Revisiting Moneyball(djpardis.medium.com)
djpardis.medium.com
Revisiting Moneyball
https://djpardis.medium.com/revisiting-moneyball-074fc2435b07
62 comments
As someone for whom Moneyball bought a nice house, I can say with some confidence that I have seen analytics elevate the smartest organisations to new heights, while burying the less sophisticated under their own confirmation bias. Many in the middle do good work that nevertheless gets ignored by the relevant decision makers. The mere existence of stats guarantees you absolutely nothing.
I won the Premier League's first analytics hackathon (at Man City, 2016, IIRC), and all we did was some regression using some data that nobody had really looked at in detail before.
Back then, it was relatively easy to just look around and prod relatively available data and find new angles if you went in eyes open. If you were trying to justify spending all that money on that kid who isn't working out with you, the approach lets you dig the hole you want.
Back then, it was relatively easy to just look around and prod relatively available data and find new angles if you went in eyes open. If you were trying to justify spending all that money on that kid who isn't working out with you, the approach lets you dig the hole you want.
Moneyball highlights an inefficiency that I would think would've stopped existing sometime in the 80-90's as data driven approaches have become standard.
One of the analytics leads for the Red Sox came to Harvard to give a presentation. I asked if he could quantify the effects analytics was having compared to the conventional wisdom developed over the course of 150 years of pro baseball.
He thought that analytics was changing the probabilities of discrete events by single digits. Essentially, nobody was doing anything wrong, there were just optimizations that were/are available.
Remember that the book/movie is about the A’s, who were eliminated in the first round of the baseball playoffs.
He thought that analytics was changing the probabilities of discrete events by single digits. Essentially, nobody was doing anything wrong, there were just optimizations that were/are available.
Remember that the book/movie is about the A’s, who were eliminated in the first round of the baseball playoffs.
Each baseball game is so high variance that even a 5 or 7 game series is still largely a crap shoot. Unlike the NFL or NBA, any MLB team that makes the playoffs has a puncher's chance to with the title. It's one of the beauties of the game. (Unless you're a Dodgers fan.)
This highlights one of the things I've always loved about baseball as a sport ... and always bugged the shit out of me about college football.
Why would a game of baseball be any more high variance than an NFL game?
On the face of it, NFL’s playoff of a single game deciding forward progress seems more high variance, rather than a best of x series.
Football also seems more high variance just due to the explosive, physical nature of the game.
I wonder what the stats say about lower seeds winning the tournament for MLB vs NFL playoffs.
On the face of it, NFL’s playoff of a single game deciding forward progress seems more high variance, rather than a best of x series.
Football also seems more high variance just due to the explosive, physical nature of the game.
I wonder what the stats say about lower seeds winning the tournament for MLB vs NFL playoffs.
https://www.vox.com/videos/2017/6/5/15740632/luck-skill-spor...
Why it’s so much harder to predict winners in hockey than basketball
Why it’s so much harder to predict winners in hockey than basketball
The way I look at it is the Patriots in '85 had maybe a 1 in 10 chance to beat the Bears in the Super Bowl. And I think that's being charitable.
Meanwhile the best regular season team of my lifetime, the '01 Seattle Mariners, still only win maybe 75% to 80% of the time vs. the historically terrible 2024 White Sox.
Now put the '85 Bears against a historically inept team instead of one of the other best teams in the league that year. I'm not sure the '76 Buccaneers win 1 game out of 100.
It's just not possible to physically dominate a game in baseball the way it is in football.
Meanwhile the best regular season team of my lifetime, the '01 Seattle Mariners, still only win maybe 75% to 80% of the time vs. the historically terrible 2024 White Sox.
Now put the '85 Bears against a historically inept team instead of one of the other best teams in the league that year. I'm not sure the '76 Buccaneers win 1 game out of 100.
It's just not possible to physically dominate a game in baseball the way it is in football.
I think part of the point though was that the As were performing far better, and progressing further than you'd expect given a dramatically lower budget for players.
If you can have a competitive team filling 80% of the seats as your competitors at 50% the payroll..
If you can have a competitive team filling 80% of the seats as your competitors at 50% the payroll..
Before Moneyball there were other controversial stats, largely driven by Bill James. By and large they had the same footprint you're describing. In any particular game they were such a small effect as to be meaningless. But of the course of a season? That could be the difference between playoffs or not.
The one that pops into mind offhand was putting the "cleanup hitter" as leadoff, even though at the time "leadoff hitter" was a very specific physical archetype as was the "cleanup hitter". Yet the latter was often the best hitter on the team, and by putting them top of the batting order they'd get more at bats over the course of a season.
