Thursday, April 28, 2005

Questionable calls

I didn't get to watch the game yesterday, but I did follow the first 4 innings online and listened to the rest on Radio.

Let me start off by saying that the radio guys were pretty awful. I was listening to XM but it was the Nationals' home broadcast. I just didn't feel they added anything to the game, and they were clearly more old-school baseball guys who loved to throw around old adages than anyone who respected new thinking and stats and the like. And they clearly shied away from criticizing anything Nats related.

Which ties in here to a couple of what I thought were HIGHLY questionable calls made by Frank Robinson yesterday (and a highly questionable quote as stated by the radio guys).

1) Maybe this didn't come from Frank, but I would guess it did. While the steal attempt is usually a bad move, often when you have a guy on first and two outs and a bad count, you try to steal figuring that even if you get out, you are saving the batter who can start fresh next inning with an 0-0 count. However, what were the Nats thinking trying to steal Guzman with Loaiza up in the 7th? You want to make sure Loaiza can bat again in the 8th? It doesn't make any sense. The only even potential argument would be if you planned on pinch hitting for Loaiza anyway in the 8th and you had someone on your bench who could actually hit (which we don't--I brought up at the game the other day that I couldn't remember/think of a single bench weaker than the one we have right now). Anyway, that clearly wasn't in Frank's mind because of...

2) Why did Loaiza pitch the 9th? He had just gotten out of a rough inning where he had over 20 pitches, his pitch count had reached 100, and according to those that were watching, he looked tired. His last outing against the Mets, he had 4 really rough innings where he had to work to get out of jams, and Frank didn't pinch hit for him, letting him pitch the 5th--where he promptly gave up a lead-off homer (no one at the bar I was at knew buck seyz or I would have had an easy win there). Here, Frank leaves him in to pitch the 9th, and he gives up the leadoff homer again. I brought this up to friends a couple of days ago--Frank is very loyal to these old school mentalities of--you let your pitcher go 5 for the chance at the win, you bring your best pitchers in at specific times when you have the lead, not based on the situation, you give your starter a chance to win even if he is tired, you bat a guy like Cristian Guzamn second, etc. That just isn't the way to win ball games and it is going to cost you a few games a year.

3) Some proof of this is a quote the announcers read of Frank's last night. When asked whether he was concerned that Schneider is batting around .200, Frank answered (I'm paraphrasing, but attempting to remember it exactly) "No, he'll go in ups and downs, but it doesn't matter. His job is to catch our pitchers and call ball games. If he gets even a single hit in a game, that's a bonus, but I'm not going to worry if he doesn't." Well, it is true that a catcher catches. But if you are going to have the mentality that your #7 hitter isn't supposed to hit, then you clearly don't care about your #8 hitter, and #9 is the pitcher...not sure we can win too many games if you are willing to have a black hole from a third of your lineup. Especially when the other 6 guys aren't murderer's row.

4) It's funny how the announcers went out of their way not to be critical at certain points. If they meant it, it's even funnier. After the attempted steal with Loaiza up, one of the announcers said "I'm not sure you want to try to steal with your pitcher up and 2 outs. We don't want Loaiza starting off the next inning." The other announcer then countered: "That's what makes Frank Robinson so good. He does things when you least expect them." Let me tell you. When I'm driving, I least expect someone to dive in front of my car. It doesn't make it a smart move.
They also tried to make it seem very exciting that the Nats were going to be in a bunch of scoreless games like yesterday--the old "I'd rather see a 1-0 game than a 12-10 game." I actually tend to agree with that statement some of the time. But unfortunately, if you have a bunch of 0-0 games late in the game, that probably means you are winning half and losing half and the non 0-0 games, you are losing 5-0, making for a pretty bad overall record. We just can't keep hoping to win 2-1 and 3-1 games.

Ok, I had another thought but I lost it, so I'll just post it later if I remember. 11-11, so I can't complain too much anyway!

Final thought: Random digs at Cristian least I'm not the only one.


At 11:52 AM, Blogger DM said...

As I told you on the phone, Dexys, I think Charlie and Dave are pretty good announcers, especially at describing the play of the game. Sure, the banter you describe here is comical, but as long as it does not obscure the goings on, I'm ok with that. When the banter becomes the main attraction we are in trouble.

