Originally published at The Tenth Inning Stretch

---

I think that it is pretty safe to say that nobody expects the Minnesota Twins to be close to contending in 2013. As a matter of fact, a lot of fans and experts would find the new Accuscore predictions that were released today and show the 2013 Twins with a predicted record of 72-90, optimistic.

I usually do not try to predict the way the Twins will finish with an analysis. The last time I did it was in the aftermath of the Santana trade and before the 2008 season, when the sky was supposedly falling (you can find that analysis in parts: Part 1, Part 2 and Part 3.) My quite unorthodox calculations predicted the team to have an 89-73 record, while everyone was predicting a sub .500 record. That season, as in this season, the major breaking point is Starting Pitching.

When I was asked before Spring Training by Cardinal70.com in their playing pepper series, my prediction (gut feeling) was that the Twins will compile an 86-76 record and will be competitive in 2013. Additionally, I recently noted the good energy that I saw this team have in Fort Myers, which makes me optimistic for 2013. Add to this, this excellent analysis by Phil Macky at ESPN 1500 based on individual projected performances that project the Twins to win 80 games in 2013, and it was about time to look at the possibilities formally.

I looked at it in 2 different ways and this post is part one, looking at the possibilities of the new starting rotation. The second way will be a WAR- and RAR-based analysis that will be presented some time this weekend or so, otherwise this post will be a monster.

This analysis is pretty unorthodox but pretty close to what I did before the 2008 season. I am looking at the differences of the Twins' rotation in 2013 vs the Twins' rotation in 2012 (Starting pitching is the driver of the Twins' performance folks... They led the AL Central in position player WAR by a lot, 25% ahead of the second team, in 2012.)

I am looking at the following two measures for each starting pitcher in 2012:

• Bill James' Game Score (which is what I used to tabulate my Spring Training Dashboards, and use the same convensions for above average, below average and average as there; explanation is here)
  
• Earned runs a pitcher allowed in each game
  
• Then I tabulate the number games that each pitcher had game scores better than, worse than or average; and the number of games that each pitcher allowed 2 runs or less, 3 to 4 runs and 5 runs or more. (Note: The Twins scored about 4.3 runs a game, so =< 2 runs is high probability or win and 3-4 runs good probability for a win.)
  
• The 2012 Twins' starters used for the calculations are Pavano, Blackburn, Liriano and Marquis
  
• For the 2013 Twins' startes I used the 2012 numbers for Correia and Worley and the 2011 for Pelfrey and Harden
  
• I assume that the rest of the rotation in 2013 (Diamond, De Vries, Deduno etc) will perform as in 2012.
  

Here are the results by pitcher:

It is obvious that the 2013 rotation had many better games in 2012 and would have kept the Twins competitive in most of their games (4 runs or less) than the 2012 rotation. But how much?

• The 2013 rotation had above average game scores 48.3% of the time vs 23% for the 2012 rotation
  
• The 2013 rotation allowed less than 2 runs in 42.7% of games vs 26% for the 2012 rotation
  
• The 2013 rotation allowed less than 2 runs in 82.2% of games vs 61.2% for the 2012 rotation
  

What does this mean in wins for the 2013 Twins?

• In 2012 the Twins starting pitchers won 41 and lost 72 games
  
• Extrapolating based on above average game scores the 2013 rotation will win an additional 23 games for a total of 89 wins
  
• based on less than 2 runs allowed 2013 rotation will win an additional 26 games for a total of 92 wins
  
• based on less than 4 runs allowed 2013 rotation will win an additional 15 games for a total of 81 wins
  
• These 3 average to 87 wins, giving the 2013 Twins a projected 87-79 record.
  

So here you have it. These Twins can compete in 2013.

Next: WAR and RAR-based analysis

Well that would be fun.

Now all they have to do is read this and get on board!

Thanks thry

Thanks for the work. I would be interested in seeing more detailed math. How exactly do you apply those extrapolations, both to the new starters and the returning ones? For example, the new starters had 99 starts for their old teams, and they are replacing Twins who only started 54 games in 2012.

Interesting read, thrylos. Thanks. Obviously, whether it's based on math or based on your sense of the energy at spring training, it's conjecture. I'll place more faith in the observations you and others have made, and in the comments from staff and players than anything else. The distinction I make is between being competitive versus contending. I'm optimistic that, from game to game, we are much more competitive, and your math supports this optimism. But I'm very pessimistic about the chances of us contending. Still, this team should be watchable much more often this year than last. We'll just have to occupy ourselves with a second ntertainment option to fill the void on nights when Correia is on the mound, right?

I'm not very convinced by this. A lot of optimism thinking this team could even come close to 80 wins. The starting pitching just isn't good enough. Diamond is going to regress, Worley will pitch have a dip in number while changing leagues, Harden might not be a starting pitcher anymore, Correia is going to struggle with the league change, Pelfrey might not make it through the season, and Hendriks is still unproven. I hope your optimism because reality, but future looks bleak.

Unlike Badsmerf, I am an incurable optimist. (Last year at this time I was predicting a bounce-back season and a Cy Young for Liriano.) But I'm not convinced by your data, either.

1. I thingk your game score chart is a better indicator than the other part of the chart because it takes into account inniings pitched. Giving up 3 runs in 6 innings does not represent a good chance of winning; it represents a slightly below average chance (4.5 runs per gave vs. 4.3 scored.) I'm pretty sure the starters averaged fewer than 6 IP per start.

2. The guys you chose to represent 2012 only started 50 games last year. Based on game score, I'd guess the record in those starts was 15-39. Even if we replaced those 54 games with a couple of guys from a video game who pitched shutouts every time out, that iimproves us to 90 wins (+24). If they are average (which seems like a stretch), that's +12 wins, or 78.

