Jump to content

Sabermetrics, Moneyball, and the Rise of Quantitative Sports "Fandom"


TedE

Recommended Posts

31 minutes ago, DonRocks said:

I know this too, and I don't agree with it - we could *easily* start a separate thread for this topic. W-L records will *always* have deep meaning - even if only psychological meaning - to the average baseball fan. Sabermetrics has gone overboard, I think.

BTW, this topic *is* something I'd like to learn more about, so if you have any suggestions, I'm all ears.

The rise of quantitative sports "fandom" is worth it's own thread.  It has its use in two places, and in those scenarios has been absolutely revolutionary:

1) The front offices of professional sports teams

2) Fantasy sports

Everywhere else it's just that asshole at the bar trying to convince you that your home town hero wasn't really all that great because his lifetime BABIP or fWAR+ was merely above average.

Link to comment
Share on other sites

17 minutes ago, TedE said:

The rise of quantitative sports "fandom" is worth it's own thread.  It has its use in two places, and in those scenarios has been absolutely revolutionary:

1) The front offices of professional sports teams

2) Fantasy sports

Everywhere else it's just that asshole at the bar trying to convince you that your home town hero wasn't really all that great because his lifetime BABIP or fWAR+ was merely above average.

"Quantitative Sports 'Fandom'" is a great phrase!

Here's one existing thread on sabermetrics.

Link to comment
Share on other sites

1 hour ago, DonRocks said:

"Quantitative Sports 'Fandom'" is a great phrase!

Here's one existing thread on sabermetrics.

I come not to bury advanced statistics, but to praise them!

When it comes to being a fan of a sport I look at it this way: advanced stats cannot make you a fan, but it can enhance your appreciation of the sport.  In this day and age in order to fully understand the ins and outs or pro sports you NEED some insights into how the decisions are being made in the front office in order to make sense of the product on the field.  Pythag in baseball is remarkably accurate over a season (or, as the stat heads would day, "Duh!").  Knowing that Danny Espinosa had a really bad April and May (sub-.200, 6 HR)) makes his June surge (.300+, 9 HRs) all the better because based on past performance you literally knew he isn't that bad.  And seeing Drew Storen's repeated failure in season-defining situations is all the more heartbreaking because it leaves you wondering what extra-statistical factor (I just made that up!) drives those losses.

What I have a problem with is letting stats be the only way you access the game.  You can't cheer for stats.  Well, OK, if you are one of those people who follow sports via your fantasy teams, I guess you can cheer only for stats.  But that also means you are a horrible human being and are dead inside.

And when you go out of your way to argue that Jim Palmer is the most overrated pitcher of all time, what is the point?  Did he run over your dog?  This is the anti-thesis of fandom. It's quantitative chest-thumping.

Link to comment
Share on other sites

On 7/5/2016 at 4:27 PM, TedE said:

I come not to bury advanced statistics, but to praise them!

When it comes to being a fan of a sport I look at it this way: advanced stats cannot make you a fan, but it can enhance your appreciation of the sport.  In this day and age in order to fully understand the ins and outs or pro sports you NEED some insights into how the decisions are being made in the front office in order to make sense of the product on the field.  Pythag in baseball is remarkably accurate over a season (or, as the stat heads would day, "Duh!").  Knowing that Danny Espinosa had a really bad April and May (sub-.200, 6 HR)) makes his June surge (.300+, 9 HRs) all the better because based on past performance you literally knew he isn't that bad.  And seeing Drew Storen's repeated failure in season-defining situations is all the more heartbreaking because it leaves you wondering what extra-statistical factor (I just made that up!) drives those losses.

What I have a problem with is letting stats be the only way you access the game.  You can't cheer for stats.  Well, OK, if you are one of those people who follow sports via your fantasy teams, I guess you can cheer only for stats.  But that also means you are a horrible human being and are dead inside.

And when you go out of your way to argue that Jim Palmer is the most overrated pitcher of all time, what is the point?  Did he run over your dog?  This is the anti-thesis of fandom. It's quantitative chest-thumping.

This is a great post, Ted, and I'm with you. 

Transferring sabermetrics to a sport I better understand - tennis - there is *so much to be gained* from the use of sabermetrics in analyzing tennis players, and there is a *fortune* to be made in developing them. 

