30 November 2013

First-Base Defense - A Look at Triple-A


Travis Ishikawa receiving a pickoff, a play I don't examine in this article. Photo courtesy of Norfolk Tides/Elaina Ellis.

During a rain delay, the Norfolk Tides' media relations intern, myself, and the Tides' official scorer, killed time by discussing the divide between the traditional evaluation methods and the new sabermetric methods. While all three appreciated the goals and efforts of the new analysts, I said that I was concerned that the advanced methods provide the illusion of knowledge without a strong foundation. Many of these statistics are proprietary - the data on which they are based are not available for examination. None of these statistics are verifiable - with offensive statistics, we can confirm that they "work" at a team level, but we can't check these statistics at a team level. And the statistics are inconsistent - it's common for a player to be rated an outstanding defender by one measure and a below-average defender by another.

In order for the newer methods to achieve true recognition, we should first focus on solid, easily understandable counts than on all-encompassing, single-number ratings. For example, before we can determine which of the various advanced defensive metrics is best ,we need to know some basic information that we don't really have - such as how many batted balls defensive players actually field and turn into outs.
During the discussion I referenced above, we all agreed that first-base defense is not measured well by traditional statistics. First basemen are credited with a putout on every play on which the batter-runner is retired at first base, although their contribution to the play is often simply catching a throw from another player. Those plays likely tell us relatively little about a first baseman's defense, because there is (presumably) relatively little difference among first basemen in their ability to catch such routine throws. I said that, before I can appreciate the value of the advanced evaluations, I want to know some fundamental information about first basemen - specifically, the number of batted balls first fielded by the first baseman and turned into outs. This article will look at the first basemen's defense in 2013 games played in Norfolk's Harbor Park.

There are four types of batted balls which first basemen turn into outs - ground balls (taken to the base himself, flipped to the pitcher covering, or thrown to another base to retire a runner); fair pop flies; foul pop flies, and line drives. The table below provides the totals of each type of batted balls that first basemen turned into outs in the Norfolk games for which I have data. First, the totals for both the Tides and their opponents:


Games
Innings
Ground Balls
Fair Fly Balls
Foul Fly Balls
Line Drives
Norfolk
71
634
78
6
14
6
Visitors
71
602 2/3
73
15
13
8

Or, converted to a per-nine-innings basis:


Ground Balls
Fair Fly Balls
Foul Fly Balls
Line Drives
Norfolk
1.107
0.099
0.198
0.085
Visitors
1.090
0.224
0.194
0.119


Before I looked closely at the play-by-play results I didn't know this. The first thing that struck me is there probably isn't much of a difference between the Norfolk first basemen and the visiting first basemen in their defense at Harbor Park. The one possibly noteworthy distinction is that the visitors recorded substantially more outs on fair fly balls than the Tides, but given the closeness everywhere else that's probably a random variation, not reflecting a real ability. The second thing that struck me is how few outs are recorded by first basemen by catching batted balls in the air. The Tides first basemen caught a fly or line drive about once every 2.6 games; the visitors, about once every 1.86 games. In contrast, first basemen fielded a ground ball that was turned into an out about once every game.

However, it does seem likely that the typical first baseman in Harbor Park fields about 1.1 ground balls a game (or about 160 in a 144-game AAA season) and catches about 0.2 foul fly balls a game (about 30 in a AAA season.) So we're not talking about a lot of plays.

The data above aggregates bad defensive first basemen and good defensive first basemen. It's certainly possible that good defensive first basemen are significantly better, and bad defensive first basemen significantly worse, than the total. Next we'll look at the individual Tides first basemen, to see if there's a significant difference among them.

Four Tides players played what I consider to be a significant number of first-base innings at Harbor Park The first table shows the total number of plays by each player, the second shows the plays per nine defensive innings:


Innings
Ground Balls
Fair Fly Balls
Foul Fly Balls
Line Drives
198
27
1
3
0
164
17
2
1
2
98
10
1
4
0
77
10
1
3
1



Innings
Ground Balls
Fair Fly Balls
Foul Fly Balls
Line Drives
Russ Canzler
198
1.227
0.045
0.136
0.000
Travis Ishikawa
164
0.933
0.109
0.055
0.109
Alex Liddi
98
0.918
0.092
0.367
0.000
Buck Britton
77
1.167
0.116
0.350
0.117

The other first basemen included Chris RobinsonBrandon WoodSteve Clevenger, and Dan Johnson, none of whom played more than five games at first base.

