Article: The Twins Binary Hope

[John Bonnes]

You can view the page at http://twinsdaily.com/content.php?1168-The-Twins-Binary-Hope

[jm3319]

I'd rather the Twins build a team that has a chance, or is projected, to win 95 and miss the mark and end up with 90 rather than aim for 86 and end up with 81. No one knows what it will take to win the division, but I'd rather aim for the best team possible rather than building an okay team that might squeak it out in game 163.

[OldManWinter]

Interesting. Thanks.

[chamoman]

Hi John, great idea and legwork looking up the data and running the correlations!! Although I have read Twins Daily since you all set it up, this is the first post that has caused me to actually register in order to post.

 

You seem to be showing what baseball pundits and statisticians alike believe. However, I think the correlations may be limiting because it merely shows, well, correlation, and not causation. And since in this case, you clearly have a testable hypothesis (more regular season wins causes greater postseason success) with a clear causal arrow and no chance of reverse causality (e.g. postseason success causing more regular season wins), wouldn't it make more sense to run a regression? That way you could determine the precise statistical significance as well as the substantive significance of the effect (iff statistically significant, of course).

 

You could begin by running a bivariate regression with the two variables you just used, and then could *really* play around to see what factors increase/decrease postseason success--or if it is indeed a crapshoot. You could add pitching, hitting , and fielding numbers and basically play with the specification ad nauseum. Something interesting it bound to pop up, right? Maybe fielding is more important, or player age, or postseason experience, or whatever. But this is a pretty simple model that savvy front offices must be familiar with.

 

Of course, this may all be for naught, seeing as your correlation coefficient is so small! But with extra time, it might be a fun game.

[Top Gun]

Record doesn't matter once you are in the playoffs. It's a new season. Anything can happen in a short series and usually does.

[glunn]

I would have thought that the correlation between regular season wins and postseason success would have been a lot higher.

 

As a Twins fan, I am used to not going far in the playoffs. I wish that someone would figure out how the Twins could squeak into the playoffs then go to the World Series instead of getting swept by the Yankees in the first round.

[Seth Stohs]

Since the teams play different schedules, I think that the W-L record of division champs and playoff teams matter less and less.

 

The cause of the Twins losses in the playoffs were many, but it had nothing to do with the regular season. Against the Yankees, the Twins lefty-heavy lineups struggled against Sabathia and Pettitte. There were also a couple of bad calls (Mauer fair ball/foul ball), bad bounces (Koskie's ground rule double), and bad pitching (from guys who generally pitched great).

 

To summarize, I'm not at all surprised by the 0.07 number.

[Kwak]

Are all Wins the same? Example: some teams crush the lesser teams but struggle against the better teams. Matchups? Some teams are dominated by one or two outstanding players, while others are far more balanced. The playing field--higher winning teams get to play an extra game on their field--or some teams are configured for their home field, ergo NYY has many LH power hitters. How is that reflected in the analysis?

[drivlikejehu]

This wasn't what you were arguing with Aaron about. His point was just that a team with more wins, in a tougher division, is a better team that one with similar or fewer wins, in a much weaker division. That is obviously correct proposition, so on the podcast you were trying to change the topic a bit into something about whether getting more wins matters (a completely separate issue).

 

Your new topic is bogus for several reasons, but the most obvious is that teams can't predict the regular season outcome with specificity. They don't know exactly how their players will perform, who will get hurt, and how other teams will do. So the dominant strategy is to attempt to build the best team budgetary constraints allow.

[Brock Beauchamp]

This wasn't what you were arguing with Aaron about. His point was just that a team with more wins, in a tougher division, is a better team that one with similar or fewer wins, in a much weaker division. That is obviously correct proposition, so on the podcast you were trying to change the topic a bit into something about whether getting more wins matters (a completely separate issue).

 

Your new topic is bogus for several reasons, but the most obvious is that teams can't predict the regular season outcome with specificity. They don't know exactly how their players will perform, who will get hurt, and how other teams will do. So the dominant strategy is to attempt to build the best team budgetary constraints allow.

 

You're missing the point. No one tries to build an 85 win team and says "Well, that's good enough. My job is done." Every team tries to build a 100 win team (within reason). John's point is that it's unimportant whether you win 85 or 95 games. What's important is whether you make the playoffs or not. Once you're in the playoffs, it's a crapshoot... After all, the winner of the supposed "worst division in baseball" went to the World Series again while the Yankees, Rangers, and A's were sitting at home.

