Wins since ’98 Expansion

Wins since '98 Expansion

Here, we can see the distribution of win totals since the most recent expansion. Notice that the distribution of wins is essentially bimodal, because teams are incentivized to either perform extremely well and make the playoffs, or extremely poorly to secure a good draft pick. As such, teams have responded to the incentives and created a bit of a “black hole” in the diminished marginal utility patch that is the 75-85 win zone. Also note that each bin label corresponds to the right endpoint of the bin, so teams with 81-85 wins would be in the 85 win bin, and teams with 51-55 in the 55 win bin.

Image | Posted on by | Tagged | Leave a comment

A new era of outfield defense? Not so fast!

Earlier, members of the MCC community brought up an assertion; good defenders are nearly ubiquitous in the upper minors. This is a common belief, but is it true?

Let us assume its truth for the moment. It follows that if defense has been undervalued until recently, then these good defenders in the lower minors would find themselves more playing time as the league shifts toward valuing defensive runs the same way as offensive runs. Of course, there are two assumptions here, and neither of them are necessarily correct. Perhaps the league is not shifting, and perhaps there just are not as many good defenders in the upper minors as previously thought, and maybe both.

To clear things up, we move to the data! I looked at all the outfield defensive seasons from qualifiers in the UZR era, 733 in all, and found that there were 118 seasons in which an outfielder was worth more than 7.5 runs defensively, including the positional adjustment. I included the positional adjustment in order to fairly compare corner outfielders and center fielders; the assumption is that the 10 run defensive difference between the corner outfield spots and center is approximately equivalent to the “switching” ability of a seasoned player to play both positions. In other words, a +5 fielder in CF and a +15 fielder in a corner outfield spot would both be right at the cut.

If we are correct in assuming that defensive runs saved were a market inefficiency before recent technological advances, then ubiquity in minor league outfield defense would yield a more densely populated upper-echelon. Yet another assumption is that the rising tide may not raise all ships; offense is still valuable, so only a select few might be good enough defenders and hitters in order to get enough playing time and accrue enough defensive value to make the cut. Here is a table:

Year Count Avg Def/600
2013 9 15.8
2012 9 12.3
2011 6 13.7
2010 9 14.8
2009 12 15.7
2008 11 13.8
2007 6 18.8
2006 14 13.1
2005 12 13
2004 12 14.1
2003 11 13.4
2002 7 13.3

The “count” column is a running total of how many people made the +7.5 defense/600 PAs cut, and the “Avg” column just denotes what the average was of players who made the cut. Here is a chart depicting the same data:

It appears that there is little to no year to year movement in the distribution of upper echelon defenders. Of course, the sample size is rather small. Most years only had around 10 players make the cut, and most statisticians will tell you that is just not nearly enough to estimate a population parameter.

Meanwhile, the major assumptions that (1) not all ships are raised by the tide and (2) the tide is rising also serve to muddy the picture if they prove to be false.

So the result? Inconclusive. However, we do know that good defenders of this nature are extremely valuable. Only 16% of qualifying outfielders met the threshold, and that is already a highly-skilled sample to begin with.

In the end, it might be difficult to prove whether or not good defensive players are constantly available in the minors, but it is notable that there is about as much evidence for it as against it, which, on both hands, is just not a lot.

Posted in Statistical Analysis | Leave a comment

Projecting 2013’s Top 10

The top 10 players in 2013 featured a great variety of players. There were the incredibly young, highly touted, rising stars (Mike Trout, Manny Machado), the breakouts (Josh Donaldson, Matt Carpenter, Carlos Gomez, Chris Davis, Paul Goldschmidt), and the consistent contenders (Andrew McCutchen, Miguel Cabrera, Evan Longoria). Naturally, those are all incredibly arbitrary groups, the fact remains that baseball’s most productive players are not cut from the same mold. As such, going forward, each player will have his own trajectory. Here are the BEANS projections for their 2014s:

Player wRC+ Fielding Positional BSR WAR
Mike Trout 166 10.4 -1.0 9.7 10.3
Andrew McCutchen 146 0.1 2.1 3.6 6.7
Evan Longoria 132 13.2 0.1 -2.1 6.4
Miguel Cabrera 169 -11.6 -2.3 -3.6 5.8
Carlos Gomez 109 18.3 1.9 6.5 5.6
Manny Machado 104 21.2 2.5 -0.1 5.4
Paul Goldschmidt 143 2.5 -12.5 1.5 5.3
Matt Carpenter 130 -3.1 0.2 2.4 5.1
Josh Donaldson 119 7.5 2.1 1.2 5.1
Chris Davis 139 -2.8 -12.2 0.7 4.0

All totals are made with the assumption that each player will get the same amount of Plate Appearances as last year. As you can see, BEANS sees all of these players as very good next year, but there are two standouts. First, Mike Trout leads the pack by a country mile, because he is spectacular. Second, Chris Davis is a full win out in last place, largely because his breakout was so unprecedented, and his previous years do not belie the same type of player as his 2013.

