SB Nation User Blog: Sky Kalkman

A Note About Becoming a BtB Author: Contributing to the Community Helps

Sky Kalkman — Wed, 01 Jul 2009 14:03:25 -0000

When I recently put the call out for more contributors to BtB, I got a fair number of responses. (I think I've replied to everyone, so if you haven't heard from me, bug me again.) Some guys have already started writing (Tommy and Graham) and some I'm still exchanging emails with. One reason I'm hesitant to add some of the folks I've chatted with is that I'm not familiar with them or their work.

Now, I'm not the eternal judge of who's who in the baseball world, but when I can only find a few articles by you from six months ago, that's not a great resume, even if the articles are good. So, if you're somebody I've emailed with or another lurker wondering how you can write for BtB, here are some tips:

Be more active in the BtB community. It's kind of funny to hear people say they love the site and want to write for it, but they've never commented here before or don't even have an account. By commenting, you earn the respect of the authors and readers and can make your own niche in our community. Even better, write some FanPosts which would be along the lines of what you'd want to write for the site itself. They get almost as much attention as front page posts, and even get bumped to to the front page when they're good. If people are liking your writing, it's an easy transition over to being an official author. Probably the most straight-forward job interview out there.
Write consistently. That could be here, at your own blog, or at a group blog. Even writing once a week can be tough and it's nice to know you can do it. More writing also means you'll probably get better at writing, which, well, the perks of that are obvious.
Write from a unique angle. While we hate to admit it, there are certainly some stathead groupthink cliches: Joe Morgan sucks, fielding matters, regression regression regression. Those cliches have merit, but it's nice to read something different, and people like different.

Finally, a word on commenting. There are many ways to add to the conversation beyond supplying new information. Asking questions is a great idea -- ask for clarification, ask how you could take an idea in a different direction, ask if the assumptions made in the article are good ones, etc. Even us authors ask questions a large percentage of the time (you should see our email threads.) Or just say that you really liked a certain piece of the article and want to see more on that topic. Keep those comments coming -- they're awesome.

Saber Sunday Game Thread: Yankees at Mets 6/28

Sky Kalkman — Sun, 28 Jun 2009 15:32:43 -0000

Happy East Coast Bias Night!

We took a break from our Sunday night game threads for the NBA Finals and Fathers Day, but we're back. Count on this being a weekly feature from now until the end of the year. And count on the NY/ESPN Yankees being a part of things a few more times. Tonight's starters could provide a lot of scoring opportunities:

	W-L	G	GS	CG	SHO	SV	BS	IP	H	R	ER	HR	BB	K	ZiPS RoS FIP	tERA
2009 - Chien-Ming Wang	Feels like 0-72	10	7	0	0	0	0	31.1	56	40	39	6	15	25	3.98	6.62

	W-L	G	GS	CG	SHO	SV	BS	IP	H	R	ER	HR	BB	K	ZiPS RoS FIP	tERA
2009 - Livan Hernandez	dumb	15	14	1	0	0	0	86.2	97	40	39	11	25	48	4.93	4.89

As usual, the comments section is for discussing anything baseball-related from a saber-slanted perspective, whether it's the Yankees-Mets game tonight or not. Game time is 8:05 EDT. Go Pirates!

Weekly Box Score: Through June 26th

Sky Kalkman — Sun, 28 Jun 2009 17:26:26 -0000

In case you missed anything on BtB over the past week, here's your weekly round up. (You can always check out the Archives, which will also list any new comments as long as you're signed in to SBN.)