Speaking of announcers, the gold standard is Vin Scully. If anyone has DirecTV or, treat yourself to a Dodger home game with Vin. In the past they were simulcasting them with the radio, so Vin talked an awful lot for a TV guy, but I realized it is a great study in calling a game. One fun tip: try to catch Vin in saying "umm" or "ahh" -- he doesn't, instead he speaks in complete sentences and paragraphs, as if he is composing an article about the game. Truly remarkable, and I can't stand the Dodgers.

At 12:28 PM, Anonymous John said...

Any manager who thinks that the batting performance of his catcher isn't an issue doesn't understand how the game works and apparently doesn't care to learn. I'm growing increasingly convinced that F-Rob is the weakest link the ballclub.
p.s. Very cool getting the RV box scores, though I'm still trying to figure out exactly what goes into producing the RV stat. Tips on where to look to get that info?

At 4:57 PM, Blogger DM said...


You're welcome on the ERV boxscores. On the sidebar you'll find links re: ERV. The two operating links contain posts from January on this blog that do a fair job of explaining how RV works. I'm working on a more complete guide and an FAQ that will also be linked to from there.

We'd also love to answer any specific questions you have, because they help us improve the system and find errors.

At 5:11 PM, Blogger DM said...

Also, the RV in the boxscores is based on the 2004 Season Expected Run Value matrix as provided on the Baseball Prospectus website (in the susbcription only section). They have data for the 2005 season up now, but I'm not using it because the sample size is too small.

At 6:46 PM, Anonymous John said...

DM, thanks again :) So let me see if I understand some of this correctly... RV looks at the predicated average value of what a player does in a particular situation?

If that's the case, while this may be a good measure of what a player happened to contribute for a particular game, isn't it a poor indicator of actual accomplishment? In a world of truly useful baseball statistics (and boy golly, that's not the typical stats we get), I want to have measurements of a player that assess performance over the things he has control over.

A batter doesn't have control over who is or isn't on base before him, and so I'd rather not give him more credit just because he hit a single during the condition that another runner was on base (e.g. RBI's irk me as a stat). So... do I understand RV correctly in that it gives different credit to batters who may have had the same performance, but did it under different field conditions (conditions that they didn't actually very directly affect)?

Thanks again. I'm a fan of stats, and a fan of baseball, but am fairly new to bringing the two worlds together :)

At 10:12 PM, Blogger DM said...

That's a really important observation about ERV. It asseses the value of past events, to help you understand what happened and why. Others who have studied systems like ERV (such as Win Probability scoring) have generally concluded that they are not very good at helping you predict the future, i.e. tell you what a particular player might do in _different_ circumstances. In fact, the basic counting stats (Hits, Total Bases, Home Runs, etc.) are pretty good for that, as Baseball Prospectus's PECOTA projections show.

To me, the purpose of ERV Scoring is to help you understand the game better as you watch it, not primarily to assess the value of a player in the abstract and predict the future. It seems like a better thing to do when scoring a game or building a boxscore than simply recording the counting stats, because they don't give you much insight into the context in which they happened.

So, I think, to understand what happened in a game, you do have to saddle the players with some context for which they have no control. That is the reality in which they work, and a good measure of their success or failure.

At 2:08 AM, Anonymous John said...

Ok, thanks, that does clear it up. I can definitely see your point that "the purpose of ERV Scoring is to help you understand the game better as you watch it, not primarily to assess the value of a player in the abstract and predict the future."

Although I'm not so sure 'understand' is the quite the right word... maybe better appreciate the typical expected value of what was accomplished by a player in a particular moment/scenario of a game, certainly.

Then again, if I really want to get into situational happenstance, I might as well go nutty for RBIs and whatnot. :)

I suppose what I'm most curious about is what ERV has a limited ability to shed light on: I want to assess the value of a player in the abstract and projected future, looking at the performance that they directly control. It's part an ethos of meritocracy--I like cheering for performannce, not the luck of the situaiton--and part because one of the reasons I'm curious about stats is knowing what players I hope the Nats keep and what players I hope they trade or send back down.

*But* all that said, I can see how ERV can better help me appreciate and (accurately) value what I see go down during a game, and so I really appreciate the info (and your time and clarity in helping explain it).