It used to be things like titles, Grand Slam victories, aces, double-faults, etc., and these are *all* fine, but if you're playing someone with $100,000 on the line tomorrow, wouldn't you want to know if that person missed a disproportionate number of backhand approach shots in the 3rd set? Or hits 20% more down-the-line forehand winners (compared to cross-court forehand winners) than the average player on the tour? Or loses 65% of all points played when the stroke-count gets above 20? There is *so much* of value here in this high-dollar sport - if someone out there sees this post, and wants a whiz-kid programmer (who, granted, is no longer a kid), but truly understands statistics, programming, and the game of tennis ... get in touch with me - I would *covet* working on such a system.

Even though the odds of anyone ever seeing this are about zero, and the data collection from previous matches would be *absolutely brutal*.

Link to comment
Share on other sites

15 minutes ago, DonRocks said:

This is a great post, Ted, and I'm with you. 

Transferring sabermetrics to a sport I better understand - tennis - there is *so much to be gained* from the user of sabermetrics in analyzing tennis players, and there is a *fortune* to be made in developing them. 

It used to be things like titles, Grand Slam victories, aces, double-faults, etc., and these are *all* fine, but if you're playing someone with $100,000 on the line tomorrow, wouldn't you want to know if that person missed a disproportionate number of backhand approach shots in the 3rd set? Or hits 20% more down-the-line forehand winners (compared to cross-court forehand winners) than the average player on the tour? Or loses 65% of all points played when the stroke-count gets above 20? There is *so much* of value here in this high-dollar sport - if someone out there sees this post, and wants a whiz-kid programmer (who, granted, is no longer a kid), but truly understands statistics, programming, and the game of tennis ... get in touch with me - I would *covet* working on such a system.

Even though the odds of anyone ever seeing this are about zero, and the data collection from previous matches would be *absolutely brutal*.

It's a data problem.  The stats you are looking to build off of may not exist.  But I guarantee one thing: in the 21st century if there is money to be made using quantitative math somebody is already looking at how to exploit it.  Some sports are more ripe for the picking, and as soon as somebody finds the key it's essentially over.  Sports books will be the first to know and the "market" will figure it out and adjust, which leads people to search for deeper ways to understand the game, which in turn leads to better record keeping, analysis and more complex math.  

Baseball is perhaps unique in pro sports in two ways:

- The building blocks of the statistical revolution go back decades.  With access to box scores containing the outcomes of individual match ups you can run predictive models on a HUGE number of permutations and see how historical outcomes match up, as well as distill advanced stats out of those building blocks and run further models off of those.  Hit on a promising correlation, refine, re-analyze.  I don't think the average fan really gets how eerily stats can predict the game on a season-long basis, or the arc of a player's career barring serious injury.

- The team game is an aggregate collection of individual performances, more so than anything else.  When a pitcher faces a batter the performance of those two individuals are the greatest determinant of the outcome.  Other factors like the defense behind the pitcher, how many men are on base and where, the weather, etc. do have influences, but nothing that approaches what pitcher X does when faced with a batter like Y.  Sports like hockey and basketball are starting to catch on, but it's harder to isolate the variables enough to exploit them.  I'm honestly not sure what has been attempted with football.

I personally find the historic revisionism of whether historic greats of the game really were that "great" to be boring.  That use of the stats doesn't seem very "fan"-like and doesn't serve an ongoing purpose for me.  It's a side hobby that doesn't inform the current game, but I guess it's something people enjoy arguing about.  

Link to comment
Share on other sites

21 hours ago, TedE said:

It's a data problem.  The stats you are looking to build off of may not exist.

They don't exist; but they *will* exist - there are films of the matches, and it will be difficult - but not impossible - for a computer to analyze them.

21 hours ago, TedE said:

I don't think the average fan really gets how eerily stats can predict the game on a season-long basis, or the arc of a player's career barring serious injury.

I don't think so either - are there any articles you can cite that people can educate themselves with?

When's the last time a Presidential race wasn't over long before it began? (Well, maybe 2000, but we've come a *long* way since then ... dimpled chads?!)

