It was driving me nuts

[Sports Guy]

7 hours ago, Hallas said:

So semi related, but my shake-fist-at-clouds moment with stats involves Fangraphs and other projection systems barely budging from their preseason projections.  Fangraphs ZIPS projection still has us below .500 the rest of the way and I think that is absurd.  It's plainly obvious to me that this team is far more talented than those projections gave them credit for, in the preseason, and barring a bunch of injuries I'd be absolutely floored if we only play .500 ball the rest of the way.

 

I kind of get that the O's would be bucking trends by continuing to beat these projections but I think they're just moving way too slowly.

I have no idea why people pay any attention to projections.

You take any random group of baseball fans and they likely will give you a median record similar to any projection system and they don’t do it with 1000s of simulations. That whole simulation bs is dumb to me. 

As for this team, they have been lucky because they have been pretty healthy overall. All good teams have luck in some way or another. Health has been ours, at least the biggest part of luck.

But just like 2012, the pythag record can be explained by the back end of the pen. That means everything.

[Frobby]

20 hours ago, TopGunnar said:

A lot of these advanced analytics are stupid. 

I don’t consider Pythagorean record to be an advanced stat.  It’s more of a theoretical construct that a lot of people don’t understand well.  Nobody should expect that every team is going to be within 3 games of their Pythag.   But most teams will be close.  I would say that for more than half the teams that drastically outperform their Pythag, good bullpen performance in high leverage situations is the reason.  

[Fan4Life]

Seems like the premise is that the amount of game you get blown out in, the ones you blow out other teams, the 1-run differential is considered to generally equal out... like everything in statistics, you never know what the right sample size is. .. so it doesn't always equal out over a season.  Then you have the pitching matchup variables, the team matchup variables, injuries etc..  Luck. 

[2001OriolesFan]

21 hours ago, jcaponio said:

Pythagorean record is based solely on run differential. It takes literally nothing else into account. 

Yes.  I worked in the semiconductor industry in reseach, development and manufacturing.  We spent most of our time trying to model and explain the reality of data we collected, and trying to fit it to a model, which could predict behavior and that was related to some scientific facts and understanding.  It was always a happy day, if our data could be matched to a simple scientific equation.  For example, the resolution of a lithographic system with a constant process was found to closely follow the Rayleigh Criterion" https://www.leica-microsystems.com/science-lab/life-science/microscope-resolution-concepts-factors-and-calculation/ 

In the real world, you run into the problem where the "output of the system" (how the results wind up) depends upon TWO or more physical laws.  Things just don't make sense. You can go to the extreme experiment situations and maybe tease out a start at the understanding.

The problem with the "scientific analyitics based" predictions is that the problem is very complex and there are a lot of variable in play and often the "minor" factors become the major factors and the model fails.  Just because there are numbers and graphs involved, doesn't mean that assures understanding.  "Gut feelings" can be use to see when you are being presented BS.

 

[DrungoHazewood]

21 hours ago, Can_of_corn said:

It isn't have a good record in one run games.

It's having a significantly higher winning percentage in one run games compared to other games.

If a .600 team wins .605 in one run games that's not luck.  If a .500 team does, than yea, it's probably luck.  If the Padres win at a .495 clip and are 6-18 in one runs games, it's fair to say they've been unlucky.

Is there any significant correlation between winning percentage in one-run games and in other games? It looks like there is some, but there's far more luck in one-run records than in other games. That Eddie Epstein article is probably 20 years old, but it showed that very good teams win fewer one-run games than otherwise, and bad teams more.

This piece (mostly on the '12 Orioles) shows a very weak correlation year-to-year in one-run performance, again indicating a big luck component.

So... if a .600 team wins 55% of their one-run games, and a .400 team wins 45% that's expected. If they out- or under-perform that by a significant margin that's probably due mostly to luck.

[DrungoHazewood]

1 hour ago, sportsfan8703 said:

I don’t think these projections factor in enough that we’re a young ball club that is getting better. Gunnar in April is not the Gunnar he is now. Everyone is growing and getting better. For the most part. Even Hyde. 

The Royals are a younger team than the Orioles.

[DrungoHazewood]

20 hours ago, TopGunnar said:

A lot of these advanced analytics are stupid. 

They don't think much of you, either.

[dystopia]

1 hour ago, DrungoHazewood said:

The Royals are a younger team than the Orioles.

And they have one of the worst records in the majors and one of the worst farm systems.