[drivlikejehu]

Well home field matters, but beyond that additional wins don't "matter." No one argued that they did, certainly not Aaron on the podcast. But they still tell you something about how good a team is. It's fair game if you are evaluating front offices, since every team wants to win, and they have varying degrees of success.

[Teddy]

Clarification please. You said:

 

The closer to 1, the more regular season wins translates to playoffs success. For instance, comparing temperatures in Celsius to temperatures in Fahrenheit would have a correlation of 1. Not only does one go up when one goes down, but it goes up or down proportionally the same.

 

Did you mean to say "not only does one go up when the other goes UP..."?

[Riverbrian]

Get into the playoffs... It's that simple... Being the better team in the standings or on paper means nothing and always has. The beginning of each series matchup is a blank slate and small sample size.

 

It doesn't matter if you won 119 games during the regular season. You can still get beat in the first round against a team that won 82 games.

 

It doesn't matter if you swept the first series. You can still get beat by the team that had to eek it out in 16 innings in game 7.

[jharaldson]

Since the teams play different schedules, I think that the W-L record of division champs and playoff teams matter less and less.

 

The cause of the Twins losses in the playoffs were many, but it had nothing to do with the regular season. Against the Yankees, the Twins lefty-heavy lineups struggled against Sabathia and Pettitte. There were also a couple of bad calls (Mauer fair ball/foul ball), bad bounces (Koskie's ground rule double), and bad pitching (from guys who generally pitched great).

 

To summarize, I'm not at all surprised by the 0.07 number.

 

If you take John's point to heart and think that playoff success is binary then the odds that a team would lose 6 straight playoff series is %1.6 (%50^6). That either indicates horrible luck for the Twins or indicates that other factors besides win/loss are involved.

[Boom Boom]

Since the teams play different schedules, I think that the W-L record of division champs and playoff teams matter less and less.

 

The cause of the Twins losses in the playoffs were many, but it had nothing to do with the regular season. Against the Yankees, the Twins lefty-heavy lineups struggled against Sabathia and Pettitte. There were also a couple of bad calls (Mauer fair ball/foul ball), bad bounces (Koskie's ground rule double), and bad pitching (from guys who generally pitched great).

 

To summarize, I'm not at all surprised by the 0.07 number.

 

I think you've hit on something here - the Twins were no more successful during their run against the Yankees in the regular season than they were in the postseason.

 

The Yankees have missed the playoffs once since 1995. So if the Twins (or any other AL team) wants to be a World Series contender, they're going to have to be able to compete with the Yankees. That means that playing for one run and pitching to contact is not a recipe for postseason success.

[Jim H]

I have believed that getting to the playoffs is the goal. I am not surprised at your results. Clearly the "best" regular season team or teams do not seem to have much if any advantage in the postseason. What might be interesting is whether "how" a team is constructed gives any advantage in the postseason.

 

For example, many people have claimed that the Twins need more strikeout pitchers to have more postseason success. Personally, I think having good pitching is important in the post season. Still, having good pitching during the regular season doesn't always translate directly in the postseason. Verlander was great all season and in the postseaon. One bad game helped short circuit Detriot's chances in the World Series, however. This year in the post season, having power pitchers seemed pretty important, till you consider Barry Zito. Last year, St. Louis did well without any real power pitchers in their rotation.

 

I think having good defense in the post season as well as the regular season is important. But, staying with Detriot, they were pretty poor defensively during the regular season. They were better in the post season, at least in terms of making the plays they had too.

 

One other consideration, is a home run hitting team better in the postseason than having more of a high average type team? Does it really matter when Mark Scutaro is one of the top hitters in the post season?

[Craig in MN]

The problem with this is that you are using raw wins.....which completely neglects Gleeman's point. His point is that Win totals don't do a good job of telling what teams are better. A 85 win team in the AL Central is not equivalent to an 85 win team in the AL East or the NL West. An 85 win team that has a ton of injuries at the end of the year is different from an 85 win team that added a lot of talent in trades at the deadline, which is different than an average 85 win team.

 

If wins don't do a good job of rating how good a team really is, why would anyone be surprised that the wins don't do a good job of predicting which team will win in a head-to-head matchup? In some respects, you might be giving Gleeman's argument more strength.

 

That said, I'm surprised that the relationship isn't higher, but without knowing what other factors might be higher than .07, it's impossible to know if that's really a lot or not. If the playoffs are a crapshoot, then every correlation should be zero. If most other correlations you can think of are much smaller than .07, then .07 is a lot. If there are a lot of other factors that you can find higher, then this is surprising. By itself, this doesn't say much to me.