What we need right now are good, easily seen films of matches (not just tennis), and sometime in the future, technology will be able to analyze them. It's like holding onto fossils, or artefacts of bygone eras - one day, we'll understand what they are due to advanced technology - we're freeing people who have life sentences right now due to DNA evidence (you just know there are some aging racist judges hoping that things don't come back to bite them in their lifetime). Remember that porn site on usenet you used to post on? There's only one problem with it:

It exists.

And when I see a T-Rex walking down my street, I'm moving to Antarctica.

Link to comment
Share on other sites

Here's a statistical thing that bugs me. During the local broadcast of Nats games, the announcer will often say things like, "Espinosa is one for three against Syndergaard in career matchups...".

How is that statistically significant?   :huh:

Link to comment
Share on other sites

3 hours ago, Al Dente said:

Here's a statistical thing that bugs me. During the local broadcast of Nats games, the announcer will often say things like, "Espinosa is one for three against Syndergaard in career matchups...".

How is that statistically significant?   :huh:

It's not; they have to avoid gaps in airtime. 

Link to comment
Share on other sites

20 hours ago, DonRocks said:

They don't exist; but they *will* exist - there are films of the matches, and it will be difficult - but not impossible - for a computer to analyze them.

I don't think so either - are there any articles you can cite that people can educate themselves with?

When's the last time a Presidential race wasn't over long before it began? (Well, maybe 2000, but we've come a *long* way since then ... dimpled chads?!)

What we need right now are good, easily seen films of matches (not just tennis), and sometime in the future, technology will be able to analyze them. It's like holding onto fossils, or artefacts of bygone eras - one day, we'll understand what they are due to advanced technology - we're freeing people who have life sentences right now due to DNA evidence (you just know there are some aging racist judges hoping that things don't come back to bite them in their lifetime). Remember that porn site on usenet you used to post on? There's only one problem with it:

It exists.

And when I see a T-Rex walking down my street, I'm moving to Antarctica.

Bill James' Baseball Prospectus is the genesis and is still the grandaddy of them all.  I've only ever skimmed the very, very surface of what is contained there, but it has lots and lots of background reading.  If you are a Nationals fan in particular read Tom Boswell, and especially his weekly Monday chats.  He usually drops a gem or two in there on some obscure stat he's dug up to make sense of something that's piqued his interest. 

As mentioned above baseball stats play out over a long time period.  They are actually pretty terrible at predicting immediate outcomes and don't even apply very well to individual at bats.  If the announcer says a batter is 6-for-10 lifetime with 3 HRs against a particular pitcher does that mean that the probability of the at bat producing a hit is .600, and a home run .300?  Of course not.  10 at bats is a very small sample size, and against one pitcher it artificially extracts a non-random sample from the pool.  If that hitter has a lifetime .300 BA there are probably some pitchers he's faced where he's gone 0-for-10.  And the pitcher does not have an aggregate .600 batting average against. With 1,000 at bats between the two the resulting batting average would be a close approximation of the aggregate average for the hitter against those types of pitchers (say LH, 70% fastball touching 95) and the pitcher against those types of batters (RH, good power to the opposite field).  It's why when you hear an announcer say "Well, Bob, this is an interesting choice.  They've brought Johnson in to pitch hit, but last year he was 0-for-10 against this guy", you shouldn't pay attention.  A good manager these days doesn't care, he pays attention to the fact that his pinch hitter is 15-for-40 all time with RISP against RH relievers. That being said, there are innumerable factors at play that are not quantifiable, but the surprising outcome of all of this is work is how many ARE.

So, back to tennis (and it looks like somebody has the leg up on you).  The question to ask is how well those stats will predict individual match ups.  Those match ups contain lots of individual outcomes (how many points in an average men's match?  300?  I really don't know).  How much weight do you give to winning a point?  My completely amateur opinion about tennis from a distance is that the game is too predictable these days.  That may just be a talent gap, though.  It seem that the difference between the 1-10 ranked players and say the players ranked 30-40 is really, really big.  Upsets are a big deal, and with single elimination that's it!  A relatively small deviation in the expected outcome means that somebody goes home.  Random events can have an outsized influence.  This is also why baseball regular season results are shockingly bad at predicting playoff success.  It make some sort of perverse sense that the sport with the best predictive models for the long term is so bad at predicting outcomes in the short term, but I still can't fully wrap my head around it.

