More hilarious than ever: PECOTA projects the Orioles at 71-91

[joeyc]

On 2/7/2017 at 9:40 AM, Aristotelian said:

I don't see anything hilarious about that. The margin between a 70 win team and a 90 win team is pretty small. The Orioles have lately overperformed. It is possible that given enough chances their model will eventually predict us correctly. I had us as a 70 win team last year and was pleasantly surprised, and we have pretty much tread water this year. I would probably be more surprised by 90 wins than I would 70 wins this year. That said, they must really hate our SPs. They are basically predicting Tillman and Gausman to both have career-worst seasons, with Bundy pitching like a #5 and both Ubaldo and Miley below their norms. At least one of those guys has to step up, right?

I agree 100%. The Redsox and Yankees, with the young talent they acquired through the trades and the off season moves they made are set to  lead this division. I think we fight for last place with the Rays. Unless Bundy Gausman and Tillman turn into Greg Maddox, Tom Glavine and John Schmoltz overnight, we have not gotten better. We have the same team practically as last year. They, on the other hand have improved immensely.

[Frobby]

On ‎2‎/‎7‎/‎2017 at 9:40 AM, Aristotelian said:

I don't see anything hilarious about that. The margin between a 70 win team and a 90 win team is pretty small. The Orioles have lately overperformed. It is possible that given enough chances their model will eventually predict us correctly. I had us as a 70 win team last year and was pleasantly surprised, and we have pretty much tread water this year. I would probably be more surprised by 90 wins than I would 70 wins this year.

 

1 hour ago, joeyc said:

I agree 100%. The Redsox and Yankees, with the young talent they acquired through the trades and the off season moves they made are set to  lead this division. I think we fight for last place with the Rays. Unless Bundy Gausman and Tillman turn into Greg Maddox, Tom Glavine and John Schmoltz overnight, we have not gotten better. We have the same team practically as last year. They, on the other hand have improved immensely.

These takes seem absurdly pessimistic to me.    It's more likely that we'll lose 19 more games than last year than that we'll win one more game last year?    I think a realistic pessimistic take on the O's would be that we'll be a .500ish team.   70 wins is way beyond the pale.    As to the Yankees, I think they have set themselves up well for the long haul (especially when they spend up the wazoo in the 2018-19 offseason), but it will surprise me if they overtake the Orioles in 2017.    They did a nice job of acquiring young talent, but most of that talent isn't ready yet and/or is very inexperienced at the major league level.  It will take time for those guys to be above average major leaguers.

[Aristotelian]

2 hours ago, Frobby said:

 

These takes seem absurdly pessimistic to me.    It's more likely that we'll lose 19 more games than last year than that we'll win one more game last year?    I think a realistic pessimistic take on the O's would be that we'll be a .500ish team.   70 wins is way beyond the pale.    As to the Yankees, I think they have set themselves up well for the long haul (especially when they spend up the wazoo in the 2018-19 offseason), but it will surprise me if they overtake the Orioles in 2017.    They did a nice job of acquiring young talent, but most of that talent isn't ready yet and/or is very inexperienced at the major league level.  It will take time for those guys to be above average major leaguers.

I am not predicting 70, I said I would not be surprised by 70. I would put us at .500 +/- 10 games. I don't see Trumbo repeating a career year, and other than him we are basically the same team. Losing Pedro but gaining Smith. Unfortunately the current roster still has Trumbo playing defense at least some of the time, and Smith is not a huge upgrade.

Gallardo is addition by subtraction, but then we may need innings from Wilson, Wright, or Gunkel. Better wear protective headgear on Eutaw St if that is the case.

I disagree with Joeyc that we will be fighting for last place, but I would put us in the middle of the pack with TOR and NYY fighting for 2nd through 4th. 

[Frobby]

36 minutes ago, Aristotelian said:

I am not predicting 70, I said I would not be surprised by 70. I would put us at .500 +/- 10 games. I don't see Trumbo repeating a career year, and other than him we are basically the same team. Losing Pedro but gaining Smith. Unfortunately the current roster still has Trumbo playing defense at least some of the time, and Smith is not a huge upgrade.

Gallardo is addition by subtraction, but then we may need innings from Wilson, Wright, or Gunkel. Better wear protective headgear on Eutaw St if that is the case.

I disagree with Joeyc that we will be fighting for last place, but I would put us in the middle of the pack with TOR and NYY fighting for 2nd through 4th. 

Yes, I realized you weren't predicting 70, but to say 70 is more likely than 90 is a pretty bold statement about a team that won 89 games last year.    I don't think you and I are far off on the "mid-point" result for the team (you say 81, I'd say 84) but I think the odds of 90 wins, though not that high, are much better than 70.   

[Number5]

47 minutes ago, Aristotelian said:

I am not predicting 70, I said I would not be surprised by 70. I would put us at .500 +/- 10 games. I don't see Trumbo repeating a career year, and other than him we are basically the same team. Losing Pedro but gaining Smith. Unfortunately the current roster still has Trumbo playing defense at least some of the time, and Smith is not a huge upgrade.

Gallardo is addition by subtraction, but then we may need innings from Wilson, Wright, or Gunkel. Better wear protective headgear on Eutaw St if that is the case.

I disagree with Joeyc that we will be fighting for last place, but I would put us in the middle of the pack with TOR and NYY fighting for 2nd through 4th. 

.500 plus or minus 10 games?  So between 71 and 91 wins?    Really going out on a limb there, aren't you? 