Tommy Bennett joined the BtB team, bringing links to saber-slanted from around the web into one one place in a column we call the Daily Box Score. If you have any feedback, please leave it here.
Sky "defended" Harold Reynolds, although his essay was really about how two sides of a debate should communicate, if they ever hope to bridge the divide. Yes, statheads can be obnoxious. Stop it.
Beyond the Box Score now has a Facebook page. Become a fan!
Erik (who recently started writing for Fangraphs, congrats!) crossed party lines (did you know he's a Cardinals fan?) by helping convince the Reds they should jettison Willy Taveras from the lead off spot, the starting lineup, and the state of Ohio.
Harry busted out more pitch f/x-based leader boards. I'm not sure why these aren't more popular, but learned that Kevin Baker has the highest percentage of pitches swung at, Livan Hernandez never misses bats, Ted Lilly throws a huge percentage of pitches in the extended strike zone, and Joba Chamberlain can't get anyone to swing at his pitches out of the strike zone.

R.J. ventured into basketball territory, pondering about the value of second round draft picks in the NBA draft.
Daisuke Matsuzaka's periphals are very similar to last season, but his ERA is about thirty seven runs higher. Cosmic joke or injury issues? Tommy Rancel took a look.
Here's a great group of names, who all garnered Rookie of the Year support in 1986: Jose Canseco, Wally Joyner, Cory Snyder, Ruben Sierra, Danny Tartabull, and Mark Eichorn. Jeff gave them the career trajectory treatment in a nice graph. Other Graphs of the Day included the most and least effective sinkers, how much various power rankings match actual win-loss records, context-dependent (using WPA) stolen base cut-off rates,
Sean Smith's WAR database is awesome, but only goes back through 1995. Jeff crunched the numbers to put Win Shares on the WAR scale so we can compare players all the way back through the beginning of time.
Sky took a look at the Rockies' pitching staff, which is among the best in the majors, despite an ERA in the bottom third of all teams.
Justin's power rankings once again featured four AL East teams at the top of the list, but also recognized the Rockies as the big upward movers of the week, while the Royals took home the downward honor. Dan followed up Justin's power rankings with his Poor Man's Projected Standings.
Jack took a look at question not unfamiliar to Reds' fans -- what the heck is up with Jay Bruce's freakishly low BABIP? Is he unlucky or just unique?
Jeff noted that hitting streaks tend to take place during warmer weather, although more hitting streaks take place in the earlier part of the summer, not during the absolute hottest days. I love articles with answer and more questions.
Tommy Rancel looked at two surprising division leaders -- the Tigers and Rangers -- and preached to the old football cliche, Defense Wins Champsionships.
You want more DL database yumminess? Jeff dissected the Royals DL info from 2002 through 2008.
Harry's weekly column, New Arms of the Week, checked in on Sean O'Sullivan, Alfredo Figaro, and Brad Mills.
Sky, while acknowledging he knows as much about SQL as he does about Nigerian moth mating rituals (very little, in case you were wondering), presented an introductory SQL tutorial. If you want to slice and dice baseball data, it's really not that hard to get into, if you take it one step at a time. Look for Part II soon.
Got any ideas of what the smart people should be doing with pitch f/x and hit f/x data? Join the brainstorming thread.

Do the Colorado Rockies Have the Best Pitching Staff in MLB?

Sky Kalkman — Thu, 25 Jun 2009 16:59:14 -0000

More photos » by David Zalubowski - AP

Browse more photos »

If I were to ask which team has had the best pitching staff in the majors so far in 2009, I doubt many people would answer the Colorado Rockies, for two reasons. One, "Rockies" and "best pitching staff" have never been connected in any sentence written since Denver got an MLB franchise in 1993, and two, they have a 4.54 ERA this year, good for only 23rd in the majors. But according to Justin's implementation of tRA, the Rockies are ranked first, deserving of an ERA more like 3.70.

What do the adjusted stats like about the Rockies? And do they really know what they're talking about? Let's find out.

Adjusting ERA

ERA has a few problems as a measure of pitching performance, most notably that it doesn't separate pitching from fielding and doesn't take ballpark into account. Given that the Rockies have a poor defense and play in the most hitter-friendly park in the majors, we definitely need to consider those effects.

By UZR, the Rockies have been 15 runs worse than the average team in 2009, which is consistent with their -35 run 2008 season. Simply removing 15 runs from the staff's ERA brings it down to 4.33, which is 16th among all UZR-adjusted team ERAs. That's still a bit higher than their 4.10 FIP, but within shouting range.