These models will only get better and better.  Eventually some NFL coach is going to buck the stubborn conservative thinking that permeates that sport and just fucking go for it on 4th down a lot.  Knowing what a player is "worth" makes it easy to continue to cheer for a guy like Ian Desmond even after he failed to produce for the Nats.  On the flip side it takes a lot of the mystery out of the game.  It's a good thing that nobody has come up with a way to quantify personality.

Link to comment
Share on other sites

1 hour ago, TedE said:

So, back to tennis (and it looks like somebody has the leg up on you).  The question to ask is how well those stats will predict individual match ups.  

These models will only get better and better.  Eventually some NFL coach is going to buck the stubborn conservative thinking that permeates that sport and just fucking go for it on 4th down a lot.  

It's funny you say both of these things. Not "ha-ha" funny but "interesting" funny.

In the first paragraph, you're thinking of the stats like a bettor; I'm looking at them more from a player's point of view: "If this guy gets five backhands in a row, he has a 72% chance of losing the point over the course of his career." But as with all sports, there is a wide breadth of utility for good statistics - that's why I think there's such a promising future for them.

When I was playing tournaments, I had about a one-year period, before I got badly injured, where I embarked on my own form of "non-statistical sabermetrics" based on gut-feel alone: I would hit two first serves, on almost every single point. As a result, I double-faulted much more than average, but I also hit a lot of aces and won a lot of easy points on my second serve. It's like playing blackjack: you can't look at individual hands (or, in stock-market terms, you can't look at day-by-day results), or you'll go insane; I just turned my mind off and *did it*, and in my opinion, it worked in the long run - I was starting to take down some players who were better than me. It took *serious stones* to use this strategy in tournaments, and people thought I was reckless and crazy. A modified version of this: I did it on the very first point of the match (a hard, flat, second-serve right down the middle, which would almost always result in an ace if I hit it in), and had my opponent thinking "WTF?!" for the next several games, making my normal, kick second-serve that much more effective since he'd be looking for the fastball. A variation of this was that I'd use a second serve as my first serve, and a first serve as my second serve, which also wreaked some mental havoc on the other side of the net.

As for someone having a leg up on me? Ain't no way I've got the time - I'm too busy running this website!

Link to comment
Share on other sites

On July 6, 2016 at 6:45 PM, DonRocks said:

It's funny you say both of these things. Not "ha-ha" funny but "interesting" funny.

In the first paragraph, you're thinking of the stats like a bettor; I'm looking at them more from a player's point of view: "If this guy gets five backhands in a row, he has a 72% chance of losing the point over the course of his career." But as with all sports, there is a wide breadth of utility for good statistics - that's why I think there's such a promising future for them.

Stats are stats.  Somebody will find a way to use them to their advantage.  Bettors just make different use of them; their only concern is for the outcome of the combined probabilities of the match ups.  Nobody (I hope!) is betting on the outcome of an at bat.

Speaking of gambling, the widespread availability of stats in general (not just derivative advanced stats, but free distribution of raw numbers) gave rise to the popularity of fantasy sports. The natural evolution of that pursuit is daily fantasy sports.  These companies are like the lottery in that they basically print money from mathematical ignorance.  But with the lottery at least you know you really do have the same shitty chance as anybody else willing to part with their $1 for a ticket.  With daily fantasy the illusion is that a sports fan with deep knowledge of players and strategy has an advantage over somebody who knows "less" about the game.  Nope.  All of that money (minus the companies' cut) is funneling into the hands of an exceedingly small minority of players with specialized knowledge of game theory who would otherwise probably be plying their trade creating algorithms for Wall Street.  It's an arbitrage game, and without a LOT of capital and the ability to spread it out over a large enough sample size you are throwing money away in the long term.

Link to comment
Share on other sites

On August 14, 2016 at 11:04 AM, DonRocks said:

Look at Novak Djokovic's page (go there, and click on "Point Outcomes by Rally Length"): Out of 25,000+ points played, he has only won 53% of them; yet, he's the undisputed #1 player in the world right now. 

