Adrift in a sea of advanced stats

[Monkeypaws]

Hey all, 

 

I see a lot of advanced stats flashed at Twins Daily. 

 

I was a wondering which among them you all saw as a real improvement over traditional triple crown stats for pitchers or hitters?

 

 

[ashbury]

Hey all, 

 

I see a lot of advanced stats flashed at Twins Daily. 

 

I was a wondering which among them you all saw as a real improvement over traditional triple crown stats for pitchers or hitters?

Depends on your purpose.

 

For looking back, the stats you named are fine by me. They tell what happened, more or less. Most of the time I don't care about what "should" have happened.

 

For forecasting, rate stats moreso than counting stats are what I want, and the less opportunity-biased, the better. OPS works pretty well in my view for batters. I look at OPS-against, for pitchers.

 

I hope my team's analytics department has better forecasting tools, but as a fan a quick-and-dirty tool seems enough.

 

[cardsfan]

Out of 15 AL teams 7 lost 89 or more. The stats that you look at is how well did your team do against winning and losing teams?

You hit 200 Hrs against teams with losing records and 100 against winning records.

 

The AL has dominated inter-league play except for last 2 years. You beat the Marlins 2 out of 3, but, 8-12 in interleague play.

 

The Indians lost 18 more than they won against winning teams and still won 93 when I count. Just awful. Against the Twins they were 9-10.

 

Somehow it all makes sense about the Yankees and Twins hitting 300 Hrs. It was like the 1961 and 1962 expansion seasons.

 

Yes, the Yankees and Astros seem dominate still.

[Richie the Rally Goat]

Hey all,

 

I see a lot of advanced stats flashed at Twins Daily.

 

I was a wondering which among them you all saw as a real improvement over traditional triple crown stats for pitchers or hitters?

the advanced stats just like all stats have their pros and cons. I like Fangraphs, mostly for their mobile site and splits tool. Many a debate has been held here between Fangraphs and baseball reference. There are differences in the calculations between them. https://blogs.fangraphs.com/glossary/

 

 

For relievers reviewing history I like K%, BB% and WPA. The sample sizes for most relievers is so small that earned run average doesn’t tell you a lot. K9 and BB9 penalize relievers compared to starters. WPA as a stand in for leverage helps identify their usage patterns.

 

Starters reviewing history I tend to lean on xFIP, I think it does a good job of controlling for variables.

 

Hitters I rely on OPS and wRC+. OPS is quick and dirty. WRC+ controls for league and puts hitters on and 100= average scale. It’s a good comp tool.

 

Fielding, I waffle on metrics. I want to use them, but I don’t have much trust in them.

 

I don’t do a ton of forecasting, so I’m curious what the other members do there.

[The Wise One]

K% over k/9. K/9 can look good but unless the k/9 is 81 what happens with the other batters becomes more important. K% is more telling for ability. WHIP works for that. too.  Many stats need a fairly large body of work to have much validity.

[IndianaTwin]

Without getting too nerdy and geeked out, because "we were told there would be no math," I'd love to see someone take an educated shot at a glossary, with...

 

Stat, what it purports to measure, what are its limitations, and what is an average mark and an all-star mark.

[mikelink45]

I am not an advanced stathead.  I do see shifts.  To say that wins do not count in the era when starters actually completed there games is a travesty.  They took charge and earned their wins.  To say BA does not count is ridiculous.  Yes I want Rod Carew and George Brett and Tony Gwynn on my team.  To say RBIs do not count is ridiculous.  Of course they count.  If you have a player in scoring position and do not bring them in we lose.  Runs do matter - I believe the team with the most runs wins.

 

HRs are over rated.  Stolen bases are underrated.  Give me a triple crown hitter any day.

 

Yes I look at and respect the new stats.  Some are just initials to me, but those I understand are interesting and I enjoy them, but to toss out the old stats is foolishness.

 

A three hundred win pitcher is still the top of my list of accomplishments. 

