Photo

O's Scoring Distribution


  • Please log in to reply
31 replies to this topic

#1 BSLSteveBirrer

BSLSteveBirrer

    Soccer Analyst

  • Members
  • PipPipPipPipPip
  • 5,000 posts
  • LocationMS and ID

Posted 16 May 2018 - 03:20 PM

Y'all know that for several years my issue with the makeup of the O's is their all or nothing lineup. My theory being they tend to score runs in bunches or not at all. So I did a little checking. Interesting data.

 

Going back since the start of 2016 through the game today here are the O's run distribution and how they fare.

 

I chose the following breakdowns for the reason listed.

 

0-2 runs scored - expect a team to lose most the time

3-4 runs scored - expect the O's to lose more than they win due to starting pitching

5-6 runs scored - expect the team to win more than they lose (MLB average runs scored is right around 4.5 runs/game)

7+ runs - expect the team to win most of the time

 

Results

0-2 runs scored: O's are 15-94

3-4 runs scored: O's are 38-58

5-6 runs scored: O's are 51-22

7+ runs scored: O's are 70-15

 

So the W-L results are just about what I would expect to see. But I think the runs scored distribution supports my theory. The O's have played 363 games since the start of the 2016 season. In that span they have scored:

 

0-2 runs: 30% of the time

3-4 runs: 26% of the time

5-6 runs: 20% of the time

7+ runs:  23% of the time

 

Almost a quarter of the time they score runs in bunches and almost a third of the time they don't score much at all. Now I have no idea how this distribution would look for other teams? 

 



#2 Mackus

Mackus

    HOF

  • Members
  • PipPipPipPipPip
  • 36,463 posts

Posted 16 May 2018 - 03:49 PM

So this is buried in a long series of articles on a different topic, but this is the plot of the variance of runs scored per game versus runs scored per game for every team from 1981 to 1996.

 

rd5a.jpg

 

http://walksaber.blo...tions-pt-5.html



#3 Mackus

Mackus

    HOF

  • Members
  • PipPipPipPipPip
  • 36,463 posts

Posted 16 May 2018 - 03:54 PM

It makes sense, because the teams that usually lead the league in scoring are going to be the teams that explode for 8, 9, 10+ runs most often.  So they are inherently going to have a high variance because you're mixing in really high run totals in 5 or 10% of their games compared to your low-scoring teams that only explode a couple times a season and instead have a low variance since most games they don't get more then 4-5 runs.

 

On the other hand, Steve is asking a different question, I think.  Given a certain number of runs scored by the team, is it better to have high or low variance?  I can't find the articles now, but I'd read some on this before when the topic has come up, and the answer appears to be that of course it's better to have a lower variance for runs scored to lead to a higher winning percentage.  Similarly, it's better to have a high variance of runs allowed to lead to a higher winning percentage.

 

I haven't seen enough work done on the following question, but my hypothesis would be that while lower variance is helpful given a certain run production, increasing that run production is massively more important than minimizing variance in terms of winning more games. 



#4 CA-ORIOLE

CA-ORIOLE

    HOF

  • Members
  • PipPipPipPipPip
  • 6,647 posts
  • LocationSOCAL

Posted 16 May 2018 - 04:46 PM

Seems to me the issue Steve has is how they score runs so I'm not sure a compprison to teams that score a lot of runs (or a small amoount) is really the answer that Steve is looking for. So, finding a team (or teams) that score a significant/comparable amount of runs more with OBP and/or speed than they do with power/homeruns would be the best comparables. Te Red Sox from last year and the Royals from a few years ago would seem be good contemporary models to compare variances. 



#5 BSLSteveBirrer

BSLSteveBirrer

    Soccer Analyst

  • Members
  • PipPipPipPipPip
  • 5,000 posts
  • LocationMS and ID

Posted 16 May 2018 - 05:19 PM

What I am suggesting is indeed variance. My thinking is that the O's have a very large variance and it leads to more losses than a team that overall scores similar numbers of runs but has less variance. Its not how they score its that they tend to score a bunch or nothing.

 

The O's have had 109 games the last two plus seasons where they have scored 2 or fewer runs and in those games they are -79 in W/L. On the other end they have had 85 games where they scored a bunch and are +55 W/L in those games. So at the extreme scoring ends they are overall -24 in W/L.  And those extremes account for over 50% of their games. That seems like a lot?

 

For the middle two segments (scoring around average) they are 89-80 so +9 in W/L. All this tells me that they do fine when they score a bit, they do great when they score a lot, and they do awful when they don't score. No surprise there. But they have more games where they don't score compared to those where they score a lot. Which tells me that their all or nothing offense is skewed enough that it hurts them overall in the W-L column. 