[John Bonnes]

Clarification please. You said:

 

The closer to 1, the more regular season wins translates to playoffs success. For instance, comparing temperatures in Celsius to temperatures in Fahrenheit would have a correlation of 1. Not only does one go up when one goes down, but it goes up or down proportionally the same.

 

Did you mean to say "not only does one go up when the other goes UP..."?

I did. Thanks for the catch. Fixed.

[tmerrickkeller]

Many good points here. A 90-win team in a "weaker" division can still make the WS, but one would have to believe that a team playing in a more competitive division would be a stronger team, having defeated better competition all year to get to 90 wins. It has struck me that two factors seem to elevate playoff teams: 1) teams with pitchers 1-3 with the lowest xFIP during the regular season (as your 4th and 5th starters don't seem to play into playoff wins and losses as much), and 2) teams that hit more home runs per game in the playoffs than they did during the regular season. They often talk about teams winning that weren't huge power teams, but it also seems that those teams elevate their power numbers during the playoffs.

[minn55441]

Another variable is that the regular season is very long. It starts in April and runs through September if not later. Although the same season, you may have a different team in October than you did in April or May. Each of the wins counts the same, but as a group of players, a team, they will progress as a unit or regress as a unit, they won't stay the same. I'm not even talking about additions to the team at the trade deadline. There were several discussions about team chemistry last summer and much of it devoted to the fact that the team with the most talent wins. I consistently argued against that thesis. It helps to have talent, but the best team wins not the team with the best players. Oakland, Baltimore and others are prime examples.

 

Through the season, teams evolve and change. Some change for the better, some change for the worse. I wonder if a better predictor is how many wins a team has after the all star break? Of course first taking into account teams that have things locked up with a week or two to go and then coast into the postseason.

 

Great topic. John

[rambis26]

I wonder if there is a better correlation between having the worst record of the playoff teams and losing in the playoffs

[John Bonnes]

The problem with this is that you are using raw wins.....which completely neglects Gleeman's point. His point is that Win totals don't do a good job of telling what teams are better. A 85 win team in the AL Central is not equivalent to an 85 win team in the AL East or the NL West. An 85 win team that has a ton of injuries at the end of the year is different from an 85 win team that added a lot of talent in trades at the deadline, which is different than an average 85 win team.

 

If wins don't do a good job of rating how good a team really is, why would anyone be surprised that the wins don't do a good job of predicting which team will win in a head-to-head matchup? In some respects, you might be giving Gleeman's argument more strength.

 

That said, I'm surprised that the relationship isn't higher, but without knowing what other factors might be higher than .07, it's impossible to know if that's really a lot or not. If the playoffs are a crapshoot, then every correlation should be zero. If most other correlations you can think of are much smaller than .07, then .07 is a lot. If there are a lot of other factors that you can find higher, then this is surprising. By itself, this doesn't say much to me.

 

First, I should say that this was an investigation of a topic that came out of that debate. It was not meant to settle the debate. During the debate, we argued about whether "better" teams have an advantage in the playoffs. It is a topic we have touched on often, and I think we have waffled on it, most recently during the postseason. I realized afterwards that I don't have any idea what is true regarding that question, or to what extent it is true.

 

Let me be specific about what I showed above: this shows that the number of wins a team has in the regular season is a poor predictor of their chances of winning a championship. I'll be honest - I was dumbstruck by how low this factor is.

 

It might be that there are other predictors, and one of those predictors might have something to do with strength of schedule or divisional strength or starting pitching or whatever. But to conclude that any of them are important, one must define them. Maybe we should judge teams by something like wins times a factor for strength of schedule. Great. Then do so, and show it predicts success in the postseason. But otherwise, I don't see any difference in saying that a team is better because they had more regular season wins than saying a team was better because they had a higher batting average. Both are nice, but both are ultimately unimportant.

 

I'll go a step further - and I think this might be important. This suggests that the best strategy for an organization to win a championship is NOT to push all their chips to the middle of the table and become really, really good (as judged by wins) for a few years. It is to extend their window of opportunity as long as they possibly can and make the postseason as many times as they can.

 

For the record - I think the Rays are following this strategy. I think the Twins do too, though I think it's a side effect of a core belief in giving opportunities to minor leaguers.