[Richie the Rally Goat]

I am not an advanced stathead. I do see shifts. To say that wins do not count in the era when starters actually completed there games is a travesty. They took charge and earned their wins. To say BA does not count is ridiculous. Yes I want Rod Carew and George Brett and Tony Gwynn on my team. To say RBIs do not count is ridiculous. Of course they count. If you have a player in scoring position and do not bring them in we lose. Runs do matter - I believe the team with the most runs wins.

 

HRs are over rated. Stolen bases are underrated. Give me a triple crown hitter any day.

 

Yes I look at and respect the new stats. Some are just initials to me, but those I understand are interesting and I enjoy them, but to toss out the old stats is foolishness.

 

A three hundred win pitcher is still the top of my list of accomplishments.

Bert Blyleven is the classic example of why Win isn’t a good pitcher stat. He spent decades on terrible Pirates, Indians, Rangers and Twins teams. If he played for the Reds, Mets, Tigers he might have had 400 wins!

 

Instead he never got that 300 win mark and struggled to get into the Hall as clearly one of the greatest pitchers to have ever played.

 

Win is the ultimate team stat. It just doesn’t do a good job of describing a pitcher in isolation.

 

 

Of course batting average counts. It’s a completely valid stat and does a good job of describing a hitters ability to make contact. There are other stats that correlate more closely with scoring runs by weighting extra base hits and including walks. That doesn’t take away from batting average.

 

 

If scoring runs is your jam and RBI is how you measure it, WPA for hitters might give you all you want and more. Win Probability Added, accumulates odds of winning to game outcomes. I like it for relief pitchers but haven’t found enough utility for it in other position sets. It might be worth some research.

[Thegrin]

For the last 35 years I have been enamored by the "Runs Produced" stat. (Runs + RBI- HR). They have fiddled with it some, but it still works. In this century I started paying attention to OBP.  It helped me appreciate Joe Mauer. 

[Nine of twelve]

 

If scoring runs is your jam and RBI is how you measure it, WPA for hitters might give you all you want and more. Win Probability Added, accumulates odds of winning to game outcomes. I like it for relief pitchers but haven’t found enough utility for it in other position sets. It might be worth some research.

I think WPA is on the right track as a way to measure a player's worthwhile performance because the goal of playing is to win games and it addresses that. However, its strength can also be its biggest drawback. It is weighted in favor of good performance in close-and-late situations, which is good because good performance in those situations is more important. The problem is that it penalizes players on very bad teams and very good teams because such teams have fewer close-and-late situations.

It seems to me that a better way to use WPA would be to compare a player's WPA outcome in a given situation with the best possible and worst possible WPA outcomes in those situations. For example, say a player comes to bat with no outs and two on in the bottom of the ninth with his team down two runs. The best possible outcome would be hitting a home run. The worst possible outcome would be hitting into a triple play. I would guess that some stat-head somewhere has looked into doing this but I've never researched it to find out if it has been done. If it has never been done then it would be an excellent project for someone with the skill, the time, and the interest.

[Mike Sixel]

 

Without getting too nerdy and geeked out, because "we were told there would be no math," I'd love to see someone take an educated shot at a glossary, with...

 

Stat, what it purports to measure, what are its limitations, and what is an average mark and an all-star mark.

Well, Fangraphs has just such a thing for their stats.

 

I like wrc+ for hitters. Pitchers? I waffle....

[Vanimal46]

I’m not a big fan of Fangraphs going all in on FIP for pitcher evaluations. Before the change, fWAR and bWAR were similar and gave you a good base. Now there’s a significant gap in fWAR/bWAR for specific pitchers that I don’t like.

[jkcarew]

With regard to WAR...don't love it for pitchers based on Van's point above. Also don't love WAR for position players much, either...or at least I don't take it too literally. Too influenced by defensive metrics, which are still, more or less, garbage, IMO.

 

I go with my eyes on defense, and with wRC+ and OPS+ for offense...as others above have mentioned. With pitchers...I'm just confused. WHiP (combined with BABiP against) maybe? ERA still factors some in my book. And honestly...I still like innings pitched for starters (or innings per start). I know the standards are changing there...but still so valuable.

