SB Nation User Blog: PhoenixV

Arch Nemesis in 2009: Regression to the Mean?

PhoenixV — Sun, 18 Jan 2009 03:14:27 -0000

We all know that the Twins hit ridiculously well with runners in scoring position (RISP) last year. Just how clutch were they? Well, their OPS with RISP was .826 (2nd in MLB, after the ballpark-boosted Rangers), far higher than the overall team OPS of .748. Now of course, many analysts have demonstrated that teams on average tend to hit better with RISP than without, but Minnesota's performance last year was of a staggering magnitude.

In this post I discuss two topics:

Is hitting with RISP a "skill", i.e. can we expect the Twins to repeat to some extent?
If not, how "lucky" did the Twins get last year? How big of a hit does our expectations for next year take?

I think the conclusion is quite positive. If you're interested, please read on after the jump.

Is there a problem?

First I loaded the data from ESPN's Team Stats from 2000-2008 into Excel. As a measure of offensive performance I use 2OPS = 1.8 * OBP + SLG, which is a better predictor of run scoring than simple OPS and also corresponds fairly well to wOBA. For each team in each season I then calculated the ratio of 2OPS with RISP to overall 2OPS (which inherently includes the performance with RISP). Let's call this ratio, expressed as a percentage, RISP+.

How much better did the Twins hit with RISP than overall last season? By my measure, their RISP+ was 110.8, so they basically lifted their offensive game by more than 10.8% when they needed it most. I say "more than" because we must remember that the overall 2OPS in the numerator also includes the RISP hitting.

The mean (and median) RISP+ in my data is 103.2, with the 1st quartile at 100.7 and the 3rd quartile at 105.6. As we already knew, teams generally perform better with RISP.

However, it turns out that since 2000 only 7 times has a team posted an RISP+ higher than 110 (the Twinkies were the only ones to do it in 2008). In every single case, team RISP+ dropped at least 5 points in the next season. That's regression toward the mean for ya.

We can paint a clearer picture of this. I grouped RISP+ numbers from 2000 to 2007 together with the corresponding team's change in RISP+ in the following season. The plot below shows the connection between clutchness one year and the change in clutchness the next year:

Now that's a clear relationship. The lesson: If your team posts a high RISP+ one year, it is very likely that they will perform worse in the clutch the next season. All the 15 best RISP+ scores were accompanied by a subsequent drop.

I added a linear regression line because the relationship looks fairly linear. The coefficient of determination comes out at almost 0.5, which roughly can be interpreted as saying that knowing your basics about regression toward the mean can tell you as much about the variation of next year's clutch performance as all other factors you can think of, combined.

So now we know that the Twins are probably due for a setback next year, since posting a high RISP+ doesn't seem to be a repeatable skill. How big will that correction be?

How big is the problem?

I will try to give a catious estimate about how lucky the Twins got last year when there were runners on. By simply plugging the Twins' 110.8 figure into the regression equation shown above we get an expectation of a 6.7 point drop in RISP+ next year, or almost right in line with the average (as the concept of regression toward the mean would suggest). The standard error for the regression coefficient is 0.063, so there's a lot of room for measurement error, but let's just keep things simple here.

This suggests that the Twins' "fair" or "luck neutral" RISP+ last year should have been around 104, which means that if we fix overall 2OPS, the Twins should have had a 2OPS with RISP of around 1.061, whereas their actual 2OPS with RISP was 1.130.

How many runs would that decrease in clutch performance have cost the team last season? That's hard to say. 2OPS is just a scaling of The Hardball Times' Gross Production Average (GPA). In their stat glossary (see the previous link) THT lists a way to convert GPA into runs, so I did that for the two 2OPS figures. By this method, the Twins seem to have gained 28 runs from pure luck. But we need to remember that THT's run conversion formula is calculated based on an "average" setting, and when there are RISP, the setting is decidedly not average. Thus, 28 runs is at best a lower bound, since the runners which are already on base benefit enormously from better hitting results.

To get another estimate, I ran a quick linear regression of the GPA vs. the number of runs scored per plate appearance with RISP (2008 data only, Twins numbers excluded). The R-squared was 0.25, so the relationship isn't all that strong, but let's keep playing. When inserting the Twins 2OPS figure, I get that with their number of PA with RISP they should have scored 48 runs less than they did (of course, their PA would also have declined a bit if they had been less clutch).