Screenshot 2016-08-14 at 11.06.15.png

This is interesting.  You would think that seemingly dominant players would have a more skewed distribution when they win, e.g. they win dominantly more often (winning 6-0/6-1/6-2 sets) than when they lose (more close outcomes), leading to a point win percentage larger than recorded there. I only skimmed that page and had a hard time determining the range that it represents (whole career vs. recent dominance, I don't believe that is EVERY professional match of his since it relies on volunteer assessment of recorded matches).  But, if I'm reading it correctly, that 53% overall points won is backed by "critical" point win percentages all well over 60%.   This equates to those stats you see for baseball players' percentages with bases loaded, or RISP with 2 outs, etc.  In other words, the "clutch" stats.  Maybe there is something to that, and Moneyball ain't everything.

Link to comment
Share on other sites

15 minutes ago, silentbob said:

I don't follow tennis nearly as closely as I did 25 or 30 years ago, but FiveThirtyEight (based on skimming their articles) seems to engage in fairly quantitative analyses of the sport.

Funny, I just saw this article today:

"Key to Winning at the U.S. Open Is in the First Four Shots" by Craig O'Shannessy on nytimes.com

Link to comment
Share on other sites

On 7/5/2016 at 4:48 PM, DonRocks said:

Transferring sabermetrics to a sport I better understand - tennis - there is *so much to be gained* from the use of sabermetrics in analyzing tennis players, and there is a *fortune* to be made in developing them. 

It used to be things like titles, Grand Slam victories, aces, double-faults, etc., and these are *all* fine, but if you're playing someone with $100,000 on the line tomorrow, wouldn't you want to know if that person missed a disproportionate number of backhand approach shots in the 3rd set? Or hits 20% more down-the-line forehand winners (compared to cross-court forehand winners) than the average player on the tour? Or loses 65% of all points played when the stroke-count gets above 20? There is *so much* of value here in this high-dollar sport - if someone out there sees this post, and wants a whiz-kid programmer (who, granted, is no longer a kid), but truly understands statistics, programming, and the game of tennis ... get in touch with me - I would *covet* working on such a system.

Even though the odds of anyone ever seeing this are about zero, and the data collection from previous matches would be *absolutely brutal*.

Click

On a completely different note, one potential emerging trend that's been happening in baseball this year is the abandonment of a traditional starting pitcher.

The Rays, well known for their adherence to analytics, have started Sergio Romo (a long-time set-up man for the Giants and key to their bullpen in three WS title runs) in the first inning of certain games where the opposing team's batting line-up is top-heavy with right-handed hitters.  The Angels, with Mike Trout, are one such team.  This makes some strategic sense, because Romo has a heavy platoon split -- his career OPS against RH batters is a staggeringly low .561 (compared to a decent but hardly dominant .687 against LH batters).  If a manager would do something like this in the 7th or 8th inning, why not in the 1st?  That said, Trout is one of the few MLB hitters with a reverse-platoon split -- his career OPS is 50 points higher against RHPs.

Zack Cozart of the Angels has already been critical of this practice, saying that you would only see something like this in spring training.  His complaint?  "I don't think that's good for baseball, in my opinion. It's definitely weird, not knowing who you're going to face in your first couple of at-bats. … Usually, you have a starter and you think you're going to have three at-bats probably. So you're going to use the first at-bat and you want to have success, see what he has if you haven't faced him before, stuff like that."  You know what, Zack?  It's not the other team's job to make life easier for you.  They're trying to win too.

This development should also come as no surprise because the Rays started experimenting with "bullpen days" a few years ago.

It should also come as no surprise that the Dodgers have followed suit and used one of their relievers in the first inning last Friday.  Andrew Friedman, who used to head the front office in Tampa, is President of Baseball Operations for the Dodgers.  And only 41 years old!  Data-minded people like him and Theo Epstein are examples of why more major pro sports franchises are hiring "kids" to run their front offices.

I'll be curious to see if this "opener" practice starts to grow and become ubiquitous in the same way that infield shifts have become in recent years.  Baseball is a sport full of unspoken rules and taboos.  Who knows where the so-called traditionalists view something like this on their acceptability spectrum.

  • Thanks 1
Link to comment
Share on other sites

33 minutes ago, silentbob said:

Click

On a completely different note, one potential emerging trend that's been happening in baseball this year is the abandonment of a traditional starting pitcher.