  • RShack likes this

#6 RShack

RShack

    http://tinyurl.com/fake-news-BS

  • Members
  • PipPipPipPipPip
  • 19,483 posts

Posted 16 May 2018 - 05:33 PM

What I am suggesting is indeed variance. My thinking is that the O's have a very large variance and it leads to more losses than a team that overall scores similar numbers of runs but has less variance. Its not how they score its that they tend to score a bunch or nothing.

 

The O's have had 109 games the last two plus seasons where they have scored 2 or fewer runs and in those games they are -79 in W/L. On the other end they have had 85 games where they scored a bunch and are +55 W/L in those games. So at the extreme scoring ends they are overall -24 in W/L.  And those extremes account for over 50% of their games. That seems like a lot?

 

For the middle two segments (scoring around average) they are 89-80 so +9 in W/L. All this tells me that they do fine when they score a bit, they do great when they score a lot, and they do awful when they don't score. No surprise there. But they have more games where they don't score compared to those where they score a lot. Which tells me that their all or nothing offense is skewed enough that it hurts them overall in the W-L column. 

 

I enjoy seeing you do this.... so,, thanks....

I agree variance is interesting.   

I agree that the sample you have used says the O's variance is hurting, not helping...

 

I'm not sure that this tells us that having a bunch of HR guys vs OBP guys is hurting, not helping...

What do we see when we look at the good O's seasons of late... their HR-vs-OBP approach is not new.

So,, how did they succeed in the seasons in which they succeeded?


 "You say you've lost your faith, but that's not where its at.

  You have no faith to lose, and ya know it" - Bob Dylan


#7 CA-ORIOLE

CA-ORIOLE

    HOF

  • Members
  • PipPipPipPipPip
  • 6,647 posts
  • LocationSOCAL

Posted 16 May 2018 - 05:46 PM

What I am suggesting is indeed variance. My thinking is that the O's have a very large variance and it leads to more losses than a team that overall scores similar numbers of runs but has less variance. Its not how they score its that they tend to score a bunch or nothing.

 

The O's have had 109 games the last two plus seasons where they have scored 2 or fewer runs and in those games they are -79 in W/L. On the other end they have had 85 games where they scored a bunch and are +55 W/L in those games. So at the extreme scoring ends they are overall -24 in W/L.  And those extremes account for over 50% of their games. That seems like a lot?

 

For the middle two segments (scoring around average) they are 89-80 so +9 in W/L. All this tells me that they do fine when they score a bit, they do great when they score a lot, and they do awful when they don't score. No surprise there. But they have more games where they don't score compared to those where they score a lot. Which tells me that their all or nothing offense is skewed enough that it hurts them overall in the W-L column. 

You really can't use W/L like that with isolating/neutralizing other variables (runs allowed/leveraging. run environment etc.). If you don't care how they score (offensive design), then Mackus has answered your question imo. 



#8 BSLSteveBirrer

BSLSteveBirrer

    Soccer Analyst

  • Members
  • PipPipPipPipPip
  • 5,000 posts
  • LocationMS and ID

Posted 16 May 2018 - 07:29 PM

You really can't use W/L like that with isolating/neutralizing other variables (runs allowed/leveraging. run environment etc.). If you don't care how they score (offensive design), then Mackus has answered your question imo. 

Disagree. Thats 363 games of data. That is statistically relevant. Of course those other factors influence the outcome. But I think the data is pretty clear. The O's are a hot and cold team and that does hurt them in the W-L column. Now I suspect that if they had a better OBP and team speed that it would reduce the variance. But that there is no rule or law of physics that says that you can't be a HR hitting team and not have less variance. But for whatever reason the O's seem to take it to the edges.



#9 CA-ORIOLE

CA-ORIOLE

    HOF

  • Members
  • PipPipPipPipPip
  • 6,647 posts
  • LocationSOCAL

Posted 16 May 2018 - 07:39 PM

Disagree. Thats 363 games of data. That is statistically relevant. Of course those other factors influence the outcome. But I think the data is pretty clear. The O's are a hot and cold team and that does hurt them in the W-L column. Now I suspect that if they had a better OBP and team speed that it would reduce the variance. But that there is no rule or law of physics that says that you can't be a HR hitting team and not have less variance. But for whatever reason the O's seem to take it to the edges.

Yeah. I'm really not following. The variable with defense/runs allowed and leveraging etc., are integral to any analysis with regards to winning and losing. If the team had better OBP/speed (which shouldn't matter as you've already dismissed offensive design) it would mean they'd create more runs not necessarily create a better or more stable variance. 

 

If anything, there may be a better argument the offense has been more efficient than inefficient with respect to wining. 