The one that pops into mind offhand was putting the "cleanup hitter" as leadoff, even though at the time "leadoff hitter" was a very specific physical archetype as was the "cleanup hitter". Yet the latter was often the best hitter on the team, and by putting them top of the batting order they'd get more at bats over the course of a season.
>He thought that analytics was changing the probabilities of discrete events by single digits.
That is actually huge in baseball. As an example, the player who was least likely to get a hit last year did it in 19.6% of their at bats (this typically isn't represented as a percentage and would instead be listed as .196 batting average), while the player who was most likely to get a hit did it 33.2% of the time, meanwhile the league average was 24.3%. That means "changing the probabilities of discrete events by single digits" is what separates the average player from the outliers at both the top and bottom of the talent pool.
That is actually huge in baseball. As an example, the player who was least likely to get a hit last year did it in 19.6% of their at bats (this typically isn't represented as a percentage and would instead be listed as .196 batting average), while the player who was most likely to get a hit did it 33.2% of the time, meanwhile the league average was 24.3%. That means "changing the probabilities of discrete events by single digits" is what separates the average player from the outliers at both the top and bottom of the talent pool.
Famously, Billy Bean said "his (stuff) doesn't work in the playoffs." This is because top end talent still generally wins championships. The regular season is a slog and you can do very well by always beating the teams you should beat. But in the playoffs when everyone is giving maximum focus and effort, the talent gap is much more important than in the regular season. In basketball, I think of teams like this year's Cavaliers or the Budenholzer era Hawks. These teams won a lot of regular season games, but never felt like a legitimate threat to win the championship.
That might be true in-game (though I suspect the marginal gains compound fairly strongly) but I think the numbers when applied to recruitment and avoiding costly mistakes are much more impactful.
It was just after the 90s in 2002 so not too far off. Teams were collecting more data but looking at it wrong and now it's likely much less effective because it's been brought out into the open by both the book and the movie. Though small cap teams are still paying way less per win than the big boys like the Yankees who can pay almost 3x per win.
I think a common misconception of Moneyball is that it's about analytics. The broader lesson is that people need to systematically evaluate undervalued assets in sports/business etc.
One of the interesting 'post-Moneyball' stories is when old-school scouting methods came back onto the scene. People started overvaluing the new popularized statistics, and the market advantage was to combine the analytics and traditional approach in a cost-efficient manner.
One of the interesting 'post-Moneyball' stories is when old-school scouting methods came back onto the scene. People started overvaluing the new popularized statistics, and the market advantage was to combine the analytics and traditional approach in a cost-efficient manner.
The 2014/2015 Royals capitalized on this to some degree, picking up players who didn't strike out or walk much, at a time when players who walked a lot were at a super premium.
Some of the smarter teams in the NFL seem to be figuring out that maybe running backs aren't completely fungible, as has been the mantra for a while.
Some of the smarter teams in the NFL seem to be figuring out that maybe running backs aren't completely fungible, as has been the mantra for a while.
Markets are a decentralized adaptive system, so the overall lesson is to have a process to identify what is under/over valued and adapt over time.
There is no durable thing you can simply identify as your edge in metrics that you can stick to for years.
There is no durable thing you can simply identify as your edge in metrics that you can stick to for years.
Any BJJ player will understand Moneyball.
There are things that work with a very high percentage ...and there are things that are enormously satisfying and exciting to do.
If you're interested in winning you will methodically do the "correct" things.
The problem is it's just so much fun to do a firemans carry ...
In the Moneyball narrative, the non-analytical scouts are branded as "stupid" or "thick" or even "bigoted". I see them as more human and less robotic. I bet they have more fun than Billy Beane (and the book suggests they do).
There are things that work with a very high percentage ...and there are things that are enormously satisfying and exciting to do.
If you're interested in winning you will methodically do the "correct" things.
The problem is it's just so much fun to do a firemans carry ...
In the Moneyball narrative, the non-analytical scouts are branded as "stupid" or "thick" or even "bigoted". I see them as more human and less robotic. I bet they have more fun than Billy Beane (and the book suggests they do).
As someone who has practiced and followed sports before, during, and after the revolution in analytics and optimization, I can say that “back in the day,” sports were definitely more fun, interesting, and creative, whereas now they are more regimented and boring, but individual athletes and teams are much better at winning. There is no longer much room for eccentric, flamboyant characters, “geniuses,” or charismatic competitors who, thanks to their instincts and intuition, were able to make the winning move.
But, as in business and personal relationships, finding what works best and putting it into practice tends to be more likely to succeed than simply hoping it will work--look at me, it has never been tried before!
