THE BOOK--Playing The Percentages In Baseball

I just wanted to let you know, something seems wrong with the data tracking.  The first table is correct but the second table seems to be reversed (5’s are being counted as 1’s, etc.).

Yes, what marcello said is accurate, there is a small bug there.

Fixed, thanks for pointing it out.

Is there a way to see the evaluations without giving one yourself?  I already submitted an evaluation for a team, but at the time only a couple people had submitted them, I’d like to check it out later after there are more replies.

Would it screw things up if I submitted an evaluation of entirely ‘I dont know’ in order to get to the page of seeing what the collective evaluation is?

Ah, I see it now, click ‘entries’, instead of ‘sumbit evaluations’.

The thing I’ve always been confused about is the idea that we should not consider any defensive stats we have seen. How, after having looked at the UZRs, TZs, of the 2009 Cubs and their stats in prior years, am I supposed to be able to put that out of my mind, and go only by the games I have seen and my visual impressions? And then, unless I started out the season trying to do this, and kept some sort of log of every play a guy made, I’m only going to recall some great plays and some terrible plays from each guy. Sure, I might be able to integrate all of this into an overall impression of defensive competance, but the idea that I will be able to break that down into all of these fielding components doesn’t ring true for me. I might be able to identify players who are great or horrible in a specific category which is fairly easy to see, such as a throwing arm, but that’s about as far as it goes.

Has Tango attempted to evaluate how much of the Scouting Report is ‘contaminated’ by responders like myself, who have looked at the defensive stats, or seen articles on the fielders on all of the baseball sites, everyday of the week?

Personally, as someone who looks at UZR a lot, i don’t find it difficult to ignore it when filling out this survey for my team (the Mets).  Sometimes you just know certain things.  For example, I rated a certain player’s arm accuracy fairly low, but the rest pretty good...and that’s just from my eyes.

But that’s just me, maybe you have more trouble.  Odds are what you think is bias is just your eyes agreeing with UZR.

I have to believe that most people who vote aren’t even UZR-savvy.  Look how long it took them to come down on Junior, and how much they simply ignored the metrics on Jeter, to take two prominent examples.  And Ichiro too.  He’s once again #1.  Indeed it is bUZR that has come around on Ichiro, in contrast to sUZR.  (I have some other circumstantial evidence that STATS is mis-scoring batted ball types when it comes to Ichiro.  I have to do that complete study one day.)

And there are many cases (prior to this year anyway), where UZR wasn’t even available for rookies, and yet the Fans rate those guys in-step with UZR.

The concern however is legitimate and it’s up to me to disspell the notion in a systematic fashion.

800 ballots filled in 24 hours.  Fantastic turnout!

Bravo, Tango, for organizing this annual enterprise.

Cool, thanks!

***

Oh, you can also do sorts.  This for example sorts by arm strength:
http://www.tangotiger.net/scout/index5.php?sortid=14

Tied for worst arm strength is Juan Pierre and Johnny Damon.  At the top are several players, among the Tulo, Furcal, Ankiel and Yadier Molina.  When you combine that with arm accuracy and release, the best throwing arm is Yadier, followed by Scott Rolen.

I also like to look at traded players.  Jack Wilson for example ranks high by both Pirates and M’s fans.  Betancourt ranks low with Ms fans, and passable (though low for a SS) with Royals fans.

Finally, only two CF got a rating of below 3: Vernon Wells and Gary Matthews.  This makes sense, since anyone who can’t field CF would quickly move elsewhere.  Matthews is an outfielder who spends most of his time in CF.  Wells, well, we know his story.

Its interesting to look at guys who have been traded part way through the season.  Betancourt for example is graded horribly by Mariners fans and just grades as bad by Royals fans.  Consistent but not identical.  Would be interesting to look at all the other instances of this.

I wonder if the new fans (Royals in this case) will trend toward the previous ranking as they accumulate more data or if this is an instance where humans (M’s fans) are just irrational.

Or the traded guy happened to perform worse for the first team, which is why he was traded maybe?

I’m super short on DBacks and Marlins votes.  If you’ve got a blog focused on them, please help me out.

Focus this weekend is on Marlins and Astros.  Let’s find those fans.  DBacks fans came through.

***

I’m inspecting junk ballots now.  Sigh.  I was hoping to get no assho!es, but no such luck. At least the rate went down this year.  I’ll get those cleaned up soon enough.

How do you spot the junk?  That’s unfortunate people would do that, but I guess people are just that way.

I can’t tell you how I do it, except for the obvious ones: putting in “1” or “5” for every player in every category.

I love this project. While I do promote it at Driveline Mechanics and plug it elsewhere, I must admit that I don’t fill it out myself, given that 1) I spend most of the year obsessively following the numbers; 2)I don’t get to watch the same players frequently enough; and 3)I wouldn’t trust my judgment, anyway.

I know this is a “public service,” but can I just put in a request? Would it be possible, when the results are complete, to put them out in a .csv file along with one of the universal player IDs so that people could privately use them in our research/databases? Again, I know this is your project, and I have no idea how difficult that would be to do one way or the other. Just a friendly request, since the information gathered is very useful.

Drive: if you click the “SEE ALL PLAYERS” link at the bottom of the main page, you will see that the first column is the MLBAM ID.  And of course, all of the data is there as well.