3. We have three guys coming from the NL to AL, which tends to make numbers worse, although Marquis was dealing with bigger issues at the beginning of last year and I think the blank slate [edit - after he left and went to SD] helped as much as the return to the NL.

4. Harden and Pelfrey are coming off major injuries. I have no idea what to expect from Harden if he ever makes it into the rotation, which is a 50-50 proposition. Pelfrey is less than a year removed from TJ surgery. Maybe he has Adrian Peterson's larm and is completely healed and ready to perform at a level above anything he has done before, but I'm guessing he's an ordinary guy who won't get back to within 90% of his old adequate self before next year.

5. Diamond is hurt. He was a great candidate for regression anyway.

Despite all of that, I'm going to agree with your overall record prediction just on blind faith and/or wishful thinking. My way is more scientific.

Thanks for the work. I would be interested in seeing more detailed math. How exactly do you apply those extrapolations, both to the new starters and the returning ones? For example, the new starters had 99 starts for their old teams, and they are replacing Twins who only started 54 games in 2012.

Good question. And I apologize because I wrote this whole mess up early this morning in an airport lounge down in FL waiting for my flight up North...

I got the percentages (the ones in those bullets) and I did simple math: (percentage 2013 rotation in 2012 / percentage 2012 rotation in 2012) X 2012 rotation wins (i.e. 41). And then added to 66.

There are a few assumptions that actually work the other way: This assumes that the 2013 rotation will pitch as few innings as those guys in 2012 rotation pitched and the rest will be pitched by Diamond/Hendriks/Deduno/Walters/De Vries/Duensing/Swarzak (as in 2012 - which will not happen) So I think that it is a pretty good approximation.

Kid of a teaser for the next installment: If the Twins keep Duensing and Swarzak on the pen exclusively this season, they will gain about 4 pitcher WAR (big deal) and this was not reflected in this analysis...

I'm just happy that Scott Baker didn't resign with the Twins since he is now hurt.(what a shocker).

I got the percentages (the ones in those bullets) and I did simple math: (percentage 2013 rotation in 2012 / percentage 2012 rotation in 2012) X 2012 rotation wins (i.e. 41). And then added to 66.

 

There are a few assumptions that actually work the other way: This assumes that the 2013 rotation will pitch as few innings as those guys in 2012 rotation pitched and the rest will be pitched by Diamond/Hendriks/Deduno/Walters/De Vries/Duensing/Swarzak (as in 2012 - which will not happen) So I think that it is a pretty good approximation.

I'm not sure I follow. If you are assuming the new guys are simply replacing the same 54 starts, there is no way they are adding 15 to 26 team wins. The Twins were 20-34 in those 54 starts last year. Your lowest projection has the Twins winning 15 additional games in that sample, apparently going 35-19 with the new guys, which is a winning percentage even higher than the Detroit's winning percentage in Justin Verlander starts last year. That's not happening.

I think innings and runs allowed with Pythag records might be the best way to do this. Or even just eyeballing WAR -- the four guys last year posted a combined -4.4 WAR in 54 starts. The replacements posted 2 WAR over 99 starts, or 1.1 prorated to 54. That's a 5.5 WAR spread, so it suggests 5-6 additional wins with the new guys. That seems a lot more reasonable (unfortunately, it only takes the team record up to 71 or 72 wins).

I'm not sure I follow. If you are assuming the new guys are simply replacing the same 54 starts' date=' there is no way they are adding 15 to 26 team wins. The Twins were 20-34 in those 54 starts last year. Your lowest projection has the Twins winning 15 additional games in that sample, apparently going 35-19 with the new guys, which is a winning percentage even higher than the Detroit's winning percentage in Justin Verlander starts last year. That's not happening.

 

I think innings and runs allowed with Pythag records might be the best way to do this. Or even just eyeballing WAR -- the four guys last year posted a combined -4.4 WAR in 54 starts. The replacements posted 2 WAR over 99 starts, or 1.1 prorated to 54. That's a 5.5 WAR spread, so it suggests 5-6 additional wins with the new guys. That seems a lot more reasonable (unfortunately, it only takes the team record up to 71 or 72 wins).[/quote']

 

There is more to it. Part of the issue with the old starters is that they just did not pitch enough (because of injuries or because they were ousted) so additional below replacement players made those starts they should had make. If the new guys pitch 30+ games for the 3 of them and 15 for Harden, that is what will make the difference.

 

I did a WAR analysis and I will post it this weekend. In addition to that change of WAR you mention, you have a huge diff or WAR for both Swarzak and Duensing as relievers vs starters, then you take into consideration that Hendriks will probably be at least replacement level this season (instead of -1.2 last season) and things are really looking up....

Sorry, I am still not seeing this. Last year, the Twins had 3 starters with positive WAR (Diamond, DeVries, and Deduno), and they posted a collective 3.5 WAR in 59 starts. That leaves 103 starts and approximately -8.2 WAR (ugh) for our other 2012 starters.

Replace that with the four new guys (2 WAR over 99 starts in 2011/2012), and that's a 10-win improvement. That's a solid improvement, probably optimistic compared to most projection systems, but still less than your lowest projected improvement of 15 wins. (And Duensing and Swarzak aren't making up the remaining 5+ WAR in the bullpen either - probably 2 WAR max there.)

You pretty much have to project significant improvement for Hendricks, a strong debut by Gibson, and big rebounds for Worley and Pelfrey to get to 15. I'm not sure how you'd get to your mean projection of 21 -- even if all 162 starts were started by equivalent pitchers to Diamond 2012 (a pretty crazy optimistic projection), that would still only be a 19 win improvement over 2012.