speed on the bases makes this offense significantly better, leading to more wins for this team. We saw the previously mentioned "havoc" on the bases last night when Taveras got an infield single, stole two bases, and was driven home with a sac fly.

Sure, one player's performance on any given night is most likely not going to make or break the team, but over a period of time, a hot streak by Willy will result in more wins.