Isn’t it simple enough to copy/paste into Excel?

Oops! Sorry, I missed that button. Is there an analogous function for the past years? If not, that’s fine, just wondering if I missed it.

Yes, it is easy enough to copy/paste into Excel and go from there. Now, if I can only figure out how to update your “universal player ID” matcher myself…

Thanks.

I will, hopefully this year, launch the full scouting reports for the last 7 years all interconnected.

***

What do you mean about updating the “matcher” yourself?  Are you referring to that file I posted last year that had the 8833 pro baseball players in 2008 with MLBAM IDs, with corresponding STATS/BIS/Lahman/Retro Ids?

Very cool.

Yes, I was talking about the list of 8833… but I guess I was also making fun of myself. Its a great tool, but given how long it takes me to do (seemingly) relatively simple tasks in SQL, it was more a joke than anything else. I suppose I’m just hoping you or someone else will do it and share it (again). It’s unbelievably useful service, butI wouldn’t even know how to start doing it myself.

I hope these aren’t taken as “complaints,” just questions and requests. I realize that time is valuable, especially “unbilled” time. I appreciate all the information and interaction that is provided here and elsewhere for all of us. Thanks.

Sure, no prob.  Glad to do it… otherwise I wouldn’t do it!

***

So far, these are the best fielders according to the Fans:

2 Mauer, Joe

4 Phillips, Brandon

5 Beltre, Adrian
5 Figgins, Chone
5 Inge, Brandon
5 Longoria, Evan
5 Rolen, Scott
5 Zimmerman, Ryan

6 Andrus, Elvis
6 Aybar, Erick
6 Izturis, Cesar
6 Rollins, Jimmy
6 Ryan, Brendan <<-------
6 Tulowitzki, Troy
6 Wilson, Jack

7 Morgan, Nyjer

8 Beltran, Carlos
8 Bourn, Michael
8 Cameron, Mike
8 Gutierrez, Franklin
8 Kemp, Matt
8 Upton, B.J.

9 Suzuki, Ichiro

To me, there’s only one big surprise.  It’s interesting that UZR sees Ryan in a positive light, but this is the first time the Fans were so glowing on him.  In the past, they rated him around an average 2B or SS.

It is possible that Ryan has “learned” his position.

Tango,

I think the story on Ryan is that he is finally getting enough playing time to really show off his abilities. There were never enough consistent chances to rate him in the past. As a Cards fan who watches nearly every inning, I can say that the more consistent playing time has led to more consistent fielding from Ryan. Also, more chances means more opportunities to make the spectacular play as well. Ryan makes me say “WOW!” at least once a game.

I would also say that the Cards organization has stressed fielding and teaches fielding better than most organizations. It is George Kissel’s legacy.

Not sure where else to post this…

Derek Jeter has posted pretty decent UZRs the last 2-3 years, despite being an age where most shortstops are moved off the position and terrible defensive stats for most of his career.

Has he actually improved at this age?  Is he just lucky now?  Just unlucky for many years before that?

He just made an incredible play to rob Maicer Izturis of a single on a ball hit up the middle.  All I can say is that from years of watching Jeter, that ball is EXACTLY the kind of hit that inspired the phrase “Pasta-Diving Jeter”.  But this time he made it.

"Has he actually improved at this age?  Is he just lucky now?  Just unlucky for many years before that?”

Who knows?  One of those things that is almost impossible to know.  Likely a combination of all those things.  What the percentages are (of the combination), who knows?  I suppose that if you spent a few weeks watching video of almost every game from the last 5 years, you might be able to shed some more light on it, but from the data it is impossible to know why players’ numbers do what they do, no matter what the pattern or trend.  All we know is that from the numbers we have an estimate of the player’s current true talent and an estimate of his past true talent at and over any period of time.  Those numbers constantly change.  When a number gets better, we never know how much it means that a player got better or has been lucky now or unlucky in the past.  Again, the most likely explanation, in general, is all of the above. But we really have no idea from the numbers. No idea at all. Did I say that we have no idea?

From the numbers, the most likely explanation for someone like Jeter is that his true talent was better in the past than his numbers suggested, he got better (that may not be part of the most likely explanation - I don’t know), and his true talent is worse in the last couple of years than his numbers suggested.

But, any of the many permutations of luck and true talent are possible.

I suppose that because it is pretty unlikely that a player in his 30’s will actually have his true talent defense improve significantly, there is a point at which it is more likely that a player like Jeter did not actually improve in true talent, but that it is a combination of his true talent being better in the past than we thought and worse in the present (the last couple of years) than we think.

So, if a player in his 30’s looks like this, UZR-wise:

5 years ago: -15
4 years ago: -12
3 years ago: -16
2 years ago: -3
1 year ago: -1

Maybe we can say that it is more likely than not that his true talent defense improved, but if that same player looked like this:

5 years ago: -15
4 years ago: -12
3 years ago: -16
2 years ago: -8
1 year ago: -5

Maybe we can’t (say that his true talent improved).  I have no idea where that line is.  It is a complicated Bayesian statistical problem that involves some numbers that are really hard to know or estimate, such as, “What is the probability that a player in his 30’s will have his defense improve by x,y,z, etc. points in UZR from one time period to the next, and what is the measurement and sample error of UZR for a SS (or other position) for any given number of opportunities, etc.?