[COtwin]

I understand your point John and even agree with the data analysis. But, I think there is something else that others have obliquely alluded to. The Twins made the playoffs for years in a fairly consistent fashion. Teams like the White Soxs and the Tigers had to squeak in or missed all together. But those teams have had far more recent success in the playoffs. I know this proves your point, but maybe we need to be looking at how our team is built. Verlander tanked and maybe helped kill their season, but he also made their postseason. I would always put my money on Verlander over anyone else, because most of the time he wins. The Soxs (in the past) and the Tigers also went all in on offense, even to the detriment of their defense. Cabrera at 3rd had nothing to do with defense, only providing a spot for Fielder. Worked out well. Meanwhile people are talking about how we need some defensive upgrades. Pitching and Offense. Period. I am tired of a successful season being 3 and out in the playoffs. I think Twins fans used to be a bit smug about how our team was so consistent as opposed to the Tigers and Sox. I know I no longer am. Maybe the were right....

[Boom Boom]

I'm not going to crunch this myself right now, but I'd be interested to see the correlation between playoff success and regular season success vs. other playoff qualifying teams.

[Curt]

The champion marathoner doesn't necessarily win the 100 yard dash. Or even advance past the qualifying heats.

[Craig in MN]

I'll go a step further - and I think this might be important. This suggests that the best strategy for an organization to win a championship is NOT to push all their chips to the middle of the table and become really, really good (as judged by wins) for a few years. It is to extend their window of opportunity as long as they possibly can and make the postseason as many times as they can.

 

You are probably right, but I don't see where Regular Season Wins leads us anywhere in the argument. I wonder if can figure it out with simple probability and some thought exercises. Hopefully my math here is right:

 

Let's say the playoffs are a crapshoot and every team (excluding one wildcard) has a 1 in 8 chance of winning. If an average/crapshoot team makes the playoffs (with a a .125 probabilty of winning) five times, they've got a 50/50 chance of winning at least one World Series. What about if it's not a crapshoot? Say you can create a playoff team with 1 in 5 chances of winning the series: .200 probability. They team would have to have about 59% chance of winning each playoff series to get to 1 in 5 odds overall that year. That's a really good team. If that elite team make the playoffs 3 out of 5 years, they have basically the exact same odds....50/50 chance of winning at least one World Series in the same stretch.

 

So, this team can rebuild for 2 years and go all in for 3 years, or just try to maintain being good for 5 years and their odds are almost identical. But how hard is it to create a team that will be 1 in 5 favorites to winning the World Series (after making the playoffs)? How hard is it to just make the playoffs 5 years in a row? Both are very tough. I'd say its tougher to build a team to be the clear WS favorite 3 years in a row than to make the playoffs 5 years in a row.

 

Now that I think about this, these don't really have to be discrete 5 year periods. It could 5 years of crapshoot playoffs in 8 years of the steadier franchise versus 3 years of playoff favorite in 8 years for the boom-or-bust franchise. The odds stay the same. I'd still say it would be easier to just get to the playoffs 5 out of 8 years than to build a clear favorite 3 out of 8 years, but it's hard to say for sure. But that is basically where the odds lead us.

[LaBombo]

You're missing the point. No one tries to build an 85 win team and says "Well, that's good enough. My job is done." Every team tries to build a 100 win team (within reason). John's point is that it's unimportant whether you win 85 or 95 games. What's important is whether you make the playoffs or not. Once you're in the playoffs, it's a crapshoot... After all, the winner of the supposed "worst division in baseball" went to the World Series again while the Yankees, Rangers, and A's were sitting at home.

 

Not buying it. First, it matters a great deal whether you win 85 or 95 games when trying to reach the playoffs, even in the "worst division in baseball".

 

85 wins in the Central Division going back to 2003 (the first season after revenue sharing took on its present form) gets you a division title one time. 90 wins since 2003 would win the division just three times (88,89, and 90 wins). And the supposedly now mythic, nearly unobtainable yet statistically meaningless total of 95 wins or more was the amount it took to win the Central in half of the past 10 seasons. In fact, the average number of wins to capture the AL Central since 2003 stands at just above 93.

85 wins almost certainly doesn't get you the Central.

 

Also, although it has practically no meaning to AL Central teams, it's worth noting that 6 out of the last 11 AL wild card teams since 2003 have won 95 games or more. And no team has won an AL wildcard berth since then with fewer than 91 wins, or would have won the new second wildcard spot with just 85 wins. 85 wins doesn't get you either wildcard.