Conclusion

So it seems like Lady Luck helped the Twins to around 30-50 extra offensive runs last season, corresponding to about 3-5 wins. The Pythagorean Formula thus puts the Twins' "fair" record somewhere between 84-78 and 86-76 for the regular season (excluding the fateful one-game playoff).

This really doesn't seem bad to me. I had expected the analysis to show that the Twins played waaaay over their heads last year. If Span and the pitching staff don't regress too much overall next year, we might actually still have a decent shot at the division title!

The number juggling above leaves a lot to be desired in terms of error estimates and so on, but I will still venture the assertion that based on core talent, the Twins are one level better than a .500 team right now, which is more than adequate to compete in the AL Central.

Any comments?

Twins 2008 Baserunning Stats

PhoenixV — Wed, 07 Jan 2009 05:29:29 -0000

I had actually planned to spend part of my winter break on determining whether or not Justin Morneau is truly an RBI Machine by parsing through Retrosheet event files and letting other batters "stand in his stead", so to speak, to see how many RBIs they would generate if given the same opportunities as our favorite Canadian. However, this project proved too onerous.

Instead, I used my half-baked program to spit out some 2008 baserunning numbers for the Twins, more specifically how often each offensive player scores on a single or double from 1st, 2nd and 3rd base, respectively. I present the data after the jump – feel free to ignore it if you're not really in the mood for reading a bunch of decimal numbers today.

I focus on the 9 Twins hitters who were on base more than 100 times last season, namely Mauer (252 times on base), Morneau (243), Young (199), Gomez (174), Kubel (153), Harris (151), Span (150), Casilla (134) and Punto (126). Note that these numbers don't include bases reached via the home run or defensive errors.

For reference, and to explain the data format, I first list the American League averages; then the Twins data.

AL Averages

In the 2008 AL, no player failed to score from any base on a triple, so we only need to look at data for singles and doubles. Here are the average scoring percentages for the league:

	1st	2nd	3rd
1B	0.95±0.28 (4517)	56.03±2.56 (1442)	98.69±1.14 (382)
2B	37.78±2.56 (1379)	96.56±1.77 (407)	100.0±0.00 (106)

The way to read this is illustrated by the following example: If you want to know how often runners on 2nd base scored on a single, you go to the cell "1B-2nd", yielding that 56.03% of such runners scored. The number after the plus-minus sign gives a crude estimate of the 95% confidence interval, while the number in the parenthesis is the sample size. Hence, with 95% certainty, we believe that the true probability of an AL runner scoring from 2nd on a single lies in the interval [53.47, 58.59], and we conclude that from 1442 total chances of this kind.

If you want the 67% confidence interval instead, just knock half off the number after the plus-minus sign. Please remember that the approximation is very rough, and some confidence intervals actually come out with negative lower limits. However, it is much better to have the simple uncertainty estimate than no estimate at all.

Twins Data

All right, let's break the data down into singles and doubles.

Singles (1B)

	1st	2nd	3rd
J. Mauer	4.44±6.02 (45)	53.85±19.16 (26)	100.0±0.00 (13)
J. Morneau	0.00±0.00 (27)	56.52±20.26 (23)	100.0±0.00 (18)
D. Young	0.00±0.00 (37)	66.67±17.78 (27)	100.0±0.00 (14)
C. Gomez	0.00±0.00 (20)	80.00±15.68 (25)	100.0±0.00 (12)
J. Kubel	0.00±0.00 (30)	42.86±25.92 (14)	100.0±0.00 (12)
B. Harris	2.86±5.52 (35)	61.11±22.52 (18)	100.0±0.00 (15)
D. Span	3.57±6.87 (28)	41.18±23.40 (17)	100.0±0.00 (12)
A. Casilla	0.00±0.00 (25)	73.33±22.38 (15)	100.0±0.00 (13)
N. Punto	5.26±10.04 (19)	52.94±23.73 (17)	100.0±0.00 (7)

Doubles (2B)

	1st	2nd	3rd
J. Mauer	61.90±20.77 (21)	100.0±0.00 (7)	100.0±0.00 (2)
J. Morneau	18.18±22.79 (11)	100.0±0.00 (7)	100.0±0.00 (6)
D. Young	33.33±30.80 (9)	100.0±0.00 (7)	– (0)
C. Gomez	42.86±36.66 (7)	100.00±0.00 (7)	100.0±0.00 (6)
J. Kubel	20.00±24.79 (10)	100.0±0.00 (3)	100.0±0.00 (5)
B. Harris	60.00±42.94 (5)	100.0±0.00 (5)	100.0±0.00 (1)
D. Span	20.00±35.06 (5)	87.50±22.92 (8)	100.0±0.00 (6)
A. Casilla	28.57±33.47 (7)	100.0±0.00 (5)	100.0±0.00 (3)
N. Punto	28.57±33.47 (7)	100.0±0.00 (5)	100.0±0.00 (3)