And then, of course, there's Coors Field. It's not as hitter-friendly as it used to be, but still increases run scoring by 21% according to Terpsfan1's park factors. Colorado's played eight more road games than home games so far this year (which is good news going forward) so we need to reduce their runs allowed by 8.7%, which is the weighted average of their road and home park factors. That puts their ERA at 3.98, very close to the top five. Not the best, but certainly quite good.

Of course, we haven't considered the fact that Rockies pitchers haven't had to face many designated hitters. In general, an NL pitcher switching to the American League will see an ERA bump of half a run, meaning the Rockies team ERA would probably fall outside the top ten, but not out of the top half.

How Have They Done It?

For years the Rockies have tried to acquire talented ground ball pitchers and it seems they've finally succeeded. The staff leads the majors in GB/FB ratio, and has managed to hover around league average at 10.5% of fly balls leaving the yard. Even while playing home games in Coors, they've allowed under one home run per nine innings, good for tenth in the majors, thanks to all those ground balls.

The staff's strike out rate isn't stellar at 6.7 K/9, but that's to be expected given that the Colorado air doesn't help pitches break as much. Their walk rate is quite good, at 3.2 BB/9, and their overall 2.1 K/BB ratio is tied for 8th in MLB.

Four-fifths of the starting rotation have been especially impressive, and the one-fifth isn't who you might think. Ubaldo Jimenez, whom Tommy recently profiled, sports a 3.31 FIP, eighth best in the NL. Jorge De La Rosa, despite a poor ERA, sits a 4.02 FIP and is striking out nearly 10 K/9. Rays cast-off Jason Hammel sits at a 3.67 FIP and journeyman Jason Marquis is at 4.37. Aaron Cook's been the worst starter, with a 4.80 FIP -- striking out 4.5 K/9 just isn't a recipe for success -- but that's still better than replacement level, especially in Coors.

Are They Really The Best Pitching Staff So Far?

Probably not. Even using Justin's version of tRA instead of adjusted ERA or FIP leaves us needing to adjust for the DH factor. However, the Rockies do, quite possibly, sport the best pitching staff they've ever had, one that rivals the Dodgers, Braves, and Giants for best in the National League. As has been the case for a few years now, the strength of the Colorado team lies in their arms and not the position players who are overrated by their home park and some of who struggle in the field. This is a good team, led by their pitchers, who should be in the NL wild card race into September.

How Do You Like the New Daily Link Roundup Posts?

Sky Kalkman — Thu, 25 Jun 2009 18:21:49 -0000

Hopefully you've noticed a new daily feature around here --the Daily Box Score. Tommy Bennet started rounding up saber-slanted links on Monday and it's been full steam ahead ever since. Since the goal is to provide a service for readers, we'd love to hear your feedback on the column so far.

Are there enough links? Too many links?

Is there enough commentary? Too much commentary?

Are we covering a wide enough range of articles? Wide enough range of teams? Too wide?

Is there a certain subsection of saber-slanted content we should pay more attention to?

Do you like the publish time of about noon eastern time?

Any other suggestions?

Poll

How do you like the Daily Box Score so far?

Graph of the Day: Comparing Power Rankings

Sky Kalkman — Wed, 24 Jun 2009 15:29:35 -0000

Power Rankings are funny. They all try to measure something different and cater to a different audience. And no matter the system, fans of some team will disagree vehemently. Nonetheless, i was curious to see how Justin's BtB rankings compared to other systems.

The comparison set: Pythagorean wins, second- and third-order wins from BPro, BPro's Hit List factor (which is simply an average of actual record, Pythag, and second- and third-order winning percentage), and RPI. BtB_lg removes the league adjustment.

All systems are compared to actual records as of late last week using both correlation and total absolute error. Now, "error" isn't a bad thing here, because these power rankings aren't trying to predict past record, they're trying to extract true talent and ignore fluke.