At 10:02 AM, Blogger Chris Needham said...

Two points... not really rebuttals, just playing devil's advocate, I guess.

I didn't have a huge problem with Loaiza starting the 9th. Like I said when I wrote about the game, it was obvious he was tiring, but maybe you can squeeze another inning out of him.

That being said, I would have had the shortest leash possible -- as soon as one runner got on (or in this case, scored), I'd Yank him.

If Frank had done that, it's probably 1-0 entering the bottom of the inning -- and then, you've got a chance.

RE: The Steal Attempt
It's certainly unconventional, but I like the move, actually. That was at the point in the game where you need to start worrying about scoring runs, and less about clearing the pitcher.

No, Loaiza's not likely to get a RBI hit, but the possibility does exist and Guzman on second dramatically increases those odds.

Guzman's an average basestealer, and Lieberthal's not an especially strong thrower. If he makes it great.

If he doesn't, and Loaiza has to lead off the next inning, it's probable that Frank would have PH for Loaiza there anyway, getting him out of the game, just in time.

RE The Announcers.
I like them, Charlie, especially. They're not especially insightful (and some of the trivia they use has actually been wrong!), but they've got good radio voices and do a good job of describing the action, even though they don't give the score often enough.

Damn, that was too long!

At 10:06 AM, Blogger SuperNoVa said...

I disagree with DM's proposition that ERV can be used solely for historical understanding of games already played. I find that it could be used as a predictive tool as well.

Consider the case of Timo Perez with the White Sox last year. On his pure numbers alone, he had a VORP of -14 or so. But, having anecdotally understood him to have had a lot of high-value hits (he hit nearly .400 with RISP), I recalculated his value to the White Sox from an ERV perspective - and he was only -7 ERV.

So, if you were Billy Beane looking at the White Sox, you might say, well, VORP tells me that replacing Timo Perez (-14) with Jermaine Dye (+20) in Right Field will add 35 runs to the team this year. In reality, based on Timo Perez's actual performance, he could only predict an increase of 28 runs for 2005, rather than 35.

Now, in making the evaluation of Timo Perez for 2005, you wouldn't want to use the ERV, because you couldn't say "well, Timo will only cost me 7 runs." In reality, he will, on average, cost you 14 runs...he just had a lucky performance last year.

So ERV is a little better at understanding where your team was ACTUALLY last year versus purely predictive stats like VORP.

Now, you could also create an ERV based on the predicted value of each element based on the frequency of such situations, which is what I did in the offseason coming up with the ERV number for various pitchers that was based on the weighted-average value of a walk, strikeout and home run.

At 11:14 AM, Blogger DM said...

SuperNoVa, like the good lawyer he is, mischaracterizes my position as being that ERV's "sole" utility is for historical analysis. My point was simply that its focus is historical analysis, and that by itself it doesn't tell you much about future events. However, as SNV ably demonstrates, when you take ERV and compare/contrast with "counting stat" based predictors, you can get information that helps you make good predictions.

Also, John, if you want an "situation-indepedent" RV for batting, you can easily calculate it by multiplying a player's counting stats by the following modifiers:

1B: .261
2B: .423
3B: .617
HR: 1.0
BB: .261
SB: .17
Out (AB-H): -0.18
CS: -0.18

These are derived from the ERV matrix, isolating the value of going from Home to First, First to Second, Second to Third and Third to Home. I simply averaged the three out scenarios, though their might be a more accurate weighted average. For SBs, the value of going from 1st to Second is 0.16 and from 2nd to 3rd is 0.19, so I picked 0.17 as a reasonable average of the two.

One other benefit of ERV is that you can compare the value of fielding and pitching with batting more easily, because everything is described in terms of a common language of runs.

At 6:01 PM, Blogger Sam said...

1. Vin Scully really is all that.
2. I think Charlie's OK -- a lot of cities have a lot worse.
3. How does someone not know "buck seyz"?

At 5:45 PM, Blogger DM said...

Correction: The modifier for Caught Stealing listed above should be -0.5. Getting caught stealing is worse than making an out at the plate.

At 11:46 AM, Blogger Yuda said...

A bit late to the discussion, but the Cubs probably have a worse bench than we do.


Post a Comment

<< Home