Conclusions

So... Those were the numbers. I don't think there's a lot that stands out, but here are a few things that I consider notable:

Gomez and Casilla (and perhaps Delmon) were significantly better than average at going the extra mile from 1st base on singles. This jives well with my personal perception of the speed distribution on the Twins roster.
Generally, the Twins did a good job of not getting gunned down when trying to score from 2nd base on a double. Of all the players examined above, only Span had a failed attempt of this kind.
Mauer was astoundingly good last year at scoring from 1st on doubles. Maybe Morneau's doubles are of a special quality?
The only below-average baserunners on this team (judging from my data) seem to be Kubel and, oddly enough, Span. You can maybe toss Morneau in there as well. It's possible that Denard simply got unlucky last year, but I'll make sure to revisit these numbers next offseason to check up on him.
It's an open question whether the data reflects actual player ability or simply Scott Ulger's trust in his guys.

Your thoughts?

(Image courtesy of Alexander Ross and TwinsGeek.)

They Grow Up So Fast: Carlos Gomez

PhoenixV — Tue, 09 Sep 2008 01:46:42 -0000

Carlos Gomez kills me. With reckless abandon he has been hacking his way through plate appearances ever since Opening Day. If he somewhere down the line turns into a decent hitter, I will give the Twins' coaching staff full credit, because he simply doesn't seem to be able to adapt to anything by himself.

Yet, a couple of days ago someone on this site (I forgot who it was – sorry) suggested that Gomez might have altered his approach recently and actually patched together some good at bats along the way. This got me thinking a bit, and I decided to investigate the surmise.

To evaluate Gomez' "maturity" at the plate, I look at two stats: pitches per plate appearance (P/PA) and the percentage of plate appearances in which he swings at the first pitch (1PS%). This is obviously a gross simplificiation, but bear with me here. It's impossible to find (free) monthly splits for these numbers anywhere on the web, so I created a so-called spider which downloaded the necessary data from MLB.com's 2008 Gameday data. I then extracted the parts I needed for my research and the final output is a couple of interesting (albeit ugly) graphs.

I think you'll find my findings about Gomez' maturation encouraging, so please take a minute to review them.

Data and graphs

Broad overview

First I broke down Gomez' 2008 season into bins of 10 games (in which he came to bat at least once), meaning I lumped his 1st through 10th game together, then his 11th through 20th game, and so on. This gives us some more meaningful sample sizes to look at.

Here is the evolution of Gomez' "maturity stats":

Games	1-10	11-20	21-30	31-40	41-50	51-60	61-70	71-80	81-90	91-100	101-110	111-120	121-127
PA	46	45	42	39	53	47	48	47	45	32	33	36	23
P/PA	3.48	3.64	3.24	3.64	3.42	3.06	3.33	3.30	3.07	3.09	3.61	3.69	3.91
1PS%	43.5%	48.9%	42.9%	35.9%	43.4%	46.8%	41.7%	51.1%	51.1%	43.8%	45.5%	41.7%	39.1%

I created two graphs to help illustrate the development:

Detailed breakdown

Next I decided to trade a more fine-grained view of Gomez' season for smaller sample sizes by narrowing the bins down to 5 games each. Please note that this results in only 21 PA per bin on average, so you may choose to disregard the next graphs completely if you think the samples are too insubstantial. Anyway, here are the same graphs as above, but now with 5 game bins and an added rolling average (unweighted average, unfortunately):

Conclusions

There are a number of interesting insights that can be gleaned from these graphs (and for that I'm very thankful since the data collection took five hours). Most notably, we get our suspicions confirmed that Gomez was indeed pressing and having worse-than-usual at bats during his horrible slump in June and July (around games 50-90). He was seeing approximately 15% fewer pitches per PA than his average line and swinging at the 1st pitch more than 50% of the time.

Since his demotion to the 9th spot in the batting order (somewhere around game 90), however, both maturity indicators have trended dramatically in the right direction. His P/PA during the last couple of weeks has actually been almost at league average and he's done a much better job at laying off the first offering once in a while. Furthermore, the positive development hasn't seemed to slow down, so maybe there's even more to come.