Alas, I cannot see the future, and these are only projections. In any case, the top 10 in 2014 may not be an exact repeat, but you will be seeing much of these names for years to come.

Posted in Uncategorized | Leave a comment

Explaining Terms: WAR

Explaining Terms is a many part series in which I explain certain Sabermetric terms in order to make them more accessible.

WAR, or Wins Above Replacement, is a statistic which is intended to give a reader a holistic view of a player’s performance.

  • WAR is a cumulative counting stat, so higher is better, but be wary of high WAR totals in a large number of PAs.
  • The average WAR per 600 plate appearances is 2.0.
  • WAR is measured in wins, which are meant to be directly applicable to team wins and losses.
  • 1000 WAR is handed out every year by Baseball Reference and Fangraphs, each. Both websites have proprietary techniques to calculate WAR; to differentiate, Baseball Reference WAR is called rWAR and Fangraphs WAR is called fWAR.
  • WAR is league adjusted, context neutral, and park adjusted.
  • For pitchers, fWAR is intended to be fielding independent, while rWAR is intended to be fielding dependent. Both have their detractors, but both are incredibly useful.

WAR is a combination of every commonly quantifiable aspect of a baseball player’s game. WAR for position players combines hitting, fielding, baserunning, a positional adjustment, and the nebulous value over replacement level. For pitchers, WAR is calculated by ascertaining runs saved relative to replacement level.

But what is replacement level? That’s the theoretical point of talent at which players are ubiquitously available at the league minimum of approximately $500,000. In other words, replacement level players can theoretically be attained at any time, so any player worth playing has to be more valuable than that. A team full of replacement level players is assumed to attain a .294 winning percentage, or 47.7 wins in a 162 game season.

Calculating WAR for hitters is rather easy once you have the components. Finding the components might be tricky, but I will explain those in another post.

Batting WAR = (Batting Runs + Fielding Runs + Positional Runs + Baserunning Runs + Replacement Runs) / lgRuns Per Win

lgRuns Per Win is the league average rate of runs that each win above replacement is worth, and it varies slightly from year to year; however it does vary nontrivially over long periods of time. Here is a table with the values. Replacement Runs is assumed to be 20.

WAR for pitchers is substantially more difficult. I may make a post about it in the future, but it is quite a bit of pain for an unsubstantial gain to explain, and my patience with it wanes. I hope you don’t find my rhymes lame.

As always, feel free to ask questions in the comments.

Posted in Explaining Terms | Tagged | 5 Comments

Explaining Terms: wRC+

Explaining Terms is a many part series in which I explain certain Sabermetric terms in order to make them more accessible.

wRC+ stands for Weighted Runs Created plus. It is a statistic that attempts to give a reader a one-shot glance at a player’s hitting ability by boiling down all contribution at the plate into one concise number. 100 is average, and higher is better.

wRC+ is weighted, which means that each outcome (single, double, triple etc.) is weighted by how much it helps a team score runs, on average.

wRC+ is context independent. It is not meant to encapsulate a player’s situational value. Instead, the statistic gives each player the average run production value for each outcome, so the player is not unduly rewarded or punished for the performance of his teammates.

wRC+ is park adjusted. Every ballpark is different, and wRC+ uses Fangraphs’ component park factors to essentially attempt to annul the undue effects of a ballpark on a player’s visible statistics. In doing so, wRC+ becomes directly comparable between teams.

wRC+ is league adjusted. The offensive environment varies year to year, so you can see how Babe Ruth performed in his time against how Barry Bonds performed in his, and you do not need to do the mental math to see which season was better in the context of each player’s peers. wRC+ does it for you.

wRC+ is housed at Fangraphs, and you can find it on every player page.

For the math inclined, wRC+ is calculated thusly:

1. Find each player’s wOBA, Weighted On Base Average. At this point, you can either divide every term by the component park factors, or divide the answer by the total park factor. The component park factor will correct for everything the park changes, whereas the total park factor would only correct for the park as a whole. The former is more valuable to predict future results with a different team, and the latter more useful in determining how useful a player was to the team at hand. Both approaches have their benefits. The equation for 2012 was:

wOBA = (0.691×uBB + 0.722×HBP + 0.884×1B + 1.257×2B + 1.593×3B +2.058×HR) / (AB + BB – IBB + SF + HBP)

uBB is unintentional walks.