So, in a full 144-game season, Alex Liddi would have fielded about 132 ground balls and Russ Canzler about 177. Britton and Ishikawa would have fit between these totals.

It's interesting that Travis Ishikawa, who has had a good-glove reputation at first base, ranked below the average in the number of ground balls he fielded and turned into outs. The fly-ball data is too small to really allow us to draw meaningful conclusions. One point that may be worth further exploration is that Travis Ishikawa, who had been trained as a first baseman since his amateur days, caught fewer foul fly balls than did Russ Canzler, Alex Liddi, and Buck Britton, all of whom moved to first base from other infield positions.

This is a first step toward providing a better understanding of first-base defense. The biggest conclusion is that more research is needed. To me, the most interesting results are (1) how few fly balls and line drives are caught by the first baseman; and (2) that first basemen, on average, field one ground ball each game.

Of course, there are plenty of complicating factors - the ground-ball and strikeout tendencies of the pitching staff, the number of left-handed and right-handed batters in the opposing lineups, the size of foul territory, etc. The advanced fielding metrics attempt to account for all these factors. However, I don't think we can evaluate the metrics until we see how they compare with the more elementary components of first-base defense. This is a first attempt to quantify these elementary components.

6 comments:

Tom Fowler said...

Interesting. I'm generally positive on the "new stats," but the disparity between defensive metrics is vexing. If someone is a good (or bad) defender, objective analyses should come to the same conclusions. It seems that C and 1B are the outliers in measuring defense because those players do so many nonstandard things. How good is the C at blocking pitches in the dirt? How well does the 1B scoop out low throws, or come off the bag and do the sweep tag on a wide throw? Metrics have to account for plays like that. The closer we get to total information in baseball, the better our stats will get at measuring all the little things that go into evaluating a player.

Philip said...

I have a question about defining first base defense:
Wouldn't the right fielder have a role in defining how effective the 1B is?
Balls get to RF via ground ball, fly ball, or line drive. the 1B has no control over FBs, but a line drive may only get to RF because the 1B missed it, and 1B would have much more rersponsibility for a grounder that gets through.
Presumably, if a right fielder never fields a ground ball, it is only because the first baseman succeeded in stopping it, and even if the runner is safe on the play, it's only one base and a runner on second is unlikely to score.
so that would indicate range, speed, etc.
Im coming to the party late, so perhaps this has already been discussed, but when determining the quality of a 1B, how much are plays to RF taken into consideration?

Jon Shepherd said...

Phillip...right fielder does not matter with these metrics because they assume an average defense outside of the player. A correlative would be that metrics tend not to consider the OBP in front of a masher. The goal is to find worth in a context neutral envoronment. There is some discussion as to how much context drives results, but no one jas figured out how to express it qualitatively or quantitatively.

Jon Shepherd said...

Same is true for RF... you are responsible for what gets into your field and are compared to what the rest of the right fielder in the league have done with similar plays. Out might be good for you to read a primer on UZR or DRS or find a copy of Lee Panas' book on stat explanations.

Joe Reisel said...

#Tom Fowler - I applaud the motivations and goals of the new statistics. I think that part of the problem is that defense is truly a team effort while offense is truly an individual effort. To use your examples, in order for a first baseman to scoop a low throw, there must be another fielder making a play and then making a low throw. This article is trying to start to get a sense of the scope. Based on what I found here, if some rating proclaims that one first basemen is, say, 30 runs better than replacement, I'd be suspicious.

Anonymous said...

Great stuff. I love this site. My thought is that great first base defense is actually measured more (subjectively) by factors such as decision making (where to throw the scooped ground ball. Second base or run to bag), the type of ground ball scooped ( a routine grounder toward second or a hard shot down the line that saves 2-3 bases), and most of all gathering errant throws (which may turn two base errors into outs.) How to measure?