The question now is whether these improvements stem from the fact that he is playing less regularly and therefore is more rested, or because he doesn't feel the same pressure in the 9th hole that he does in the lead-off spot. Or maybe he has actually adjusted his approach and listened to some good ol' reason.

What do you think? Please tell me if there's anything you think needs to be done to improve the reliability of the data.

Note: There seems to be small inconsistencies in my data. I get 3 more PA for Gomez than Baseball Reference has recorded.

Poll

Has Carlos Gomez altered his hitting approach for the better recently?

Yes

40 votes

No, but he will learn soon

19 votes

No, he's a lost case

19 votes

Undecided

18 votes

96 votes | Poll has closed

The Stats: Stretching Nathan Out - Part II

PhoenixV — Sat, 09 Aug 2008 20:03:37 -0000

A couple of days ago I made a half-assed attempt at evaluating Nathan's abilities with regards to pitching in the eighth and pitching more than one inning. As both BeefMaster and Johnny Safron correctly pointed out (I like to refer to this as the "Saffron Beef Critique"), simply grouping Nathan's performances by the length of his outings gives false impressions. For example, a (rare) appearance in which he gives up a walk-off HR after having gotten just one out in the 9th ought to be considered as the same type of outing as his trademark 1-2-3 9th innings.

Thus, I decided to run the numbers one more time, but this time I parsed the Retrosheet event files from 2004 to 2007 (Nathan's four full seasons as a Twin) in order to break down his pitching line by the situation in which he entered the game (the entry point).

I use the following abbreviations:

#P: Number of pitches thrown (may only be 98% correct I guess)
SO%: Strikeouts per batter faced
BB%: Walks per batter faced
#P/BFP: Pitches per batter faced

You will notice that I do not list Nathan's ERA below. In order to calculate that, I would not only have to look at Nathan's own pitching events but also the events for preceding and subsequent pitchers (due to the twisted nature of the earned run), and the mere thought of this programming exercise made my brain melt a little bit.

Now, without further ado, here are the...

Stats

Basic breakdown

Entry point	G	IP	BFP	#P	oppAVG	oppOBP	oppSLG	SO%	BB%	#P/BFP
Somewhere in the 8th inning	14	11.0	75	284	.203	.297	.234	26.7%	10.7%	3.79
Somewhere in the 9th inning	224	133.2	874	3346	.186	.239	.290	32.4%	6.2%	3.83
Extra innings	36	19.0	156	612	.168	.265	.198	33.3%	11.5%	3.92

Fine-grained breakdown

Entry point	G	IP	BFP	#P	oppAVG	oppOBP	oppSLG	SO%	BB%	#P/BFP
8th inning, 0 outs	5	4.0	27	98	.320	.333	.360	22.2%	0.0%	3.63
8th inning, 1 or 2 outs	9	7.0	48	186	.128	.271	.154	29.2%	16.7%	3.88
9th inning, 0 outs	215	131.0	854	3269	.184	.238	.288	32.7%	6.2%	3.83
9th inning, 1 or 2 outs	9	2.2	20	77	.263	.300	.368	20.0%	5.0%	3.85
Extra innings	36	19.0	156	612	.168	.265	.198	33.3%	11.5%	3.92

Conclusions

I guess the first conclusion is that when you become too fine-grained, you get down to ridiculously small sample sizes so most of the last table above must be thrown out the window.

Nathan is an absolute beast when entering the 9th with 0 outs. Look at that SO/BB ratio. It sends shivers down your spine. It is clear that Nathan is better at pitching in the usual 9th inning save situations - but not better by much. He tends to lose control a bit in extra innings, possibly due to nerves, but the resulting batting line doesn't really reflect this.

Also, I think that the almost constant value in the #P/BFP column shows that Nathan doesn't alter his approach much in terms of "nibbling" with the strikezone.

There. I guess we didn't learn a whole lot. We knew that he was good. We all have a feeling that he will be good in the 8th inning as well.

Anyway, I promise that it will take a while before I start spamming the forum with stats once again. ;)

The Stats: Stretching Nathan Out

PhoenixV — Wed, 06 Aug 2008 19:59:37 -0000

I thought it would be interesting to check out Nathan's stats broken down by number of IP, so here they are (courtesy of Retrosheet and a little tinkering with R). I've lumped all Nathan's four full seasons (2004-2007) as a Twin together in the following.