In general, as correlation goes up, aboslute error should go down. One system where that doesn't happen is RPI, which correlates the closest to actual record, but has a much tighter set of ratings, making its absolute error quite high. The system that matched actual records the least was BtB's, followed by the non-league-adjusted BtB number and the two most adjusted BPro systems. Personally, I'd like to see a power ranking NOT match actual record that closely, because I want new information. Of course, not matching closely doesn't mean the system is any good.

Anyway, my major takeaways from this graph are that BtB's system definitely doesn't just match win-loss record (as many commenters have creatively pointed out) and that RPI and BPro's systems DO.

Looking For SQL & Tech Geek Help For Collaborative Projects

Sky Kalkman — Wed, 24 Jun 2009 15:37:59 -0000

I have a ton of research ideas running through my head, but lack the technical abilities to make most of them happen. With the other BtB authors pretty busy with their own research, I thought I'd see if anyone out there was interested in working with me on some projects. I think most of the them are fairly straight forward, but hey, what do I know?

If you're interested in collaborating, combining my ideas with your tech skillz, even if you don't consider yourself a big statgeek, send me an email: skyking162@gmail.com

Help Me Expand Who I Follow On Google Reader

Sky Kalkman — Mon, 22 Jun 2009 19:47:57 -0000

I switched to using Google Reader within the past few months, mostly because of the Sharing feature. Google doesn't make it easy to find people who share articles, however. So, I'm calling out to anyone who reads this site and uses Google Reader -- tell me (and therefor all of us) who you are so we can "follow" you. My gmail address is skyking162 (@gmail.com). If you read this site, chances are we have common interests and I'd find your Shared items interesting.

Now if only they'd make it easier to organize RSS feeds...

Sky's SQL Tutorial, Part One (Which is Like a David Ortiz Bunting Tutorial)

Sky Kalkman — Mon, 22 Jun 2009 14:22:15 -0000

When I challenged folks to write a complex SQL query using the Baseball Databank database in the last installment of Saberizing a Mac, I anticipated we'd all start at ground level and work our way up to more complex queries until we eventually figured it out. Well, sometimes with a lofty goal, taking all those baby steps can be overwhelming and we just see one large, overwhelming step. I decided to start from the beginning, with the help of this SQL tutorial and write a bunch of simple SQL queries, each adding just a bit of complexity. Not only does starting simple help, but forcing myself to write the queries instead of copying them from somewhere else helped my brain process what was going on. All the mistakes along the way were productive.

I'm going to paraphrase the tutorial I used and show the SQL queries I came up with. My suggestion to you is to follow along, manually typing in each and every query, and modify them a bit (different tables, different stats, different qualifications, etc.) so your learning path is as different from mine as possible.

SELECT, FROM, WHERE -- SELECT tells the program what data you want to use, be it names, teams, or stats. These are field names, which you can look up in the individual tables at the top of the columns of data. FROM tells the program which table(s) you're pulling data from. WHERE tells the program what conditions you're setting -- you'll almost always want to pull only certain data, not all of it.

My first query looked like this:
SELECT playerID, yearID, HR
FROM Batting
WHERE HR >= 50;
It pulls the player id code (not name, that'll come next time with JOINs), the year, and number of home runs. The data is in the Batting table. And I only wanted to pull instances (rows) where the home runs were at least 50. (This is not all the 50 HR seasons, though, as I'll explain later.)

