12-13-2013, 12:54 PM #1
- Liked 12 Times in 7 Posts
Favorite Advanced Stats for Pitchers
Okay sabermetrically inclined - I'm ready to be enlightened by your wisdom... What are your top three favorite stats to look at when evaluating pitchers and why? Feel free to post different lists for starters and relievers.
12-13-2013, 01:17 PM #2
I prefer strikeout and walk percentage (rather than per 9 stats). Also GO/FO is a good thing to look at for all pitchers.Staff Writer for Tomahawktake.com, come check it out!
12-13-2013, 01:23 PM #3
- Liked 430 Times in 261 Posts
12-13-2013, 04:21 PM #4
Have you done any SIERA applications to past seasons? Or is this metric anywhere?
e.g. I could look at say the top 10 Cy Young Candidates over the years and was Bob Gibson's 1968 much better than the pitchers that year to even Nolan Ryan's 1981 (it was a shortened season) to Cy Young's seasons...the 'dead-ball era'...if you will. How would it fare for a guy with similar ERA...but 500 IP....100 BB 130 K.
I admit...pitching is tough to evaluate, I believe much harder to evaluate than hitting.
I'd be curious if there was a poll of hitters each year - to see how they ranked pitchers. Although they'd likely rank on 'difficulty' versus 'success' they had. John Kruk against Randy Johnson type of thing..."Well, I came out alive and he didn't hit me, so I'm thankful..." and maybe somehow he walked (success Kruk, but was extremely difficult AB). Whereas next at bat he strikes out against Jose Canseco pitching during junk innings (unsuccessful, but difficulty was easy).
12-13-2013, 04:33 PM #5
- Liked 39 Times in 25 Posts
PWAA, RA+. My favorite quick counting stat for pitchers is K-BB.
12-13-2013, 04:34 PM #6
Also, does SIERRA then correctly hand Johan the 2005 CY Young?
12-13-2013, 05:54 PM #7
- Liked 430 Times in 261 Posts
To answer the previous questions:
You can find SIERA in fangraphs. Hard to compare between seasons and really makes no sense for a million reasons other than fantasy (like comparing Walter Johnson's 1919 to Santana's 2005.)
12-13-2013, 11:47 PM #8
I wanted to know if it works across seasons...not so much for fantasy but for validity. If it shows that Boof Bonser was the best pitcher one year due to the formula or that George Cobb now is better than Cy Young in 1892.
For it to be a legitimate formula, advanced metric, it must hold for all seasons. Even when guys pitch 400 innings.
The difference between linear and exponential or logarithmic function.
12-13-2013, 11:56 PM #9
According to SIERRA, Clayton Kershaw was the 8th best pitcher in MLB last year...
I would have said it was one of the top 5 single seasons of the past 25 years!
Batters his .195 off of him. His WHIP was 0.915. His ERA was 1.83. These are retarded numbers.
It also says Rick Porcello was the 13th best pitcher in MLB, despite a 4.32 ERA and batters hitting .270 against him (the MLB average was .253) so anyone who hit against him did better than average? And yet according to SIERRA, he's the 13th best pitcher in MLB? Using a line of regression...the league average ought to be about .320 if he's the 13th best. Madison Bumgarner and David Price were 17th and 18th respectively.
That's pretty tough to stomach. I'll search out more seasons though.
Last edited by twinsfan34; 12-14-2013 at 12:00 AM.
12-14-2013, 12:28 AM #10
Interestingly, Fangraphs just posted an article/tool that lets you explore pitching metrics, evaluating which are most closely related. For example, you can try to figure out how well K% might predict ERA. Ultimately, the findings are pretty interesting (ERA is not as valuable of a predictor as other metrics for projecting future ERA). All of this is great info if you are interested in projecting guys. Here's the link: http://www.fangraphs.com/blogs/tool-...t-correlation/
One of the simpler advanced metrics I look at is xFIP, which normalizes unstable factors like BABIP and HR/FB ratio to a league average. You can think of xFIP as a more sophisticated version of ERA.
K:BB is huge in my opinion, though not particularly advanced. Fangraphs has an excellent library for understanding advanced metrics, making them much easier to comprehend for someone new to that area of the game.
Last edited by B Richard; 12-14-2013 at 12:32 AM.Bring a song and a smile for the banjo,
Better get while the gettin's good
Catch a ride to the end of the highway
Where the neons turn to wood
This user likes B Richard's post and wants to buy him/her a steak dinner:
12-14-2013, 12:57 AM #11
Top 38 SIERRA seasons since 2002 (all the farther Fangraphs goes back).
Top 38 you ask, because Kershaw's 2013 was the 38th best season since 2002.
SIERA = 6.145 – 16.986*(SO/PA) + 11.434*(BB/PA) – 1.858*((GB-FB-PU)/PA) + 7.653*((SO/PA)^2) +/– 6.664*(((GB-FB-PU)/PA)^2) + 10.130*(SO/PA)*((GB-FB-PU)/PA) – 5.195*(BB/PA)*((GB-FB-PU)/PA)
At this point not sure I could give any validation to SIERRA. I could acknowledge it as a work in progress, but you can't 'miss' on an all-time historic season by that much.