In his first four seasons with the Twins, Nathan has never pitched more than 2.0 innings and he has never had an appearance in which he failed to record an out.

Note: "SO%" and "BB%" mean strikeouts per batter faced and walks per batter faced, respectively.

Update 08/07/2008 04:52 ET: Added ERA column.

Stats

Overall pitching line as a Twin

G	IP	BFP	ERA	oppAVG	oppOBP	oppSLG	SO%	BB%
274	282.1	1104	1.94	.181	.242	.269	32.2%	7.2%

When pitching less than one inning

G	IP	BFP	ERA	oppAVG	oppOBP	oppSLG	SO%	BB%
17	9.0	52	13.00	.395	.500	.605	17.3%	19.2%

When pitching precisely one inning

G	IP	BFP	ERA	oppAVG	oppOBP	oppSLG	SO%	BB%
236	236.0	907	1.56	.172	.228	.252	32.6%	6.4%

When pitching more than one inning

G	IP	BFP	ERA	oppAVG	oppOBP	oppSLG	SO%	BB%
21	37.1	145	1.69	.169	.236	.269	34.5%	8.3%

Conclusions

First of all: Gardy has experimented very little with Nathan's role. 86% of his appearances as a Twin have been of the 3-outs variety.

Obviously, his numbers when pitching less than one inning are pretty horrible due to the fact that those appearances are mostly ones in which he has struggled (we can see that a sudden lack of control has probably been the main culprit in these situations).

The real lesson learned here is that Nathan pitches almost equally well when being "stretched out" as he does when he simply goes one inning. When pitching one inning Nathan has faced an average of 3.8 batters; that average is 6.9 when pitching more than one inning. Thus, it seems that Nathan can actually handle double his usual workload without experiencing a drop in performance. Nathan is on pace to log only 66.1 IP this season, which is about 4.2 IP less than his average totals from 2004-2007.

Now, the sample size for the "more than one inning" numbers is pretty small, so this is not a definitive result. A more thorough study would also look at Nathan's performance when entering the game with runners on base, and maybe adjust the numbers according to the leverage index.

But I'm still pretty certain that Gardy's decision to let Nathan handle longer outings in the future is a very good one - and not only because the rest of the Twins bullpen consists of little schoolgirls.

via www.ohmvfiredept.org

The Run Value of Swapping Liriano and Livan

PhoenixV — Thu, 31 Jul 2008 08:29:52 -0000

Pretty much everybody around here seems to want to ditch Livan Hernandez in favor of Francisco Liriano, the Second Coming of Johan "Should've Won 3 Cy Young Awards By Now" Santana. So, much like I did with the question of batting Gomez 9th, I decided to grab some perspective by calculating the added run value of simply releasing Hernandez and calling up Liriano to take his place.

My question is this: How many more runs will the Twins prevent the opposition from scoring if Liriano gets Livan's remaining 11 starts?

Creating projections

First we need to gauge each player's true talent level. This is easy to do for Hernandez; I just plug his stats and birth date into The Hardball Times' in-season Marcel projection spreadsheet. This yields that Livan should be good for a 4.74 (fielding independent) ERA over the remainder of the season, meaning that his aggregate ERA for the 2008 season would end up at 5.11.

It's a bit more difficult for Liriano, since I want to base my analysis on his excellent work in Rochester this year as well. It's not right to just plug in his minor league totals, since the level of competition obviously doesn't match what he would face in MLB, and furthermore, Marcel bases its projections on major league averages.

Thus, I went to Baseball Prospectus to look up Liriano's minor leagues stats translated to a major league context. This transforms the Franchise's 3.10 International League DERA (defense independent ERA) into a 4.09 major league DERA. I typed the relevant translated statistics into the Marcel pitcher projection spreadsheet as a separate 2008 row, and voilà: We get a 3.63 true talent ERA for Liriano.

The big (?) difference

Having done the above, it's easy to see how many runs Liriano would save more than Livan. Over 9 innings, Liriano is projected to allow 1.11 runs less than Hernandez, which is 0.123 runs per IP. The Twins have 55 games left to play in the 2008 season, so that's probably around 11 remaining starts per spot in the starting rotation. Hence, if the Twins swapped Francisco and Livan right now, it would probably affect approximately 66 remaining IP (6 innings a start seems about right to me).

Multiplication shows that Liriano is then worth 8.14 runs more than Livan from this point of the season and out.