But, as in business and personal relationships, finding what works best and putting it into practice tends to be more likely to succeed than simply hoping it will work--look at me, it has never been tried before!
Totally anecdotal, but you see in Sim Racing versus Real Racing, the best sim racers are very regimented and analytical, and whilst it's mostly true of Real Racing, the confluence of death, consequence and mechanical failure lends some advantage to "characters" that can ride their waves better than others.
In both, it's way more interesting to watch a bunch of less talented people with personality and style duke it out, than it is to watch perfect, best of the best execution go toe to toe.
In both, it's way more interesting to watch a bunch of less talented people with personality and style duke it out, than it is to watch perfect, best of the best execution go toe to toe.
In the end, apart from the nerdiest among us, we don't watch sports because we want to see technical perfection or tactical precision.
We watch them to be moved by a narrative, by the "hero" that is about to be defeated but finds a way out, for the creative solution that has rarely, or never, been tried before.
That's why we look at a short video of an unbelievable Forward 3½ Somersault combined with a Pike dive, and we say to ourselves, "that's great", before quickly switching to look again, for the 100th time, at a less technically impressive but more emotionally charged goal scored by our favorite team 25 years ago.
I rarely watch sports nowadays, also because it is difficult to get really into a team or an individual when they are younger than you are and make millions.
That's why we look at a short video of an unbelievable Forward 3½ Somersault combined with a Pike dive, and we say to ourselves, "that's great", before quickly switching to look again, for the 100th time, at a less technically impressive but more emotionally charged goal scored by our favorite team 25 years ago.
I rarely watch sports nowadays, also because it is difficult to get really into a team or an individual when they are younger than you are and make millions.
>I can say that “back in the day,” sports were definitely more fun, interesting, and creative…
I’ll agree. Was kind of a big sports fan in the 80s (nba, nfl, mlb). The characters were pretty fun. It was a little crazy but fun. Some recent documentaries/books on the Celtics I checked out reveals more things going on behind the scenes. The press had more access. The Celtics broadcaster “Johnny Most” was really not very partial to other teams, but came up with fun nicknames. One game on the radio they stopped announcing and started laughing because he lit his pants on fire with a cigarette https://en.m.wikipedia.org/wiki/Johnny_Most
I tried watching sports now, the players are so good but it lacks the fun / heart. It’s professional and big business but somehow when you don’t care it’s a lot less fun. Though I did enjoy the women’s world cut in France quite a bit.
I’ll agree. Was kind of a big sports fan in the 80s (nba, nfl, mlb). The characters were pretty fun. It was a little crazy but fun. Some recent documentaries/books on the Celtics I checked out reveals more things going on behind the scenes. The press had more access. The Celtics broadcaster “Johnny Most” was really not very partial to other teams, but came up with fun nicknames. One game on the radio they stopped announcing and started laughing because he lit his pants on fire with a cigarette https://en.m.wikipedia.org/wiki/Johnny_Most
I tried watching sports now, the players are so good but it lacks the fun / heart. It’s professional and big business but somehow when you don’t care it’s a lot less fun. Though I did enjoy the women’s world cut in France quite a bit.
Sports "analytics" is just the meta narrative bullshit of fools and a scientistic society willingly and collectively fooling itself via non-ergodic stochastic process taking the place of actual scientific progress.
The trick is to make games where doing the thing that wins the game is also the most interesting thing to see and do.
it's only a matter of time before some team owner instructs his analytics team to "build me the team that gets me the best ratings/sells the most tickets" rather than "build me the team that gets me a championship".
> The problem is it's just so much fun to do a firemans carry ...
what about grabbing a folding chair and hitting your opponent with it?
the problem is when 'entertaining' optimizes away reality.
what about grabbing a folding chair and hitting your opponent with it?
the problem is when 'entertaining' optimizes away reality.
It’s true in boxing, too. I almost never enjoy a Floyd Mayweather bout because he plays for points better than anyone. It’s effective but not entertaining.
The "highest expected value"-ification of MMA has made it uninteresting to me. Fighting styles have become so homogenized that I fight back tears of boredom, especially when my mind drifts back to the days of seeing random fighting styles like Art Jimmerson fighting Royce Gracie
Brazilian jujutsu player?
This is a postmodernist, market fundamentalist, meta narrative that has basically nothing to do with reality.
As if there is some kind of no-arbitrage condition arbitrage, to be made between humans playing baseball.
This is nothing new though. Baseball is America's pastime in a very deep way. I collected baseball cards in the 80s like I was trading stocks in another bullshit, market fundamentalist delusion.
