Monday, March 28, 2011

Rox Talk - Player X

The "X"-Factor
In this past weekend's NY Times Keeping Score, there was an article on the intangibles.  It is well known that stat heads claim there is no "X"-Factor.  But everyone has tales that Player X took that extra base, worked the count, stole a base, framed a pitch, or just had that "umpf" - that presence that leads a team to victory.  Data below shows last year's team record when Player "X" starts.  Throw out those rare starts and I was a bit surprised to see Smith and Fowler at the top. 


For the Rox to succeed in 2011, the intangibles need to work.  Players like Smith, Iannetta, Fowler, and Stewart are going to have to do the little things.  If the Rox think they can only win with "Os" it is going to be a long year. 

So let's us have a great fun Summer!  Go Rox

Monday, March 21, 2011

Rox Talk - Errors

Errors and Winning Percentage
In defense (ah get it?) of Troy, I was curious about the same thing that he wrote in his column yesterday. What part does defense play in successful teams?  Pretty easy to see that not giving up extra outs allows your pitcher to be more comfortable.  But is it that black and white?  I didn't look at any exotic defensive statistics, just simply looked at errors and winning percentage.  Graph below shows the results from 2006 - 2010.  X-axis is winning percentage and Y-axis is errors.  Correlation of this data (R^2) is around 0.2 indicating that while it makes sense that fewer errors would lead to more wins, the data over the last 5 years indicates that this isn't exactly true.  

Look at the graph and one point kind of jumps out at you and that is the team with a winning percentage below 0.400 and less than 80 errors...that would be the 2009 Pirates who won 62 games.  Now of course the year the Rox went to the World Series they committed only 68 errors.  The average errors committed over the 5 year period is 99.  Teams having more errors then the average won an average of 78 games while the teams committing less than this average won on average 84 games.  So regardless of the scatter plot below, generally, committing less errors does help a team out.  Over those 5 years the teams in the top 10 won on average 86 games and if you throw out the Pirates and Jays, then that average number of wins jumps to 90 which is probably a playoff team.


And what about the Rox specifically.  See the data below.  No real correlation between defensive prowness and wins.


Links
- Ah the continuing battle between seamheads and old school stat lovers...as always Joe says it best...original article here...and I resent the quote about my mother's basement, because I didn't have one!
-
Chris Quick's article is an eyesore for Rox fans and of course makes a pretty good argument that defense does matter...maybe my analysis of errors above is an error?
- Just because I love Ichiro...

- And finally in a tweet by Mike Axisa, he gives us this little gem:
"Greg Maddux faced 20,421 batters in his career. Just 310 of them saw a 3-0 count. That's roughly one every three starts."  Wow!

Monday, March 14, 2011

Rox Talk - Relief Game Score Part II

Relief Game Score - Part II
In Part I of my treatise on a relievers game score, I noticed some possible modifications that could then be integrated with the starters game score to produce a total game score for the pitching staff of that game. The basis of the starters game score is to represent an average start to be around 50 points. Any exceptional start would be in the 80s and an extraordinary start would be in the 90s.

Basing a relief game score on the same 50 point basis would suggest a similar trend. An average relief outting would score around 50 and go up from there. To extrapolate from this it would suggest that an average start and relief appearance would end up to be around 100 points. For instance in 2010, the average relief game score for the Rox was 58 and the average starter game score was 50. But what if we scaled the relief score, so that instead of an average effort of 108 points, you would get around 75 points when you add the relief game score to the starters game score (you know a nice C effort?). So instead of starting the relief game score calculation with 50 points let's start with 15 points. Now the average relief game score for the Rox is 24 for the 2010 season and to go along with the average starter game score of 50 we now have a total game average of 74 points for 162 games.

How does this hold up in the wins and losses? Well the average total game score in wins is 85 and in losses it is 63. Not too shabby. Table below shows the top to bottom total game scores for the 2010 season.



Top Scores:
Game 82: 15 inning 4-3 win over the Giants (Hammel goes 7 innings)
Game 93: 10-0 win over Florida (Francis goes 7 inning, bullpen strikes out 5)
Game 4: 7-0 win over San Diego (de la Rosa goes 7 innings, bullpen strikes out 3)
Game 52: 11 inning 2-1 win over the Giants (Hammel goes 6 2/3, bullpen strikes out 6 and no hits)
Game 113: 6-2 win over Mets (Francis goes 6, bullpen strikes out 6 and no hits)

Worst Score:
Game 133: 11-12 loss to Phillies (bullpen gives up 9 earned runs in 3 2/3 innings)

Conclusion
Game Scores isn't a real statistical tool. It was mostly just a back of the envelope calculation to rate a starting pitcher's effort and to especially look beyond the "quality start". In much the same way the relief game score is just taking some benchmarks that you'd like to see out of your relief corp and assigning some point values. Manipulating these numbers to get a realistic number so that they can be favorably compared. Based on one season of input, the numbers would seem to indicate that you can assign a value to the relief effort much like the starters. While you can similarly rank the relief corp on a similar point scale it would seem to be more of a value to be able to take both the relief and starters score and produce a total game score and have this be on a 0 - 100 point scale.

Monday, March 7, 2011

Rox Talk - Relief Game Score

Reliever's Game Score
If you know my work then you know I have an interest in Bill James' Game Score (Part I, II, and III). I have a belief that it favors the strike out kings and tends to ignore those finasse pitchers. I even came up with an alternative. Well years ago I wrote myself a note and said what about a relief game score? Sure over the years they have gotten the save and the hold but what about scoring the relief just like the starter? Rate the bullpen for that game much like you do for a starter? So let's start with the game score for starters premise:

1.Start with 50 points.
2.Add 1 point for each out recorded, so 3 points for every complete inning pitched.
3.Add 2 points for each inning completed after the 4th.
4.Add 1 point for each strikeout.
5.Subtract 2 points for each hit allowed.
6.Subtract 4 points for each earned run allowed.
7.Subtract 2 points for each unearned run allowed.
8.Subtract 1 point for each walk.

So if the above works for a starter how might it work for a bullpen? Well to me what is important? Strikeouts! Yeah lots of those. Keeping opponents of base? Yeah amazing how that walk in the 8th inning of tight game can really bite you in the...

So here it goes
1.Start with 50 points (I will get back to this later...)
2.Add 1 point for each out recorded.
3.Add 3 points for each strikeout (how relief should be...short, sweet and to the point)
4.Add 3 points for each inning completed allowing zero runs (include partial innings)
5.Subtract 2 points for each baserunner (baserunners bad...)
6.Subtract 6 points for each earned run allowed or inherited runner to score
7.Subtract 2 points for each unearned run allowed (hey defense gotta help too)

So what does it mean? Well using the 2010 rox bullpen I ran the numbers and for the most part I am pretty pleased. I obviously played around with numbers a bit to make an average bullpen outting approximately 50 - 60 points (the blank game scores indicate a complete game from the starter).



Graphically how does it compare to the Starters? Well below graphs the 2010 Rox starting pitcher's game scores (in red) versus the reliever's game scores (in blue). The distribution looks good although maybe the numbers are a bit high. The average starter game score comes in at 51 while the average relief game score comes in at 59. I could probably manipulate the numbers a bit more to get the average closer to 50 but then maybe our relief squad was pretty good last year. I will track the numbers this year and give it a whirl. Maybe next year I will make some adjustments. Next week I will look at starter and relieve game scores as a component score and see if maybe we should adjust the relief game score a bit more...