Conclusion

As in the case with Gomez, the added run value is quite a bit smaller than I expected. 8 runs is equivalent to a little less than one win in the standings. Now, one might argue that Liriano's projections are a bit conservative, since he has pitched way better as of late, and we should probably assign more weight to these stats than to the stats from the beginning of the season. However, I don't think this would change the result much.

Still, the Twins need to do this move. If they can trade Hernandez before the deadline, they would only gain from that on the major league level and maybe receive some fringe prospect as icing on the cake. It would also help Liriano's development tremendeously, as he is obviously getting tired of fanning AAA batters.

Whatever "veteran presence" Hernandez provides, it can't possibly outweigh the positives of having Francisco take his spot in the rotation.

That's my two cents - what are your thoughts?

The Run Value of Batting Gomez 9th

PhoenixV — Mon, 21 Jul 2008 10:39:41 -0000

I have been campaigning for Carlos Gomez to bat 9th (or be demoted to AAA) for some time now. To get a more objective look at the benefits of such a move I decided to do some number crunching. I want to stress that I've completely disregarded the mental aspects of baseball in the following - I'm exclusively working with the stats here.

My question is this: How many extra runs per game will the Twins score if Gomez gets to bat 9th from this point of the season and out, while Denard Span is moved to the lead-off spot?

The method

In order to arrive at an answer I go straight to The Book. In this, the authors calculate the average number of runs scored for each possible outcome of a plate appearance. They really put a lot of care into adjusting for a variety of things, including the typical situation in which each outcome is realized as well as the ball-strike count. The results are displayed in a table on p. 28 (e.g. a HR is worth 1.397 runs on average, a single is worth 0.475 runs, a caught stealing is worth -0.467 runs above what an average team would score in the inning, etc.).

The extreme case

Now, to first consider the extreme case, I focused solely on Gomez' batting and base-stealing in June and July. I looked up his splits on MLB.com and typed them into a spreadsheet where I multiplied each stat with its run value above average, as mentioned above. The bottom line was that Gomez has cost the Twins 19 offensive runs compared to the average in 194 June/July plate appearances, or 0.100 runs per PA.

Doing the same thing for Span, I found that since his call-up he has added 1 run above average in 97 PA, which is 0.013 runs per PA.

The Book also lists the average number of plate appearances per game for each lineup slot (p. 122). In AL parks, the lead-off hitter gets 4.83 PA per game, while the no. 9 hitter gets 3.90 on average. Thus, moving Gomez down to the 9th slot would give him approximately 0.93 less PA per game.

Since his performance in June and July indicates that he costs the Twins 0.1 runs per PA, moving Gomez to the bottom of the batting order would thus yield 0.093 runs extra per game for the Twins. Similarly, moving the slightly-above-average Span to the lead-off spot would add 0.012 runs per game.

Thus, the total added run value of switching Gomez and Span in the batting order is 0.105 runs per game.

via cache.lifehacker.com

The projected case

Now, obviously, no-one truly believes that Gomez will be as horrible the rest of the season as he has been in June and July. And Span will probably regress a bit towards the mean (he seems to already be doing so). Consequently, I used The Hardball Times' in-season Marcel projection spreadsheet to generate a more reasonable batting and base-stealing expectation for both players.

I went through the exact same steps as above, and the results were as follows: Gomez is projected to cost the Twins 0.038 runs per PA compared to the average while Span is projected to be almost exactly average. This means that a batting order swap would yield an extra 0.031 runs per game for the Twins.

In conclusion

The added run value was much smaller than I expected, even in the extreme case. Batting Gomez 9th would just barely amount to half a win extra over the remainder of the season. Of course there are many other ways to change the batting order, like moving Casilla and Mauer up a spot. This would most likely be worth more in terms of run value added, but I consider it unlikely that Gardy would consider this option.

I still think Gomez should be moved to the bottom of the batting order - if nothing else it might help restore his mental balance to have some pressure taken off his shoulders. What are your thoughts?

Note: The stats considered in the above analysis were 1B, 2B, 3B, HR, HBP, unintentional BB, SH, SO, SB, CS and outs on batted balls.

Poll

Does the above analysis warrant switching Carlos Gomez and Denard Span in the Twins' batting order?

Definitely - let's do it!

39 votes

No - but I think it needs to be done for other reasons

28 votes

No - the added run value isn't significant enough

6 votes

No - the mental aspect can't be ignored

5 votes

No - the employed method is flawed

6 votes

Undecided

6 votes

90 votes | Poll has closed