#10 BSLSteveBirrer

BSLSteveBirrer

    Soccer Analyst

  • Members
  • PipPipPipPipPip
  • 5,000 posts
  • LocationMS and ID

Posted 16 May 2018 - 08:34 PM

CA

 

Not sure whats so hard to follow? You do not need to factor in other issues to make a valid comparison on runs scored variance and whether that impacts Ws and Ls.

 

I don't think its any more complex than this. The O's have a .527 winning percentage when scoring 3-6 runs.  They have a .438 winning percentage when scoring 2 or less runs or 7 or more.  That is almost 10% less winning percentage when scoring on the edges compared to the middle.

 

Over this 363 game stretch their overall record is 174-189. Had they had the same winning percentage across those 363 games as they did in the "average scoring window" (for lack of a better term) their record would have been 191-172. Their variance has resulted in more Ls than Ws. Fact.



#11 CA-ORIOLE

CA-ORIOLE

    HOF

  • Members
  • PipPipPipPipPip
  • 6,647 posts
  • LocationSOCAL

Posted 16 May 2018 - 11:30 PM

CA

 

Not sure whats so hard to follow? You do not need to factor in other issues to make a valid comparison on runs scored variance and whether that impacts Ws and Ls.

 

I don't think its any more complex than this. The O's have a .527 winning percentage when scoring 3-6 runs.  They have a .438 winning percentage when scoring 2 or less runs or 7 or more.  That is almost 10% less winning percentage when scoring on the edges compared to the middle.

 

Over this 363 game stretch their overall record is 174-189. Had they had the same winning percentage across those 363 games as they did in the "average scoring window" (for lack of a better term) their record would have been 191-172. Their variance has resulted in more Ls than Ws. Fact.

Your sample here isn't large enough to mean anything. No to mention I'm not sure what it means comparatively. We had one of the worst pitching staffs in baseball history last year and that doesn't mean anything in relation to wins and losses? Not to mention this years sample. Was run efficiency an issue when we blew away pythag in 2012 and were generally pretty excellent at winning 1-2 rung games for years?  We've had the same offense since 2012, so why are you looking back 2+ years? Like I said before, I thin the better argument is that we've been efficient in runs scoring in relation to wins using your methodology over that time frame since 2012. 

 

Given a large enough aggregate you could probably show a reltiship between winning and run variance. I think Mackus's hypothesis is probably right. That said, I doubt it's anything significant, and more importantly, even controllable. Since you've stated  a couple times now that it's not about design (OBP/Speed versus a power designed team like the Orioles), what is the point here? You can control randomness? 

 

The bigger point here is the Orioles offense really isn't and hasn't been all that great (even in the winning years), particularly in relative/environment terms. 



#12 RShack

RShack

    http://tinyurl.com/fake-news-BS

  • Members
  • PipPipPipPipPip
  • 19,483 posts

Posted 17 May 2018 - 03:13 AM

Disagree. Thats 363 games of data. That is statistically relevant. 

 

You can't say that unless you've run some stat procedure that tells you that... just a big-sounding number doesn't tell you that... you need to consider other aspects as well... what degress of freedom are there in light of the role pitching plays, etc...

 

I'm not saying you're wrong... just saying that you haven't shown that you're right....


 "You say you've lost your faith, but that's not where its at.

  You have no faith to lose, and ya know it" - Bob Dylan


#13 RShack

RShack

    http://tinyurl.com/fake-news-BS

  • Members
  • PipPipPipPipPip
  • 19,483 posts

Posted 17 May 2018 - 03:16 AM

Your sample here isn't large enough to mean anything. 

 

You can't say that unless some stat procedure tells you that... you're making the same mistake he made, just in reverse... you can't just claim that, you need to show that it's true...


 "You say you've lost your faith, but that's not where its at.

  You have no faith to lose, and ya know it" - Bob Dylan


#14 Mackus

Mackus

    HOF

  • Members
  • PipPipPipPipPip
  • 36,463 posts

Posted 17 May 2018 - 07:24 AM

If you wanted to study how much of an impact run-scoring variance makes on a team's win percentage, ideally you would have a large subset of teams that score an identical amount of runs and whose other factors (mainly, run prevention) are identical.  So you take that set of identical teams, and compare their winning percentage to their run variance (adjusted for league average...i.e. if the league is scoring 5.5 runs per game the variances will naturally be higher than when the league is scoring 4.5 runs per game).

