OaktownBear said:
dajo9 said:
concordtom said:
calbear93 said:
Well, the power of data analytics as opposed to assumptions or human analysis first became famous to regular folks due to baseball and Billy Bean. Things that people assumed were most valuable were not, and the Oakland A's were able to pay for valuable players for much less based on things that data actually showed were relevant to scoring runs, such as on base percentage and slugging percentage. And that insight did not come from some genius baseball person but from a nerdy mathematician from Harvard who barely knew what baseball was.
Right. That's my point. Just apply Billy Ball to Politics. That's what YOU were saying, no?
I don't know about AI but Billy Ball has its limitations. It clearly seems to work in June but I don't think it works in the postseason. I think the reason for that is in the postseason you are no longer playing against the average. You are playing against above average. That means your home runs per strikeout rate goes down and we end up with boring postseason games where teams strike out over and over again (using one basic metric as an example).
I believe Billy Ball may help you get into the postseason but the oldtimers have it right for winning championships. Contact. Stolen bases. Manufacturing runs. It's no surprise to me that the marquee player on 2 of the last 3 championships is a classic old school ballplayer who is known for contact, manufacturing runs, and smart baseball - Mookie Betts. The other championship team, the Washington Nationals, lost its Billy Ball hero, Bryce Harper, the previous year. In October, they were better without him. In my opinion, if your star slugger is famous for walks, you have a problem.
You mistake the principles of Moneyball.
You start with analytics. So that is finding what player traits maximize wins. That is actually not that hard to figure out. Moneyball was not about that. Moneyball's first success came because NO ONE was using analytics. Old time scouting was based on gut feeling borne out of experience. That is not to say at all that scouts were wrong about everything. But they were wrong about a lot. But guess what? Analytics would say Mike Trout and Barry Bonds, and Willie Mays are awesome players. So would scouts.
Conclusions from analytics absolutely work in the playoffs. For instance, OPS is a lot more important stat than batting average. Absolutely true in the regular season or the post season.
Moneyball isn't the championship strategy and it never was intended to be. The point is, the A's didn't have any money. (still don't). So they were looking for the traits that those that had money were not valuing. If you went back to the Moneyball era and looked at the analytics the A's were using, and asked what were the top 100 things they could do to win based on the analytics, all 100 of those line items would have screamed "SIGN BARRY BONDS!!!!!" And the next 100 would have been to sign a whole lot of guys the A's couldn't afford. The point of Moneyball wasn't to build the best team. It was to build the best team they could afford. The number one thing the A's could have done to win was spend like the Yankees. The number 2 thing they could have done was to THEN marry analytics with spending like the Yankees. But, at the end of the day, they were shopping from a list of players that no one wanted. Analytics can only take you so far. Every year the A's have made the playoffs, if given the chance they would have traded rosters with almost anyone else in the post season (assuming dollars are equal). they don't lose because the analytics are wrong. They lose because the other teams are better. The A's are buying Hondas. That is a good car. Best car they can afford. But there will always be a guy with unlimited money to spend with a better car.
The A's were never wedded to what some people have mistakenly dubbed "Moneyball". Moneyball just means whatever is undervalued at the moment. Even this year one of the color commentators was surprised that the A's weren't doing what they haven't done in 10 years. Because other teams used the same analytics and now overvalued the traits the A's were buying in the 2000's. So, the A's steal bases now (What?). The A's have the best defensive player in baseball. The A's have arguably the best defensive first baseman and center fielder (What? The A's play defense?) The A's were early on in valuing the bullpen because starting pitching was so expensive. They also go beyond the old analytics and look at how they can improve player performance. For instance, they have picked up cheap, journeyman pitchers and said, "You know that 3rd or 4th pitch that you throw 2% of the time because you think it sucks? Well, it ain't great, but when you throw it 10% of the time, your other pitches all become a lot more effective" They also worked specifically on what launch angles were most successful in Oakland on offense and got their batters to work to those. And then did the same for the road so the A's launch angle as team 2 years ago was significantly different at home than on the road.
The frequency that the A's go to the playoffs is amazing given the money they spend. Yes, they lose. But that is all about what they spend. The strategy that people said can't win in the playoffs basically immediately won when the Red Sox adopted the same strategy with a huge payroll. Saying Moneyball doesn't work in the playoffs doesn't really mean anything because the A's are constantly changing their strategy based on what they can afford. What is true is that being cheap doesn't win in the playoffs.
College sports fans should understand this. It's the old saying "It's not about the X's and O's. It's about the Jimmy's and Joe's". You can't win at the highest level without the recruits.
Fair enough that we were throwing around the phrase Billy Ball when we meant sabremetrics. Billy Ball is just so much more fun to say, but sabremetrics is more appropriate because, like the game it produces, it is deadening. So, just insert sabremetrics where I said Billy Ball.
Quote:
Conclusions from analytics absolutely work in the playoffs. For instance, OPS is a lot more important stat than batting average. Absolutely true in the regular season or the post season.
I strongly disagree with the contention that analytics work in the postseason as well as they do the regular season. Batting average and OPS both have deficiencies as a statistic. Batting average because it ignores walks and slugging. OPS because it treats walks and singles equally and overvalues slugging. Neither of them are really at the core of sabremetrics.
The key batting statistic for sabremetrics is wOBA, which applies a run producing coefficient to each method of getting on base. The run producing coefficient is based on averages of thousands of games. Regular season games. It has led to the current version of baseball we see, in which batters swing for the fences and which produces a high number of home runs, strikeouts, and walks with record low amounts of balls hit into play (boring). And it works in June against average players. Your team may strike out a lot but you can hit 3 home runs and win the game. This approach will help your team get more wins in the regular season and reach the playoffs.
But in the postseason, I believe you can throw the coefficients out the window because you are no longer playing against the average. You are playing against above average and the result is likely to be even more strikeouts and fewer home runs. Postseason sabremetric coefficients (and you can't really calculate that because every year is different and there aren't enough games to get a meaningful value in time for it to be useful) would probably bring back stolen bases, strategies for advancing runners, and opposite field hits.
Here are number of players on each recent World Series team to be in the top 100 strikeout rate (fewest strikeouts). Each year there is a wide discrepancy, the better contact team won. In 2019 it is closer, and the lesser contact team won in 7. How did the Dodgers finally win the World Series? They signed Mookie Betts and learned to make contact. I don't present this to be conclusive as it is just one stat, but I think it is evidence in support of my position.
2020
Dodgers - 6 (won in 6)
Tampa Bay - 2
2019
Nationals - 4 (won in 7)
Astros - 6
2018
Boston - 5 (won in 5)
Dodgers - 1
2017
Houston - 8 (won in 7)
Dodgers - 2
Also importantly, sabremetrics has just produced a boring game with balls not hit into play. My favorite statistic is still batting average. It tells me which player is the most exciting to watch.
A tangent - I think the walk is the most overrated stat in sabremetrics. I can't prove it (maybe this is how I will spend my retirement years) but here are my thoughts. I think the walk coefficient is likely overvalued because it is correlated with hits (a pitcher giving up walks is also likely to give up more hits). A walk to your leadoff batter is far more valuable than a walk to your cleanup hitter. This is not captured in sabremetrics. I want my leadoff batter getting on base (OBP). I want my cleanup batter swinging hard to drive in runs (batting average). If my clean up batter is up there looking to walk, we are not playing championship baseball. Don't get me started on having Aaron Judge bat 2nd in the lineup.