The Unwritten Book is Finally Written!
An in-depth analysis of: The sacrifice bunt, batter/pitcher matchups, the intentional base on balls, optimizing a batting lineup, hot and cold streaks, clutch performance, platooning strategies, and much more. Read Excerpts & Customer Reviews
A call to arms for these teams:
Arizona Diamondbacks
Chicago White Sox
Cleveland Indians
Florida Marlins
Houston Astros
Oakland Athletics
Pittsburgh Pirates
Washington Nationals
Overall, after one week, we’re at 47% of reaching last year’s totals. Brewers, Royals, and Giants fans have already submitted more ballots this year than last year.
UZR liked him for 31 runs above replacement. ... Still, UZR wasn’t alone; DRS said he was even better, at 32 runs saved, and Total Zone liked him for 27 runs. It was truly an incredible year for Guti.... Finally, the glove hasn’t played like it did last year. UZR still thinks he’s been good for 7.5 runs, while DRS is even more bullish with 16 runs…
It’s VERY POSSIBLE that Guti’s glove DID play exactly last year as it is this year. That’s because UZR has an uncertainty in classifying the degree of difficulty of a play.
Then, there’s simply random variation. For example, Pujols can be a true .440 wOBA hitter, and he swings and approached each PA exactly the same, and in the first 600 PA he has a .420 wOBA and in his next 600 PA he has a .460 wOBA, and this does NOT mean that Pujols played better. He played the same, and good/bad luck explains the difference.
I’m not saying that this is what happened in Guti case, or any player’s case. But, it’s important to understand that seeing a UZR number is not like seeing a wOBA number. wOBA has one uncertainty (random variation), while UZR has a second uncertainty (classification of batted balls).
***
You also hear about how a park “played” like a hitter’s park, even though it is a pitcher’s park. If players at PETCO for example generated more runs there than away, that doesn’t mean PETCO is now a hitter’s park. It simply means that random variation reared its head (you flipped 10 straight heads). It happens.
Good article by Jason. In the media v media|blogger, it’s like a bear choosing between Usain Bolt and Ernie Lombardo: the media will always look for the easier target.
Magic Johnson preferred to play against Larry Bird, because that’s how greatness is defined.
One thing I love about Federer (and Gretzky) is how utterly non-ashholey they are. Just a couple of guys who happen to know they are the best, possibly ever, at what they do. They don’t get involved in pettiness or minutiae. They’ve got their act together. Here’s Federer’s between the legs shot. He gives a self-pump for pride, and then deflects it off as luck or undeserving of adulation. As Andy Roddick once said: “I wish I could hate you, but you are such a nice guy.” (There’s also the previous between-the-legs shot at the 1:50 mark).
From 2007 to 2009, Howard not only has the highest Win Probability Added in September/October, but he dwarfs the second-rated player on the list--Cardinals first baseman Albert Pujols. It’s not even close. That’s true, even if you stretch the data back to 2006.
Sabrmetrician Bill James awarded Howard his “Clutch Player of the Year” award in 2009. James knows of what he speaks.
Tom Tango, the same person who devised the Win Probability Added formula, also has a formula for “Clutch Rating.” In figuring a player’s clutch rating, you look foremost at situations that are high-leverage, in which the game is on the line in that turn.
A report released this month by several education researchers warned that the value-added methodology can be unreliable. “If these teachers were measured in a different year, or a different model were used, the rankings might bounce around quite a bit,” said Edward Haertel, a Stanford professor who was a co-author of the report. “People are going to treat these scores as if they were reflections on the effectiveness of the teachers without any appreciation of how unstable they are.”
...
Dr. Sanders helped develop value-added methods to evaluate teachers in Tennessee in the 1990s. Their use spread after the 2002 No Child Left Behind law required states to test in third to eighth grades every year, giving school districts mountains of test data that are the raw material for value-added analysis. In value-added modeling, researchers use students’ scores on state tests administered at the end of third grade, for instance, to predict how they are likely to score on state tests at the end of fourth grade. A student whose third-grade scores were higher than 60 percent of peers statewide is predicted to score higher than 60 percent of fourth graders a year later. If, when actually taking the state tests at the end of fourth grade, the student scores higher than 70 percent of fourth graders, the leap in achievement represents the value the fourth-grade teacher added.
...
In many schools, students receive instruction from multiple teachers, or from after-school tutors, making it difficult to attribute learning gains to a specific instructor. Another problem is known as the ceiling effect. Advanced students can score so highly one year that standardized state tests are not sensitive enough to measure their learning gains a year later.
I disagree with the bolded part, if the reporter is reporting it accurately. Say it with me: REGRESSION TOWARD THE MEAN. I’m sure the reporter is wrong, and the professor is probably doing it right. If you beat 60% of students in a standardized test, you are probably at the 55th percentile in true talent. So, if a teacher inherits 30 kids who are each at the 60th perecentile, and then those kids once again are at the 60th percentile, then those kids IMPROVED. If they use the fact that they were at the 60th percentile two years in a row, then they are probably at the 58th percentile, and the third teacher would get evaluated against that level.
Not to mention there are aging issues to account for. We need a height and weight database on these kids as well, to see how physically they’ve matured. You’d need to know if they’ve been introduced to sex, drugs, and rock&roll, as that might influence their scores more than anything (each of those things would lead you to infer a change in thought-processing and lifestyle). (I say these things facetiously because they are kids. If they were college kids, I’d be serious.)