As if there is some kind of no-arbitrage condition arbitrage, to be made between humans playing baseball.
This is nothing new though. Baseball is America's pastime in a very deep way. I collected baseball cards in the 80s like I was trading stocks in another bullshit, market fundamentalist delusion.
Game theory. Iff you're the only one in town doing it, it's a competitive advantage. As it becomes well-known, the market gets more efficient at valuing players and this specific advantage erodes away.
This was the issue with the common meme of "moneyball". It was never about optimizing the specific stats the A's originally used. The whole point was to zig when everyone else was zagging to save money.
A better example of this is Bill Belichick. He drafted multiple TEs when nobody valued the position, paid for top special teamers, like gunners (for pennies), switched to. 3-4 defense when every other team was 4-3, then back again when team copied him.
> Home runs, walks, and strikeouts now dominate baseball, with 35% of plate appearances ending without involving seven defensive players. This has reduced balls in play by 20% since 1980, creating longer games with less action
Do baseball fans ever discuss potentially changing the rules or game setup to mitigate this?
Do baseball fans ever discuss potentially changing the rules or game setup to mitigate this?
There were notable changes in 2023
> Nearly a quarter of the way into the 2023 regular season, the rule changes MLB implemented this year are continuing to have their intended effects -- namely, a quicker pace resulting in a significant reduction in overall time of games, more hits and more steal attempts.
https://www.mlb.com/news/effect-of-new-rules-early-2023
> Nearly a quarter of the way into the 2023 regular season, the rule changes MLB implemented this year are continuing to have their intended effects -- namely, a quicker pace resulting in a significant reduction in overall time of games, more hits and more steal attempts.
https://www.mlb.com/news/effect-of-new-rules-early-2023
Yes. In the past MLB lowered the pitcher’s mound and widened the strike zone to benefit hitters, and there’s been talk in recent years of doing it again, or moving back the mound. The league also recently instituted the pitch clock, which helped hitters a little.
A wider strike zone benefits pitchers, not hitters.
You’re right of course, that was a brain fart on my part.
The actual change was lowering the height of the strike zone (to help hitters.)
The actual change was lowering the height of the strike zone (to help hitters.)
Banana Ball does a lot of rules experimentation toward the aim of making baseball more fun for the fans.
Yes, Joe Sheehan (among others) has been beating the drum for a while.
If it were named anything else besides "moneyball," the take about it glorifying underpaying players wouldn't exist.
I'm not old enough to remember a pre-moneyball world, but MLB has been mitigating the effects of the “three true outcomes” problem quite well over the last few years. The pitch clock is the best thing to happen to baseball since integration.
I'm not old enough to remember a pre-moneyball world, but MLB has been mitigating the effects of the “three true outcomes” problem quite well over the last few years. The pitch clock is the best thing to happen to baseball since integration.
There's a pitch clock now? So much for the Jamie Moyer rule that "the game doesn't start til I start it."
It's been around for a few years now. Games went from averaging around 3 hours to averaging 2.5. The long tail of 3.5-4.5 hour games were basically eliminated. The only time that happens now is if you get extra innings, high pitch counts, and lots of plate appearances that don't end in outs -- in other words, longer games are long because they're good!
Occasionally an old head will grumble about it, but even the old guys tend to be pretty happy about the pitch clock.
Occasionally an old head will grumble about it, but even the old guys tend to be pretty happy about the pitch clock.
> This randomness explains why sabermetricians often view regular-season performance as a more reliable indicator of a team's true quality than its playoff results.
That makes sense statistically, but I think most fans would intuitively disagree. When your team plays great all season and chokes in the playoffs, you rarely come away feeling that the regular season was their “true quality.” Typically, such events are seen as revealing that the team wasn’t as good as it seemed, or at least not when it mattered most. There’s probably truth to both perspectives.
That makes sense statistically, but I think most fans would intuitively disagree. When your team plays great all season and chokes in the playoffs, you rarely come away feeling that the regular season was their “true quality.” Typically, such events are seen as revealing that the team wasn’t as good as it seemed, or at least not when it mattered most. There’s probably truth to both perspectives.
Moneyball is about identifying undervalued assets, and demonstrating that value in a highly compeitive arena. Luckily this was done in a popular avenue (baseball) so the lesson was spread across society.
Several years back I ended up making "moneyball for college paintball" largely inspired by a combination of Moneyball the book and Michael Lewis' article about Shane Battier [0].
That experience led me to have a couple thoughts:
- Many people think that Moneyball was about On Base Percentage. Really, it was about how to use analytics to figure out what the undervalued metric is to focus on.