The Rays, well known for their adherence to analytics, have started Sergio Romo (a long-time set-up man for the Giants and key to their bullpen in three WS title runs) in the first inning of certain games where the opposing team's batting line-up is top-heavy with right-handed hitters.  The Angels, with Mike Trout, are one such team.  This makes some strategic sense, because Romo has a heavy platoon split -- his career OPS against RH batters is a staggeringly low .561 (compared to a decent but hardly dominant .687 against LH batters).  If a manager would do something like this in the 7th or 8th inning, why not in the 1st?  That said, Trout is one of the few MLB hitters with a reverse-platoon split -- his career OPS is 50 points higher against RHPs.

Zack Cozart of the Angels has already been critical of this practice, saying that you would only see something like this in spring training.  His complaint?  "I don't think that's good for baseball, in my opinion. It's definitely weird, not knowing who you're going to face in your first couple of at-bats. … Usually, you have a starter and you think you're going to have three at-bats probably. So you're going to use the first at-bat and you want to have success, see what he has if you haven't faced him before, stuff like that."  You know what, Zack?  It's not the other team's job to make life easier for you.  They're trying to win too.

This development should also come as no surprise because the Rays started experimenting with "bullpen days" a few years ago.

It should also come as no surprise that the Dodgers have followed suit and used one of their relievers in the first inning last Friday.  Andrew Friedman, who used to head the front office in Tampa, is President of Baseball Operations for the Dodgers.  And only 41 years old!  Data-minded people like him and Theo Epstein are examples of why more major pro sports franchises are hiring "kids" to run their front offices.

I'll be curious to see if this "opener" practice starts to grow and become ubiquitous in the same way that infield shifts have become in recent years.  Baseball is a sport full of unspoken rules and taboos.  Who knows where the so-called traditionalists view something like this on their acceptability spectrum.

I love what the Rays are doing here.  And I'm equally annoyed at the traditionalist response to things like this.  Teams should be doing more of it.  You want to play an extreme shift on a lefty pull hitter with nobody on?  Take the easy bunt single!

I think baseball players and managers will come around eventually, but it will take time (I'm convinced that the NFL will never learn, which is OK because I've stopped watching).  I was at one of the Os/Nats games last week right behind the plate and when Bryce came up with nobody on and a gaping hole at 3rd base it took a lot of willpower not to yell "Bunt the fucking ball!" at him.  I get it, he wants to be Bam Bam most of the time, but a hit is a hit and a lead-off runner is the most valuable thing in the game.

Link to comment
Share on other sites

This just popped into my head; I'm not sure why.

"OPS" (On-Base Percentage plus Slugging Percentage) is often thought of as "the ultimate offensive statistic," and if you look at the career OPS leaders, you can see why, but I think this statistic is terribly flawed.

* It assumes (via its simple addition) that On-Base Percentage is essentially equal in value to Slugging Percentage.

* It accounts for "Batting Average" twice in the same statistic (Batting Average is included as a component in both figures).

It seems to me that "the ultimate offensive statistic" should consider Batting Average + Walks + Extra Base Hits (with a multiplier for doubles, triples, and home runs) + Other ("Other" being sacrifice bunts, sacrifice flies, bases reached on error, etc.), with each of these weighted accordingly.

This way, each individual statistic is only accounted for once. As to how each of these components should be weighted, I don't know. 

There are other factors that would need to be considered - for example, if a speedster changes the way a play transpires, that needs to be considered somehow.

Also, defense is just as important as offense. It just is. Maybe more so.

Link to comment
Share on other sites

Forgive me if I am missing something, it is late in the day for me, and most brain cells have been used to capacity.  I do not see how/where batting average is accounted for twice in OPS, in that, as you say, batting average is "included as a component" in both figures.  Neither includes batting average, per se, in determining the value.  Batting average could be said to possibly have a correlation to OBP and SP, in the sense that, the higher the batting average, the higher the "starting point" for each may be.  But, for example, OBP is not batting average plus something else, it is number of times reaching safely over the total number of at bats.  Batting average uses neither of these.