Bill James had a post about how to track how “regular” usage a starter was used in a given season. Basically, it’s a way to express numerically in one number the dizzying kinds of patterns all pitchers go through. David Pinto implemented James’ idea.
While it takes years to go from once-in-a-generation catcher Wieters I to go to Wieters II (Montero) to Wieters III (Bryce Harper), it took only two+ months to go from once-in-a-generation Strasburg I to Strasburg II (Aroldis Chapman).
The difference is that he walked ALOT of hitters in the minors, so he is unlike the polished Strasburg. It’ll be interesting to see what the forecasters do with Chapman (who given his walk rate, might be more like the young Nolan Ryan and Randy Johnson, themselves also once-in-a-generation pitchers).
Poz has a very powerful style of writing, one that elevates him to one of the best sportswriters in USA and Canada, and keeps the readers mind enthralled and entertained. Me? I can wield numbers in such a way as to make your mind numb. Both ways work, though Poz’s way is far more popular. (And I agree with the majority, as I’d pick him over me.)
As a perfect example of the contrasting styles, Poz was asking if a great season (or really how many great seasons) should be enough to put someone in the HOF (or your personal HOF). This is how Poz wrote that article today titled Obviopiphany, and this is how I wrote that article two years ago, titled Observed Performance Inferring True Talent (OPITT).
Poz gave you Side A that appeals to the majority, and I have Side B that finds a niche with the minority.
I love crowdsourcing because there’s so many smart people out there who know alot, and their voice drowns out those who don’t know much. And the best part: if you disagree, you are automatically in the minority.
Dave is crowdsourcing the crowd as to Crawford’s likely contract signing (meaning what they predict he will get, not necessarily what he is actually worth).
And, he notices what I’ve noticed any time I do these surveys: you don’t need alot of people. He said results stabilized after 25 votes. And, you guys probably noticed that when I run my Polls around here, after 20-40 votes, results don’t change much.
That’s why for the Fans Scouting Report, we don’t need much participation. Once I get 20 votes for any single player, I’m happy. Indeed, even at 10-15 votes, that’s pretty good. The reason is because we already kinda know the answer. And after 15 votes, you get a very strong pull toward the consensus.
And, for this reason, something like Fans tracking plays (in parks that can’t afford FIELDf/x) would have great value. You position 9 fans across the park, and get them to record what they see. Rotate them every inning to account for potential bias. See which fans are closest to consensus, and weight them heavier for the season. Indeed, you won’t even need to download play by play, because those 9 fans can agree on what the various calls were. 9 fans x 81 home games x 30 teams x 50$ = one million dollars. Call me crazy, but I think it’s worth a million bucks to 30 teams.
There are four different setups, one is official, and the other three unofficial. Depending on how you see things, feel free to choose the setup that makes the most sense to you.
John Eric Hanson is the only forecaster to finish in the top 5 in all four setups. Hanson also won the official competition last year, and I think finished fairly high up in the other unofficial competitions I ran. So, yeah, he done good.
The baseline comparison point should be Marcel. Marcel finished middle of the pack in 3 of the competitions, and IS LEADING in one of the competitions (the head-to-head one… the setup that was the official competition last year).
The results. First the three unofficial setups:
Setup 1: All Pros: All 22 Pros in the same league, 1000 leagues.
Setup 3: 1 Pros v 21 Random Joes: Each Pro faces off against Random Joes, whereby Random Joes created based off the Consensus of Pros (Value for each player for each Random Joe is within +/-5$ from the Consensus for 95% of the players, with the other 5% of players effectively removed from the pool for the Joes); 22 leagues
Setup 4: 1 Pros v 21 Random Fangraphs Readers: Large collection of Fangraphs readers were pooled in various fashions to create 21 overall draft lists for each league; 22 leagues
Giants are winning 1-0 in the bottom of the 8th with their starting pitcher, Sanchez, leading off the inning at bat. He has already thrown over 100 pitches. Sanchez is a good but not great pitcher. Somewhere around league average.
They let him bat. Nothing happens in the inning for the Giants. In the top of the 9th, he walks the leadoff batter and they immediately take him out. What was the point of letting him bat? Either you think he is still good enough to pitch the entire 9th, whether he gets the first batter out or not, or you don’t. Not to mention the fact that the next batter (after the walk) is a lefty (Gonzalez).
IOW, if you plan on taking him out if the first batter gets on, then obviously you think that he has little or nothing left in the tank. If that be the case, pinch hit for him and then bring someone else in in the 9th.
Bochy’s decisions were clearly of the “I’ll do whatever it takes to avoid criticism,” rather than actually think about what are the best moves to help his team get into the post-season.
First, “I won’t take my starter out while pitching a shutout, lest my relievers blow the game and I get lambasted for that.”
Second, “If the first batter gets on, I’ll take him out, lest I be accused of leaving him in too long.”
Third, “I’ll bring in my closer, Wilson, against the lefty batter, Gonzalez, even though my closer threw over 30 pitches the night before, lest I lose the game without bringing in my stopper.”
The other reason for bringing in the lefty to pitch to Gonzalez, or even leaving Sanchez in, is to keep Fowler from stealing (a generally underused strategy - bringing in a lefty to keep a runner from stealing second).
Needless to say, it all blew up in Bochy ‘s face, and they lost a game that they could ill-afford to lose…