12-14-2013, 09:13 AM #12
- Liked 430 Times in 261 Posts
12-14-2013, 11:17 AM #13
But really, Kershaw...the 38th best season out of the past 11 years? and Rick Porcello is on the same breath as Kershaw?
Ok, lets go with the BABIP. Analyzing if the.251 BABIP is luck....
Let's look at his body of work the last 5 years (since he became a full time starter).
Those totals give a standard error (95% confidence) that will have Kershaw having a BABIP between .247 to .284...every year. If he plays 20 years, he'd likely only have ONE season outside of that range. BABIP .286, boom.
If going by 2013 pitchers, his 'worst' year is Homer Baily, Patrick Corbin, Jose Quintana level. Worst year...
.251 is a little outside of his mean, sure, but it's 95% predictable. It's not a surprise. Now look at this, he just turned 25. Most pitcher's best years are 27-29. He's not even there yet. He's more likely than not to post better numbers even next year and through age 29 than his career average.
Lucky...well, let Wade Boggs eat his fried chicken, I'll take the .330 to .360 batting average. It's probably the just the chicken, er I mean luck
But really, there's a reason Tony Gwynn had a high BABIP every year and Rob Deer did not - it's not luck - unless your definition of luck includes skill or some sort of personal attribute.
I believe luck comes into play when you see a season way out of the norm, Paul O'Neill's .359 season (1994). Brady Anderson's 50 HR (1996). Darin Erstad's .355 (2000). Let's see..pitching...umm, John Tudor's 1985, .193 ERA, .0938 WHIP, .209 BA - all so far out of the normal distribution range.
When Babe Ruth hit 40+ HR [the other end of HR/FB%] every year, was he just lucky?
When Babe played over 135 games, there was a 95% confidence rate he'd hit between 44-52 HR.
Here's a view that says HR/FB is not tied to skill, though. Maybe for pitchers only.
As I have this hunch...what if you increased Joe Mauer's FB rate...when he was near 30%, up about 3-5%, and he hit his high of 28 HR. The guy smacks the ball pretty hard...given 450 ABs...a 3-5% increase in FB would give him 13-23 more shots at a HR. He normally hits 9...so about 80% of those extra FBs made it over the fence.
I think, as a former pitcher, there were certain guys I was willing to let them try to 'groove' one. I wouldn't waste my pitches so much on certain hitters I felt I could get a fly ball out of that I was pretty certain couldn't leave the park. So my HR/FB rate was probably very low, but it was on purpose. When I had a few guys who were 6'3" 230-260 types, I pitched low, low, low...ground balls, ground balls, ground balls - was I successful all the time? Absolutely not, but I know I pitched certain guys into certain balls.
In softball, I pitch quite a bit. When we have a few guys out and our defense isn't as strong I'll make a few shifts and put my strongest defenders at 3B/SS and pitch inner have of the plate. Whereas maybe 30-40% of balls go the opposite way on a normal game, I'd say maybe 10-15% make it to the right side of the field when I pitch to that. I have higher walk rates too though, instead of 0-2, I might walk 2-5. So that's another factor that's influenced. My SIERRA would probably not do as well if I had to pitch a lot of games that way. But we'd win those games.
I'm sorry if I blasted your reasoning at all - thanks for sharing metrics you look at. I thought it's important (and still do) that it works for many different eras of play. Instead of hyperanalyzing a bunch of other people's stuff, maybe I should spend more energy on putting together a better one.
Last edited by twinsfan34; 12-14-2013 at 11:33 AM.
12-14-2013, 11:24 AM #14
That same article had some interesting things to say about those guys who ended up with great SIERRA...indirectly of course...as the xFIP directly relates to SIERRA.
Expected Fielder-Independent Pitching (xFIP)
xFIP = [13(.106*Fly Balls) + 3*BB - 2*K] / IP + 3.20
Developed by Dave Studeman at The Hardball Times, xFIP keeps homeruns constant at a league average rate (10.6% HR/FB) for every pitcher. The result is a more stable statistic and a more accurate predictor of future ERA in the short and long term than FIP– particularly for pitchers who change teams and leagues more than others.
xFIP is on a slightly higher scale than ERA and FIP. Therefore, a xFIP of 3.20 is more impressive than a FIP or ERA of 3.20.
Groundball pitchers, like Brandon Webb and Rick Porcello, have higher HR/FB rates than fly-ball and strikeout pitchers, and therefore xFIP is a less accurate predictor of their future performance.
For pitchers who consistently outperform (or underperform) league-average HR/FB rates, like Tim Lincecum for instance (8.8% career, above 9% in 1 season), xFIP is a less accurate predictor of their future performance.
xFIP has the highest correlation (determined by root means square error) with future ERA of all pitching metrics.
Which category does Kershaw fall into? And then...what should the modification be for him - as seen for the misses on Webb, Porcello, and Lincecum.
12-17-2013, 04:13 PM #15
- Liked 22 Times in 15 Posts
- Blog Entries