- Moneyball is great on its own. When you try to do your own version of Moneyball in a sport you care deeply about, it's as if you find another "level" of the book. For example, this quote from the book (paraphrased): "You can't see the difference between a .275 and a .300 hitter. It's an extra hit every two weeks." You realize HOW MANY parts of sports are essentially invisible unless you take the stats and then analyze them.
- Another example of the above is the quote "it's a system only a rich team could afford but only a poor team looking for an edge would buy". Having worked at a lot of startups (4), I've often told this story when people are deciding whether to pay for some 3rd party tool that will have a big impact. Sometimes it's not worth it but sometimes it can really change the outcome.
- Going in to the whole process, I wasn't sure how well I could track stats in something as complicated as college level tournament paintball. I started with just a clipboard with some paper and a stopwatch. At the end of the day, that was really all I needed to get DEEP insights on how the game worked, game theory in different scenarios and which players were adding or taking away from our win percentage. A reminder of "you can just do things"
- One unexpected outcome: I got to the point where I could watch a game and know exactly what the win probabilities were as the game went on. It was like watching the World Series of Poker where you see the odds of each hand. This was both great b/c I new the next game theory optimal call or play to do. It was terrible when I knew the odds of winning a big game went below 5% (although we did occasionally win some of those too)
0 - https://www.nytimes.com/2009/02/15/magazine/15Battier-t.html
That experience led me to have a couple thoughts:
- Many people think that Moneyball was about On Base Percentage. Really, it was about how to use analytics to figure out what the undervalued metric is to focus on.
- Moneyball is great on its own. When you try to do your own version of Moneyball in a sport you care deeply about, it's as if you find another "level" of the book. For example, this quote from the book (paraphrased): "You can't see the difference between a .275 and a .300 hitter. It's an extra hit every two weeks." You realize HOW MANY parts of sports are essentially invisible unless you take the stats and then analyze them.
- Another example of the above is the quote "it's a system only a rich team could afford but only a poor team looking for an edge would buy". Having worked at a lot of startups (4), I've often told this story when people are deciding whether to pay for some 3rd party tool that will have a big impact. Sometimes it's not worth it but sometimes it can really change the outcome.
- Going in to the whole process, I wasn't sure how well I could track stats in something as complicated as college level tournament paintball. I started with just a clipboard with some paper and a stopwatch. At the end of the day, that was really all I needed to get DEEP insights on how the game worked, game theory in different scenarios and which players were adding or taking away from our win percentage. A reminder of "you can just do things"
- One unexpected outcome: I got to the point where I could watch a game and know exactly what the win probabilities were as the game went on. It was like watching the World Series of Poker where you see the odds of each hand. This was both great b/c I new the next game theory optimal call or play to do. It was terrible when I knew the odds of winning a big game went below 5% (although we did occasionally win some of those too)
0 - https://www.nytimes.com/2009/02/15/magazine/15Battier-t.html
I like the watching NFL. I started playing fantasy football a few years ago and I see a similar phenomenon. Once I got into the weeds on current and future player ability and stats, it was like a new dimension of the game opened up to me.
Is it anti-labor to not over pay Superstars and instead hire the people that get the job done?
The whole point of the moneyball movie and system is to find players who are undervalued relative to their ability to produce runs and critically continue paying them low relative to their game utility to get a good, competitive team cheap.
There's a market distortion from the rest/majority of teams incorrectly evaluating the players. Knowing that and taking advantage of it is underpaying if you believe in paying people their real worth instead of just what you can get away with paying them.
There's a market distortion from the rest/majority of teams incorrectly evaluating the players. Knowing that and taking advantage of it is underpaying if you believe in paying people their real worth instead of just what you can get away with paying them.
But like, is that "anti-labor"? The undervalued people who actually were important now being slightly more valued -- or, in the case of the storyline in the movie, having a job at all -- seems like it could even be pro-labor.
It’s neither anti- nor pro-labor; it simply exists in the context of a strong players’ union and a league that enforces constraints on team payrolls in order to promote equity between large and small market teams.
And I agree with you; the reason to ask this question is because it is something that is quoted in the article (in a screenshot, so don't try to ctrl-F for it ;P) as part of the background zeitgeist opinion on the premise.
It's a mixed bag. Over time it's probably raised the salary of these kinds of players as teams change how they evaluate player worth but initially the first couple seasons effect is to underpay players vs the value they bring to the team and that's contrary to a long held pro-labor stance is workers should be paid more in line with the value they generate for the company rather than just what the companies do pay them.