As to your point about OBP, I doubt that any team's data analysis group these days uses that and that alone as "the ultimate offensive statistic", and instead each have slightly differing views as to what stats are viewed as most important.  It wouldn't surprise me, either, if one of those 30 teams is using something akin to your proposed "ultimate offensive statistic" either instead of, or in addition to OPS in valuing offensive output of players.

Link to comment
Share on other sites

28 minutes ago, DonRocks said:

This just popped into my head; I'm not sure why.

"OPS" (On-Base Percentage plus Slugging Percentage) is often thought of as "the ultimate offensive statistic," and if you look at the career OPS leaders, you can see why, but I think this statistic is terribly flawed.

* It assumes (via its simple addition) that On-Base Percentage is essentially equal in value to Slugging Percentage.

* It accounts for "Batting Average" twice in the same statistic (Batting Average is included as a component in both figures).

It seems to me that "the ultimate offensive statistic" should consider Batting Average + Walks + Extra Base Hits (with a multiplier for doubles, triples, and home runs) + Other ("Other" being sacrifice bunts, sacrifice flies, bases reached on error, etc.), with each of these weighted accordingly.

This way, each individual statistic is only accounted for once. As to how each of these components should be weighted, I don't know. 

There are other factors that would need to be considered - for example, if a speedster changes the way a play transpires, that needs to be considered somehow.

Also, defense is just as important as offense. It just is. Maybe more so.

It accounts for singles twice.  Perhaps adding walks to the numerator and denominator would do the same thing (but I think they actually value the dispersion that results from the current method).  As for the 'Other', I think these have fallen out of favor not because they're not valuable but because they rely as much on someone else's work (getting on base ahead of you) as your own.  That external effort is not available equally across the batting order and would influence the stats.

And I'll be sure to tell every kid that gets put in right field that defense is just as important as batting.😉

That being said, I'm sure that none of this is what they're actually measuring players on today, rather I'm guessing there are machine-learning algorithms that are directly interpreting video footage of the hits somehow.  Maybe I'm wrong, though.

Link to comment
Share on other sites

33 minutes ago, DonRocks said:

This just popped into my head; I'm not sure why.

"OPS" (On-Base Percentage plus Slugging Percentage) is often thought of as "the ultimate offensive statistic," and if you look at the career OPS leaders, you can see why, but I think this statistic is terribly flawed.

* It assumes (via its simple addition) that On-Base Percentage is essentially equal in value to Slugging Percentage.

* It accounts for "Batting Average" twice in the same statistic (Batting Average is included as a component in both figures).

It seems to me that "the ultimate offensive statistic" should consider Batting Average + Walks + Extra Base Hits (with a multiplier for doubles, triples, and home runs) + Other ("Other" being sacrifice bunts, sacrifice flies, bases reached on error, etc.), with each of these weighted accordingly.

This way, each individual statistic is only accounted for once. As to how each of these components should be weighted, I don't know. 

There are other factors that would need to be considered - for example, if a speedster changes the way a play transpires, that needs to be considered somehow.

Also, defense is just as important as offense. It just is. Maybe more so.

I'll agree with your concept that there has to be a more refined measure of OPS. One thing OPS misses by a mile is whether or not the hitter is having a hot streak. I'm a big fan of OBP over batting average, and I like Slugging Percentage as a standalone stat because of how it rewards total bases, but I don't think adding them together creates a true measure.

However, I'll hand it to OPS -- if you look at the top OPS lists, you'll find that the best hitters are always represented, and nobody who can rake is missing from the list.

Don't get me started on WAR or fWAR....!

Link to comment
Share on other sites

It's far, far more reflective of a player's offensive value than any of the traditional statistics, but no serious observers think of it as "the ultimate offensive statistic."  It's been superseded (or complemented, more accurately) by far more comprehensive and context-sensitive metrics, including various iterations of Wins Above Replacement (WAR) and other stats.

Link to comment
Share on other sites

18 hours ago, Marty L. said:

It's far, far more reflective of a player's offensive value than any of the traditional statistics, but no serious observers think of it as "the ultimate offensive statistic."  It's been superseded (or complemented, more accurately) by far more comprehensive and context-sensitive metrics, including various iterations of Wins Above Replacement (WAR) and other stats.

Aha! Maybe *you* can explain Wins Above Replacement to me (in that thread), because it seems like a load of hooey, especially on defense (and how defense is compared to offense).