 

Second, it DOES matter how good a team is in the regular season when looking at John's stated goal of winning the World Series. Since 2003 only ONE team (La Russa's '06 Cards) with 85 wins or fewer has won the World Series. Nobody else did it with fewer than 90. And only two other have even made it to the WS with fewer than 90 regular season wins. 85 wins gives you a one in 20 shot of winning the WS if you have a Hall of Fame manager.

 

In other words, let's say we do win 85 games with a Twins team whose entire projected 2013 rotation at the moment consists of Scott Diamond, and whose present middle infield stands to be yet again one of the worst in the AL.

 

That 85 win team has about a one in 10 shot at winning the Central, and a one in 20 shot at winning the Series. So despite John's interesting statistical mirage, I'm going to paraphrase Moneyball despite whatever additional scorn that might earn me.

 

"Do I care if my 85 win team didn't win the World Series because 85 win teams suck in the postseason, or because 85 win teams don't make the postseason?"

 

"No, you do not."

[Chuck Ruether]

I think even using raw wins, regardless of division or competition, is even a poor way to evaluate how "good" a team is. This year is a prime example, in that Baltimore persisted into the playoffs despite a talent level that many well-respected baseball people were baffled by their continued competitiveness. However, they still won 93 games in a decidedly difficult division and defeated a Texas team that was the class of the AL for the season in a playoff. Often cited as proof that they weren't as good as their record was their run differential, which I feel like is widely accepted as a better indicator of a team's true ability, if such a thing can actually be computed. Would run differential or Pythagorean record establish a higher correlation?

 

Even that, I fear, is going to produce a low correlation, as with so few samples - even at 8 teams a year for 15+ years, I fear the data is incomplete for conclusions - one occurrence like the '06 Cardinal run to a title might alter the data. That team was beset by injury for much of the year, only getting healthy at the end of the season, and was considered, talent-wise, much better than their record. Or the aberration that was the 2001 Seattle Mariners, which saw a 17-win jump in pythagorean record from '00 to '01, despite being no better talent-wise, and likely worse with the departure of a young Alex Rodriguez.

 

In the end, it's probably inaccurate to call the playoffs a true crapshoot, but it's hard to really establish an argument that winning more regular season games indicates talent superiority or portends greater success in the postseason.

[Riverbrian]

My Theory is just as goofy as anybody elses. The goal of every club should be making the playoffs. That's planning for the long haul 162 games to get in. You can't do anything if you don't get in... Ask the talented Texas Rangers and they will agree.

 

Once you make the playoffs... It's not a crapshoot so much... It's game by game... Grind it out... Luck is always a part of baseball but it's inning by innning... Game by Game... Grind it out... It doesn't matter anymore what you did during the regular season. It's Lohse against Zito and it's about making pitches when it matters... Moving runners over... Key Home Runs... Key Hits... Defensive plays... It comes down to the makeup of your players when it's on the line.

 

Look at a regular season. Any regular season... In 2012... The best team in baseball record wise was the Nationals... They LOST 64 games. Some of those losses came in bunches. That's a team who happened to win a mere 17 games over .500.

 

17 games over .500... Just look at that and let it roll around in your thoughts... Nothing about that percentage screams lock. .605 winning percentage... The Best team in baseball is 6 and 4 over ten games. Why would anyone assume that a team with that winning percentage is a lock. That same Washington Nationals Team lost to Colorado 3 games to 4 during the regular season.

 

My theory... If you want to build a team for playoff success... Build a team that believes in itself and one that competes with every fiber and hope you get a few bounces but you better build a team that can reach the playoffs first and always remember that the other team is trying really hard to win as well.

[Willihammer]

Let me be specific about what I showed above: this shows that the number of wins a team has in the regular season is a poor predictor of their chances of winning a championship.

I have to quibble with you here. I know what you're trying to say, but you're assuming regular season wins aren't a factor in getting into the playoffs when in fact they are more important than ever. 4 whole wildcard spots are handed out according to win total. So I think wins are a very good predictor of a teams chances of winning the WS. That was the whole point of adding a 2nd wildcard - to emphasize those marginal wins needed to win the division outright and avoid that first playoff round (and the chance of being eliminated in it). And also to make the end of the year more exciting, include more teams, profit, etc. I guarantee that if you do this exercise again in 15 years (looking at the 16 years of the 4-wildcard playoff format), you will find that regular season wins are a very good indicator of playoff success due to the fact that more regular season wins translates to more division championships which are essentially a first round bye. The play-in is a round that didn't exist during the period you're looking at here, 1996-2012.