 

I don't think you're gonna find many teams with identical runs scored and runs allowed.  So the next best thing, IMO, would be to look for teams in the middle of the pack, say 6th-10th in the AL, in runs scored (since you really care about the O's here, and this whole time we've been a middle of the pack offensive team) and middle of the pack in run differential.  You aren't going to have 5 teams ever season that fits that criteria, maybe 2 or 3.  But if you go back several years, 10 or 20 or even 30, then you've got more teams that fit your criteria.  Find each teams runs per game variance (adjusted for league average runs per game) and their winning percentage.  Plot all the points and see if how strong of a correlation there is. 

 

I'm not familiar enough with database queries to generate that set of data, but it could probably be done fairly simply.



#15 CA-ORIOLE

CA-ORIOLE

    HOF

  • Members
  • PipPipPipPipPip
  • 6,647 posts
  • LocationSOCAL

Posted 17 May 2018 - 08:29 AM

With a large enough aggregate I'm sure you can show a relationship between run variation and wins/losses without having to worry too much about the other parameters. What Steve is doing/representing simply isn't adequate or convincing (individual seasons vary widely for a number of reasons) and I'm not sure what he's shown is meaningful, assuming it even was a representative sampling. What Mackus said about stable offensive variation and more unstable defense variation being helpful (especially the latter one) seems to correlate with what I've read in the past. Even then its like insignificant, but OK I'm all for identifying this stuff and winning at the edges. Steve has taken offensive design off the table. So,what's the point here if you can't control it? 



#16 BSLSteveBirrer

BSLSteveBirrer

    Soccer Analyst

  • Members
  • PipPipPipPipPip
  • 5,000 posts
  • LocationMS and ID

Posted 17 May 2018 - 09:12 AM

As an engineer (or has been lol) here's some further thoughts.

 

1. Certainly a longer look would be more valuable. I literally went over their schedule and gathered the data I did by hand. Maybe some time I will go back in and add the data for 2012-2015 seasons. Would be interesting for sure.

 

2. A comparison looking at teams that have scored similar numbers of total runs would  also be interesting to see if they have similar run scoring variance and to see how it has impacted their W and L. Perhaps the O's have been an anomaly the last several seasons?

 

3. Now for CAs last point. Yes if you can't control it then what's the point. I would agree with that. However, I believe that it can be controlled or at least evaluated to determine if changes to approach would yield less variance. It would take much more time and effort than I would ever do but baseball analytical folks certainly have the data and tools available to make a determination. My gut feel is that having better OBP would result in less variance and more overall Ws for the same relative number of total runs scored. I believe that would result in fewer games where you score a ton of runs but more games where you score over 2 runs. All else being equal more games in the "average scoring window" results in more Ws.  In my previous posts I was trying to keep to my base premise that the O's have high variance in run scoring and that it costs them in regards to their total number of runs scored. I wasn't trying to say that it could or could not be controlled or to do things differently just to point out an issue. The only way to make changes to something is to first recognize there is an potential issue. I believe that the data I have used has demonstrated that there is a potential issue. Further evaluation would need to be conducted to determine if anything could be done to change things.



#17 CA-ORIOLE

CA-ORIOLE

    HOF

  • Members
  • PipPipPipPipPip
  • 6,647 posts
  • LocationSOCAL

Posted 17 May 2018 - 10:46 AM

Steve, I've given you a couple chances now to confirm that offensive design is the premise here and you've readily dismissed it. Yet you've referred back to better OBP a couple times now. Where are you going here? Obviously better OBP means more runs. It's the biggest factor by far in scoring runs, but it doesn't necessarily mean better run distribution/variance. Maybe it does, maybe it doesn't. If it does, I agree it's a reasonable angle to explore. Again, what's the direction here? 



#18 CA-ORIOLE

CA-ORIOLE

    HOF

  • Members
  • PipPipPipPipPip
  • 6,647 posts
  • LocationSOCAL

Posted 17 May 2018 - 11:55 AM

Now that I'm thinking about it more, I'm pretty sure this run distribution inefficiency theory was brought up by JTrea years ago during the pre-Duquette era and was a big topic on OH. 



#19 Mackus

Mackus

    HOF

  • Members
  • PipPipPipPipPip
  • 36,463 posts

Posted 17 May 2018 - 12:12 PM

Now that I'm thinking about it more, I'm pretty sure this run distribution inefficiency theory was brought up by JTrea years ago during the pre-Duquette era and was a big topic on OH. 

 

No, you're thinking of the WOW offer to Teixeira made while he was attending a Ravens game with Angelos.



#20 CA-ORIOLE

CA-ORIOLE

    HOF

  • Members
  • PipPipPipPipPip
  • 6,647 posts
  • LocationSOCAL

Posted 17 May 2018 - 12:14 PM

No, you're thinking of the WOW offer to Teixeira made while he was attending a Ravens game with Angelos.

That too.






0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users

Partners