[dex8425]

Defensive metrics: if you don't look at WAR, or DRS, or UZR, or OAA, what do you look at? Fielding percentage is worthless. Eye test? Well, that works if you watch every single inning of every game and you're a scout. I think you have to look at advanced stats but also understand the pros and cons of each. If you don't look at them, please don't say someone is an elite defender because he won a gold glove.

 

Offensive metrics: I really like OPS. It's pretty objective at how good a hitter is IMO. If your OPS is high, you hit for doubles and hit homers and you get on base a lot. I don't care about batting average. Give me Sano with his .920 OPS and .230 average over Luis Arraez and his .820 OPS but 310 batting average. Sano is the more dangerous hitter. OPS plus is useful for comparing AL and NL and guys who play at Coors Field. 

 

Pitching metrics: WHIP might be the most useful single stat. But I like ERA+ too. I always look at ERA and FIP together. I don't like fangraphs because it only uses FIP for WAR. 

 

 

I'm definitely a "counting stats are dumb" guy. 

[Otto von Ballpark]

 

I’m not a big fan of Fangraphs going all in on FIP for pitcher evaluations. Before the change, fWAR and bWAR were similar and gave you a good base. Now there’s a significant gap in fWAR/bWAR for specific pitchers that I don’t like.

Hasn't Fangraphs WAR always been FIP-based? At least, I can't remember using fWAR before that. And frankly, why would anyone? There's nothing particularly useful about two systems producing roughly the same figures.

 

I think it's good to have different takes, as long as one understands the differences. And Fangraphs publishes both fWAR (based on FIP) and RA9-WAR (based on actual runs allowed) on their site too.

[Otto von Ballpark]

 

Pitching metrics: WHIP might be the most useful single stat. But I like ERA+ too. I always look at ERA and FIP together. I don't like fangraphs because it only uses FIP for WAR. 

I mentioned in the above post, but Fangraphs actually publishes both their WAR based on FIP, as well as a RA9-WAR based on actual runs scored.

 

I understand not fully endorsing FIP, but I like having both of those figures available over only have one.

[Vanimal46]

Hasn't Fangraphs WAR always been FIP-based? At least, I can't remember using fWAR before that. And frankly, why would anyone? There's nothing particularly useful about two systems producing roughly the same figures.

 

I think it's good to have different takes, as long as one understands the differences. And Fangraphs publishes both fWAR (based on FIP) and RA9-WAR (based on actual runs allowed) on their site too.

Not sure. I’ve just noticed the stark difference in WAR value for players last season...

 

Having different takes is okay, but some tell a bad story. Take Ricky Nolasco for example... a player who consistently had a better FIP than ERA throughout his career. If we looked solely at his fWAR and FIP we would think there’s room for growth. Turns out his bWAR was more trustworthy.

[Otto von Ballpark]

 

Having different takes is okay, but some tell a bad story. Take Ricky Nolasco for example... a player who consistently had a better FIP than ERA throughout his career. If we looked solely at his fWAR and FIP we would think there’s room for growth. Turns out his bWAR was more trustworthy.

Sure, but I don't think anyone suggests we should look solely at fWAR and FIP, especially as the sample size grows to multiple seasons.

 

It's still interesting information -- I like being able to find FIP/ERA divergences, and track how those players perform.

[ashbury]

Sure, but I don't think anyone suggests we should look solely at fWAR and FIP

I have an acquaintance who works at FanGraphs. Maybe I should ask him.

 

[Otto von Ballpark]

 

I have an acquaintance who works at FanGraphs. Maybe I should ask him.

I mean, yeah, it's their default for fWAR and thus gets referenced a lot. But I doubt many Fangraphs writers will go to their grave thinking Ricky Nolasco was a better pitcher than his bWAR suggests.

 

I mean, I'm not even sure what the complaint means. Should fWAR default to RA9-WAR, and then Fangraphs could list "FIP-WAR" separately? Does it really matter? There isn't going to be one perfect number anyway.