Link to comment
Share on other sites

On 6/3/2019 at 4:14 PM, DonRocks said:

It seems to me that "the ultimate offensive statistic" should consider Batting Average + Walks + Extra Base Hits (with a multiplier for doubles, triples, and home runs) + Other ("Other" being sacrifice bunts, sacrifice flies, bases reached on error, etc.), with each of these weighted accordingly.

Does Runs Created get closer to what you're looking for? There are different versions of it, one of which also includes stolen bases.  The "technical" version seems closest to what you describe. It's described in numerous places. Here's Wikipedia.

Rendon and Soto lead the Nats in Runs Created. A few days ago Soto was ahead, but now it's Rendon. The version ESPN uses has Rendon farther ahead than whatever the Washington Post draws from.  Rendon, Kendrick, and Soto lead the team in OPS, but Kendrick is lower in RC.

Link to comment
Share on other sites

On 6/7/2019 at 1:19 PM, Pat said:

Does Runs Created get closer to what you're looking for? There are different versions of it, one of which also includes stolen bases.  The "technical" version seems closest to what you describe. It's described in numerous places. Here's Wikipedia.

Rendon and Soto lead the Nats in Runs Created. A few days ago Soto was ahead, but now it's Rendon. The version ESPN uses has Rendon farther ahead than whatever the Washington Post draws from.  Rendon, Kendrick, and Soto lead the team in OPS, but Kendrick is lower in RC.

Does Runs Created account for a runner on 2nd, none out, and the right-handed batter hits a ground-out to 2nd base, allowing the runner to advance to 3rd? That's a valuable piece of hitting in a close ballgame.

And! You gave me a chance to bring up Brooks again with this very example:

Bottom of the 3rd, Orioles leading 5-3, none out, runner on 2nd, Brooks Robinson up, 1-1 count, Wayne Granger pitching - Gowdy: "Against Minnesota [in the ALCS], he had 7 hits in 12-times up; in the World Series, he's 8-for-17 - 15 hits in his last 29 times at-bat, a nine-game batting streak. He's been just as hot with his glove." Robinson laces a line drive to 2nd-baseman Tommy Helms, who bobbles the ball, but Robinson hit it so hard (and he ran so slowly) that he was still thrown out at first - this was very close to being a base hit, and was a fine, heads-up recovery by Helms - that's one thing you must do as an infielder: block the ball any way you can, so you can have a second chance to throw the runner out. Rettenmund, on 2nd base, advanced to 3rd on the ground out. Gowdy [on Robinson going to the opposite field]: "Those are things that never show up in the box score, but the good players do them, and help their team. Rettenmund now at 3rd, and the Reds have to bring the infield in with one out." Rettunmund would score on the next at-bat, as Davey Johnson hit a line-drive single to left-field; had Rettunmund remained at 2nd base, he wouldn't have made it home, and it was all set up by Robinson hitting to the opposite field - what you'd call a "productive out." Gowdy: "And Brooks Robinson, getting that man over to 3rd, set up the run." 

Video of the play

Also, does Defensive Wins above Replacement (dWAR) account for this situation with Robinson at 3rd base? From newsday.com:

“Gil came out of the dugout and said, ‘Hey, let’s change our strategy. Let’s bunt,’ ’’ Martin said in May from his home in Matthews, North Carolina.

Martin was not thrilled. He didn’t want to bunt, but he said he told Hodges,“You’re the manager. I’ll do what you say.’ ”

Hodges told Martin to avoid bunting toward Brooks Robinson, Baltimore’s Gold Glove third baseman. On a 1-and-1 pitch, Martin squared and dumped the ball a few feet beyond the home plate circle onto the turf toward first.

Link to comment
Share on other sites

1 hour ago, Pat said:

Read this if you dare: How Is WAR Calculated, Really? Breaking Down A Single Play To Find Out.

Quite an impressive work of research and analysis. The differences between the main entities that calculate their stats from the same data is pretty fascinating.

Before I read this, and before I scroll backwards through this post, I'm going on-record as calling WAR B.S., and especially dWAR.

(I do think that the concept is somewhat sound; I don't think the formulas and algorithms to create the end-product are worth a warm bucket of spit.)

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...