Offensive Consistency - Quick And Dirty

[rluzinski]

There has been alot made of the Brewers supposed inconsistency on offense and how it has cost the Brewers wins this year. While it's true that a more consistent offense wins more games, have the Brewers been less consistant than the average team? What I'm about to show isn't exactly right, but it will point us in the right direction. I hope to write a more in depth and statistically correct article after the season ends, but I thought I'd show you guys some work so far.

 

The first step was to simply find runs per game (RPG) distribution for each team and calculate the statistical variance. A low statistical variance would indicate that a team's daily runs scored centers closely around their average. For example, a team that averages 4 runs a game would have a low variance if they played alot of games where they scored 3, 4 or 5 runs. A high variance would show that a team might score 1 run one day and 10 the next.

 

It would be tempting to just sort the variance table in accending order and proclaim the team with the lowest variance the most consistant. That would be dead wrong, however. Teams that have a lower RPG average would naturally have a lower variance. For example, a team that averages 1 RPG would have a very low variance but it really wouldn't mean much. A higher scoring team, on the other hand, would have a much higher variance, but it would be a result of normal statistical variance and not lack of offensive consistency. What we need to do is separate the variance into what would be naturally expected and what can be contributed to "consistency."

 

This is where this post branches off to the "sorta right" part. For a given RPG average for a team, we can calculate what's the expected number of games that team should score 0 runs, 1 run, 2 runs, etc... We could then compare that to their actual RPG distribution, declaring a team offensively consistant if their actual variance is smaller. I'm not going to do that here, however.

 

A simpler way (and one that might make more sense intuitively) is to simply find the average variance all teams have, adjusting for the higher expected variance a higher scoring team would have. Below is a graph doing that:

 

http://i14.photobucket.com/albums/a345/rluzinski/off_cons.jpg

 

The line is the expected variance and the points are the actual variance for each team. Teams that are found below the line are more offensively consistant than expected; teams above are more inconsistant. It confirms what we expected; the more runs a team scores, the higher their RPG variance (I took out WAS and BOS for this step, since they were outliers). Now all we have to do is show the actual difference between the expected and actual:

 TEAM        AVE        VAR      lin VAR ANA        4.63        6.1        3.54 SLN        5.04        8.9        1.81 BOS        5.79       10.7        1.78 WAS        3.90        6.2        1.75 ATL        4.80        8.4        1.68 TBA        4.70        8.4        1.48 BAL        4.51        8.1        1.36 CHA        4.72        8.6        1.30 MIN        4.30        7.6        1.28 MIL        4.44        8.6        0.63 SEA        4.40        8.6        0.57 SFN        4.16        8.2        0.35 ARI        4.34        8.9        0.05 FLO        4.43        9.2      -0.03 LAN        4.31        9.0       -0.12 DET        4.67        9.9      -0.13 TOR        4.71       10.0      -0.14 CLE        4.79       10.3      -0.27 HOU        4.21        9.0       -0.32 COL        4.45        9.6      -0.37 KCA        4.25        9.3       -0.53 PHI        4.64       10.3      -0.59 TEX        5.39       12.3      -0.77 CHN        4.51      10.4       -0.94 NYA        5.38      12.4      -0.95 CIN        5.10      11.9      -1.13 SDN        4.30       10.6      -1.67 NYN        4.58       11.4      -1.82 PIT        4.15       10.6       -2.02 OAK        4.85       12.4       -2.16 

 

The "lin VAR" column is that difference. A positive value indicates a consistant offensive team, a negative value is an inconsistant team. As you can see, the Brewers are actually a slightly offensively consistant team. The difference between their average RPG and actual is smaller than expected.

 

More to come...

[And That]

Russ, excellent piece, but would it be OK if I moved it to the Stat Analysis Forum, or did you intend for it to be here?

[rluzinski]

I hoped it could be left here for the time being so that more people would see it. After it loses steam (if there's any steam to begin with) it could be moved to the SABR forum. I tried to write it without all the statistical mumbo-jumbo that plagues some of my writing.

 

Obviously, that choice is up to you guys.

[And That]

The goal of the "SABR" Forum is that people would go there first, but if you want it left here for a few days, it'll be left here for a few days.

[rluzinski]

I was hoping the subject matter would interest the non-statheads as well, hence me writing it with that audience in mind. I just think some statistical topics are general enough to interest a wider audience. I guess we'll see.

[H0FFY]

It should be here, as that other place is foreign to many.

 

Did people say the Brewers were inconsistent and it costs them games? I only say (base on the numbers) we are poor in high-leverage situations. But give me a chance to read this.

 

Thanks, Russ, for taking the time for this number crunching. Are you sure you are not using this all for some SaberDoctorate? Are we just pawns in you life game?

[rluzinski]

Did people say the Brewers were inconsistent and it costs them games?

 

I've heard the Brewers accused of having an inconsistent offense many times this year. Not so much here as in the media in general.

 

Thanks, Russ, for taking the time for this number crunching. Are you sure you are not using this all for some SaberDoctorate?

 

Appreciate it. As sad as it may sound, It's just a hobby.

[brewers19]

Just out of curiosity, is there any reason why you clipped the graph where you did? It's not really all the important, but I just noticed that about 4 teams are not within the set range of RPG.

[rluzinski]

Just out of curiosity, is there any reason why you clipped the graph where you did? It's not really all the important, but I just noticed that about 4 teams are not within the set range of RPG.