2. Convert wOBA to wRAA.

wRAA = ((wOBA – league wOBA) / wOBA scale) x PA

League wOBA is the league average wOBA for that season. wOBA scale is a constant applied to wOBA in order to convert it to “runs”, a recognizable unit. Here you can find the wOBA scale for each season, along with the league average wOBA.

3. Convert wRAA to wRC+.

wRC+ = ((wRAA / PA) / lg RPA + 1) x 100

Lg RPA is the league average production of runs per plate appearance. It’s also available here.

I hope you find this useful. Feel free to ask questions in the comments!

Posted in Explaining Terms | Tagged | 4 Comments

Does Gregor Blanco need a platoon partner?

In short, it depends. Let me demonstrate.

Blanco’s current manifested platoon split over the course of his career is 8.4%. His 428 PAs against lefties regressed against 1000 PAs of a league average 8.6% split, in following the example set in this article, Blanco’s expected platoon split is essentially approximately his manifested platoon split. In other words, Blanco is seemingly a league average platoon lefty. What does that mean for his projected performance on his weaker side?

Blanco is projected by Steamer, Oliver, and BEANS to do the following, per 600 PAs:

System wRC+ Fielding Positional BSR WAR Peri
Steamer  97 7.2 -6.6 0.7 1.9 1.2
Oliver 101 0.6 2.2 3.0 2.7 5.8
BEANS 87 12.0 -2.9 2.7 2.5 11.8

The “Peri” column is just a sum of the Fielding, Positional, and Baserunning columns, and serves to show the peripheral value that a player, in this case Blanco, brings to the table, besides hitting. Note that BEANS has the most aggressive prediction for peripheral value, but the least aggressive hitting projection.

Now that we have a good idea of Blanco’s projections, let’s put it all together. Blanco has shown +4 points of wRC+ above his career average against righties, and -14 against lefties. Since it appears that the regression earlier in the post yielded the conclusion that Blanco’s to-date platoon split is incredibly similar to his projected split going forward, we will just use his manifested career differences. The following table shows his projected batting performance against righties and lefties for each projection system.

System wRC+ v R wRC+ v L Bat v R Bat v L
Steamer 101 83 0.7 -11.2
Oliver 105 87 3.3 -8.6
Beans 91 73 -5.9 -17.8

Here, the Bat columns denote batting runs above average against each handedness.

Now, we can put it all together, and see what type of player each projection system sees him as against both righties and lefties. First, versus righties:

System Bat v R Peri WAR/600
Steamer 0.7 1.2 2.4
Oliver 3.3 5.8 3.1
Beans -5.9 11.8 2.8

It appears that all of the projection systems see him as solidly above average against righties.

System Bat v L Peri WAR/600
Steamer -11.2 1.2 1.1
Oliver -8.6 5.8 1.9
Beans -17.8 11.8 1.5

However, the values across the board seem to point to Blanco being a below average player against lefties. However, he’s not incredibly below average. Being 4 to 5 runs below average per 600 PAs translates only to a loss of about 1.2 to 1.5 runs relative to average, if LHPs make up 30% of PAs. With that said, it may not justify the resource cost to acquire an OF solely to platoon with Blanco. On the other hand, the same could be said about essentially all platoon players; there’s just not a whole lot to be gained on the short end of the platoon. Blanco is just about platoon worthy as any solidly average to above average left handed outfielder would be, all told. However, that is not much.

However, do note that the acquisition of another outfielder is something this author supports, not because Blanco is necessarily in dire need of a platoon-mate, but because the Giants are in dire need of another outfielder. As it stands currently, the top outfielders in the organization are Hunter Pence, Angel Pagan, Blanco, Juan Carlos Perez, and then perhaps players the likes of Brett Pill, Roger Kieschnick, and underperforming prospect Gary Brown. Those are slim pickings, and another OF, perhaps with a favorable platoon split, might be able to pick up the slack of both Brett Pill and Gregor Blanco against lefties. That’s the manner in which the outfield situation should be approached: replace Brett Pill. If that means taking on another lefty, then it might still be worth it, especially if the alternative is a first baseman by trade wandering the spacious expanses of AT&T Park.

Posted in Player Analysis, Projections, Roster Analysis | Tagged , , , , , | 2 Comments

Correcting WAR Totals

Fangraphs unified WAR with Baseball Reference before the 2013 season, ensuring that both websites would only assign 1000 WAR per season. In doing so, Fangraphs cut down the amount of WAR it assigns by 14%, which, among other things, introduces a bias in their long term WAR totals. By December 9th, all totals on this site will reflect totals corrected for this bias; until then, don’t place any million dollar bets on BEANS’ long term monetary value projections.

Posted in Uncategorized | Tagged , | Leave a comment