Some details on WHERE -- When you want to set conditions based on numbers, the WHERE statements are easy. With text, you have some more options. If you want to choose a specific text string, you need to surround it in quotes and you need to put AND or OR between multiple conditions. Like this:
SELECT playerID, yearID, HR
    FROM Batting
    WHERE HR >= 50 AND (teamID = "SEA" OR teamID = "NYA");
You should see a bunch of Griffey- and Ruth-related results. You can also pick apart pieces of text strings using LIKE. (I believe the geeks call these "regular expressions".) Let's say you only want to use players whose last names start with "R". Then you'd do this:
SELECT playerID, yearID, HR
    FROM Batting
    WHERE HR > 40 AND playerID LIKE 'R%';
The % sign means there could be any number of characters after the R. If you wanted to indicate just one wildcard character, use _ (an underscore). You can also use NOT LIKE to take the complement set, use multiple combinations of letters/%/ _, and include numbers. For more tips, try this.
ORDER BY -- Ok, nice you you have your results, but what if you want to sort them? Use ORDER BY followed by a field name. This makes the most sense if you order by whatever data you're pulling but I don't think it's necessary. For example, you could sort home runs hitters by the number of doubles they hit:
SELECT playerID, yearID, HR
    FROM Batting
    WHERE HR > 40 AND playerID LIKE 'R%'
    ORDER BY 2B Desc;
Doing calculations -- When you SELECT data, some stats you want already exist, like home runs, while some don't, like OBP. In those cases, you have to tell the program to do the calculation and provide a name. (You can also provide a new name to a stat without doing a calculation, like renaming 2B as "Doubles".) Something like this, although this isn't a great definition of OBP:
SELECT playerID, yearID, HR, (H+BB+HBP)/(AB+BB+HBP) AS "OBP"
    FROM Batting
    WHERE HR > 40 AND playerID LIKE 'R%'
    ORDER BY 2B Desc;
There are also some built-in calculations you can use, like MIN, MAX, SUM, AVG, and more.
GROUP BY -- The data in the data tables is extremely granular, probably more granular than we want most of the time. For example, in the Batting table, the same hitter in the same season could have multiple entries, even for the same team. All "stints" (consecutive games played for the same team) are separated. So if a player was a Giant, then an Oriole, then a Giant again, he'd have three separate lines. Two lines would have stint listed as 1, while the second stint for the Giants would have stint listed as 2. If you want all of a player's season data aggregated, or all of his data for the same team aggregated, you need to tell the program to do just that. Hence, GROUP BY. If you only group by playerID, though, it will combine all seasons. So you need to list all the ways you want the groupings to be unique. I'm using player and season below, but you might also want to separate performance by team:
SELECT playerID, yearID, HR
    FROM Batting
    WHERE HR > 40 AND playerID LIKE 'R%'
    GROUP BY playerID, yearID
    ORDER BY HR Desc;

Ok, that's as far as I've gotten for now. I just took a peak at Colin's two SQL tutorials and he has similar stuff, plus more, so go check those out. I'll definitley continue with part two soon.

Pitch and Hit F/X Brainstorming Thread

Sky Kalkman — Fri, 29 May 2009 19:47:24 -0000

Harry and I were talking a couple weeks ago -- right after his pitch f/x chat, actually -- and one thing we agreed was a great help to furthering analysis, especially new lines of analysis, was group discussion. When a bunch of people start brainstorming and taking tangents off of others' ideas, great things can happen. With that in mind, here's an open thread for brainstorming where we can take pitch f/x and hit f/x analysis.

Before those of you who would never do your own analysis tune out, stop. One of the most helpful points of view is from folks who don't understand the ins and outs of the data. Why? You don't know what the limitations are and might come up with something nobody's thought of or something that would take a little creativity to figure out.

Here are some lines of thinking to get this brainstorming going.

New metrics. Harry's done things like line drive to swinging strike ratio for certain pitches. Maybe you'd like to see something that measures control (percentage of pitches within three inches of strike zone border?)
What are some ways to tackle pitch selection and sequencing? Simply what percentage of each pitch a pitcher will throw after a given pitch? Work count into it somehow?
Are there better ways to use graphs?
Can we not use graphs?
How do we use more pitch f/x from the hitters' point of view. What pitches does he like? Can we pick out "guess hitters"? Are there combinations of pitches he doesn't handle well? Can we detect platoon effects beyond L/R and GB/FB?
Other?

You don't have to throw out complete ideas -- the whole point is that if you throw out partial ideas, either incomplete or vague, someone else can take over where you left off