 

I meant to clip off just Washington and Boston, for the sake of finding the "expected variance" (they were outliers) but must have accidently clipped off 2 on each side. I have since updated it to include all 30 teams, as well as taking the square root of variance (standard deviation is the result) but it really didn't change much.

 

The Brewers are still slightly more consistent offensively than an average team.

[reillymcshane]

Russ, thanks for doing this - and doing it so it's fairly easy to understand.

[1992casey]

---

The goal of the "SABR" Forum is that people would go there first, but if you want it left here for a few days, it'll be left here for a few days.

---

I was going to post something to that effect on the Issues forum and still might.

 

The Statistical Analysis thread offers the added advantage of material being archived for a longer period, thus letting us reference the material for that longer time.

 

Sometimes stuff comes up over here and turns out to be good material for the Statistical Analysis thread. In those cases, I'd give the threads the larger audience over here for the time being, then move them over there.

 

I'd submit that the Hall of Fame Trial threads would be good material to archive on the Statistical Analysis forum before they're lost.

[pogokat]

i feel like everything should start here, and be archived there...no ones reading it but the sabr junkies as it is, and its not improving discourse on the mlb board...

[reillymcshane]

Quote:

---

i feel like everything should start here, and be archived there...no ones reading it but the sabr junkies as it is, and its not improving discourse on the mlb board...

---

I kind of feel the same way. The stats board is something I don't look at often. I no doubt miss some good things because of it.

[1992casey]

pogokat,

 

How about a pinned thread at the top of the Statistical Analysis forum where we could all 'nominate' threads that could eventually be moved over there?

 

As far as where a thread starts, I guess it would depend on the motivation for starting it. Something like Log5 Equation or Power/speed number are probably best served by starting over there to begin with. Other threads turn into stat discussion as they evolve and can be moved when they start to die.

[rluzinski]

Something like Log5 Equation or Power/speed number are probably best served by starting over there to begin with.

 

Agreed. If it's just a stat discussion, no point in having it here. If guys want to learn more they'll visit it. If it's a thread applying statistics to the Brewers or baseball in general, I see it having a potentially wider audience.

[dixieflatline]

Great work Russ. I only wonder what you could do with a real statistics package instead of Excel

[pogokat]

I like the idea of nominating threads to go over there...and i understand your point rluz...i just don't want people who only read the ml board (geno) to miss out on good stuff...

[endaround]

Actually the idea of whether a consitent offense is good in and of itself it an open question. Say you're pitchins staff has no variance but automatically allows 3 runs a game. If you offense averages 2 runs a game and you have no variance you'll lose every single game. In such a case a highly varible offense is much better. So your pitching staff influences how variable you'd like your offense.

 

The othe thing thats important is the zero bound. The weaker you're offense the more varibility you'd like. If you score only one run a game, the diference between 1 and zero is very small. So it might be better to score 5 runs in one game and zero in four others.

[rluzinski]

Actually the idea of whether a consitent offense is good in and of itself it an open question.

 

I'm starting to think it's not very important at all, actually. I certainly can't gleam any stats that would say otherwise. I've basically come to 2 conclsuions.

 

1. The Brewers offensive consistency is about average this year.

2. The benefit of having a consistant offense is minimal.

 

I'm sarting to wonder if anything considered "common baseball knowledge" is actually true.

[istanbultaye]

I think when people say they want a consistent offense, they don't mean that they want their offense to score the same amount of runs every game. I think they mean they want their offense to score A LOT of runs EVERY game.

[rluzinski]

I think they mean they want their offense to score A LOT of runs EVERY game.

 

I really don't think that's what they mean. The major criticism I've always heard is, "They score 6 runs one days and 2 the next." It's unreasonable to "want" a team to score 6+ runs a game. I give fans more credit than that.

[HOF4canrake]

The real key to any offensive consistency is scoring runs when you should score runs.

 

Any time there is a runner on 3rd and less than two outs, you should knock him in. Obviously, this will not always happen and I don't expect it to--the other team is paid to get key outs--but that is the difference between playoff teams and sub-.500 teams.

 

Yes, if you have a team full of mashers, you don't need to rely on Sac Flies and productive ground outs as much, but the Brewers are not good enough as a whole to miss those opportunities as they have most of the season.

[istanbultaye]

It's unreasonable to "want" a team to score 6+ runs a game. I give fans more credit than that.

 

Actually, it would be unreasonable to 'expect' even the cardinals to score a lot of runs every game. But, i sure as heck 'want' the brewers to score a lot every night. I just know its not gonna happen, hehe.

[rluzinski]

Any time there is a runner on 3rd and less than two outs, you should knock him in.

 

A runner at second and no outs and a runner on third with 1 out score about the same amount of time, 2 out of 3. So one out of 3 times an average team doesn't score in that situation. I have no idea how often the Brewers score in that situation but I bet it's around average. As a brewer fan it can get pretty frustrating to have to hear 33% of the time a fan complain that "good teams don't do that." Sure they do, but only like 28% of the time.

 

To me the definition of a consistent offense is self explanitory; it's an offense that scores around the same amount of runs day in and day out. It's not a team that scores alot of runs, because they would simply be described as a good offense. It's not an offense that that squeeks out extra runs with small ball; that's an efficient offense. What's ironic is that most teams' offenses are simply consistently inconsistent. It's the nature of the game.

[H0FFY]

I have no idea how often the Brewers score in that situation but I bet it's around average.

 

No way, Russ. You cannot rank 14th in the league in OPS with RISP and get those runners in an 'average' amount of the time. NO WAY.