[Vanimal46]

I mean, yeah, it's their default for fWAR and thus gets referenced a lot. But I doubt many Fangraphs writers will go to their grave thinking Ricky Nolasco was a better pitcher than his bWAR suggests.

 

I mean, I'm not even sure what the complaint means. Should fWAR default to RA9-WAR, and then Fangraphs could list "FIP-WAR" separately? Does it really matter? There isn't going to be one perfect number anyway.

The complaint was simple. There’s a stark difference between bWAR and fWAR. I don’t like it.

[Rich]

 

Hey all, 

 

I see a lot of advanced stats flashed at Twins Daily. 

 

I was a wondering which among them you all saw as a real improvement over traditional triple crown stats for pitchers or hitters?

 

 

Good question. Advanced Sabermetrics can be complicated. I like to keep my performance evaluations simple. To evaluate hitting OPS and War work for me. For starting pitching I use War and the traditional ERA and OBA. For relief pitching I like ERA and Saves. These can be applied to Individual or team performance. I use several sites including mlb.com, FanGraphs, Baseball Reference and Baseball Alamanac.

[Otto von Ballpark]

 

The complaint was simple. There’s a stark difference between bWAR and fWAR. I don’t like it.

They measure different things. There's a stark difference between batting average and on-base percentage too.

[mlhouse]

When you think about this and the myriad  of statisitcal information available today, imagine what it was like being a math/statistics nerd and serious baseball fan reading Bill James for the first time in the early  1980s.  Almost all of these statistics were created by Bill James, although others have modified them over the decades and other concepts James imagined were created with better data.  

 

Runs created.  Range factors.  Park Adjustments. Win shares. Secondary averages.  

 

I believe that Bill James should be inducted into the Baseball Hall of Fame for his transformational work in the game.

[terrydactyls]

I got a headache just reading these posts. I felt like I was reading Solder of Fortune magazine with all the Uzi's and RAF-97's. It seems like no one is able to appreciate a player's skills any more unless they can be quantified and used to determine the best. And then argue about about it anyway because one person used some arbitrary metric that another person feels is inferior to their favorite metric.

[USAFChief]

 

They measure different things. There's a stark difference between batting average and on-base percentage too.

Actually, they purport to measure the same thing. 

[USAFChief]

 

Defensive metrics: if you don't look at WAR, or DRS, or UZR, or OAA, what do you look at?  

I'd frame the question as "If you do look at WAR, or DRS, or UZR, or OAA, what are you actually looking at?"

 

There's no real reason to think any of them are accurate, and the fact they so often wildly disagree about a player seems proof of that.

[dex8425]

 

I'd frame the question as "If you do look at WAR, or DRS, or UZR, or OAA, what are you actually looking at?"

 

There's no real reason to think any of them are accurate, and the fact they so often wildly disagree about a player seems proof of that.

I wouldn't say they disagree about a player. They measure different parts of defense. For example, OAA (new) only measures outfield catches using catch probability. OAA doesn't differentiate outfield positions either. That's why every centerfielder has a positive OAA number. They should all be better at catching than corner outfielders.  But not every centerfielder has the same UZR (not new) because UZR is positional. Some have poor arms. Some throw to the wrong base more than others. 

 

That's why I think you have to look at all of these but also understand what they mean. You can figure out what part of defense a player is poor or elite at. Some are not elite at every facet of defense. 

[Mike Sixel]

 

I got a headache just reading these posts. I felt like I was reading Solder of Fortune magazine with all the Uzi's and RAF-97's. It seems like no one is able to appreciate a player's skills any more unless they can be quantified and used to determine the best. And then argue about about it anyway because one person used some arbitrary metric that another person feels is inferior to their favorite metric.

 

I have no idea what this post means? Not one person here has said they only enjoy skills if they can be measured. Really. Not one person.

[Vanimal46]

They measure different things. There's a stark difference between batting average and on-base percentage too.

Interesting that it’s never presented that way. Writers use the higher WAR number and ignore the other one if they want to pump up a player. Vice versa if a writer wants to throw caution on a player.