By the way, 2016 was not Trumbo's career year.  In fact, other than his cup of coffee in 2010, he has only had one season in which his WAR was less than 2016, and that was the year he broke his foot.  With no further moves, we've already more than cut Trumbo's time in right field in half, and another move may well be forthcoming.  Smith may not be a huge upgrade, but he is an upgrade.

Needing innings from 6th, 7th, and 8th starting pitchers is normal.  For every team.  Being lucky enough not to need that in any given season is the rare exception.  If we had to scramble with those type of guys to fill our fourth and fifth spots in the rotation, that would be something else.

Anything can happen, of course, but I'm not seeing the Yankees doing that well.  Yes, we are probably going to be in a dogfight with Toronto for second place, and the Red Sox have to be the favorites, but the playoffs are definitely a realistic possibility IMO and, as a result, so is the World Series.  Hey, our lineup stacks up pretty well against just about anybody, and our bullpen is top notch.  If our starting pitchers can manage league average results, I like our chances.

[Frobby]

4 minutes ago, RZNJ said:

Come now. Career year offensively for Trumbo. We all know what he meant. And yes, I know it wasn't the highest OPS of his career.

You could make a straight-faced argument that 2012 was Trumbo's best offensive year (124 wRC+ vs. 123), but what's the point?     The substance is that Trumbo probably won't be as good in 2017 as he was in 2016.    It's fair enough.     Hopefully he'll defy those odds, and/or other players who had below average years last year (Davis, Jones) will do better and pick up the slack.    I think overall our offense will be slightly down, but it will be close to the same.

[stomperspc]

The models bothers me far less than the reluctance of the main users of the models (BP, Fan Graphs, ect.) to (a) acknowledge when the model is not properly forecasting a particular stat/player/team correctly and (b) manually adjusting to correct the issue. I do some financial modeling and forecasting in my job. I totally get that no model is perfect and if the model output is “mostly” right that is a pretty good model. I also know that is not an excuse for using known bad data in a forecast/projection. If I am using a model that gets one line item incorrect on a consistent basis, I am expected to adjust for that rather than simply leaving in the bad data. I think that’s the part that is missed in baseball projections. It’s just not good enough to say “well, the model always seems to miss on the Orioles”. If you want your projection system to be valuable and useful, it is on you to figure out a solution that at the very least reduces the variance between projection and actual results.

I think the reluctance to manually adjust modeled numbers in baseball is because of the long (and still ongoing) struggle to get people to rely on objective numbers over the eye-test. I think there is this fear that if you manually adjust a modeled output then you are adding subjectivity to the exercise. That’s true, but I’d rather have a semi-subjective (ie. a modeled number adjusted based on trends/historical performance/ect. )number that attempts to get to the right answer than a bad objectively modeled number. We have a five year sample size of the Orioles substantially outperforming their runs scored predictions on PECOTA. Management has stayed consistent during that time. The general makeup of the team (ie. strong bullpen, weak rotation) has stayed consistent. A lot of the key players have remained unchanged. In my option that’s enough of a trend to justify reducing the runs allowed total by some logical and conservative number.

According to LC_O’s, PECOTA has been off on runs allowed by Orioles pitchers by margins of 73, 153, and 71 in 2013, 2015, and 2016, respectively. If BP explained that and took a conservative approach of knocking off 50 runs from the model, that would be a step in the right direction. I want to see an attempt to correct known issues rather than accepting large variances every single year.

[Spy Fox]

6 minutes ago, stomperspc said:

The models bothers me far less than the reluctance of the main users of the models (BP, Fan Graphs, ect.) to (a) acknowledge when the model is not properly forecasting a particular stat/player/team correctly and (b) manually adjusting to correct the issue. I do some financial modeling and forecasting in my job. I totally get that no model is perfect and if the model output is “mostly” right that is a pretty good model. I also know that is not an excuse for using known bad data in a forecast/projection.

Good post and you captured a lot of the reasons I enjoy the projections while still being frustrated by the way they are discussed by many writers who use them. I'm not sure if I think that they should just go in and do manual adjustments, but I do think more investigation should be done as to why the models get right what they get right and why they get wrong what they get wrong. And I think you're right that a fear of appearing too subjective might influence people to adhere to the existing models more strictly. 

[Frobby]

I think allowing manual adjustments would defeat the purpose of these systems.    Two different people could have completely different ideas of what adjustments are appropriate.    It's up to the user of the model to make whatever adjustments that user believes are appropriate.    It's not up to the creator of the model, or it's not a model at all.  

[Frobby]

Here's a good article on PECOTA and the Orioles.    It notes that PECOTA correctly predicted 8 of the 10 playoff teams last year.     There's just something about our team that doesn't fit their model.   http://www.camdenchat.com/2017/2/7/14533204/pecota-projected-standings-2017-orioles

[Number5]

24 minutes ago, Frobby said:

Here's a good article on PECOTA and the Orioles.    It notes that PECOTA correctly predicted 8 of the 10 playoff teams last year.     There's just something about our team that doesn't fit their model.   http://www.camdenchat.com/2017/2/7/14533204/pecota-projected-standings-2017-orioles

They don't normally get 8 out of 10 playoff teams.  Most years there are a few can't miss teams, and the National League was unusually easy to call last year, but 8 out of 10 is rare.  We also need to realize that it isn't just the Orioles they get wrong.  They missed big time on the Rays.

[Frobby]

2 minutes ago, Number5 said:

They don't normally get 8 out of 10 playoff teams.  Most years there are a few can't miss teams, and the National League was unusually easy to call last year, but 8 out of 10 is rare.  We also need to realize that it isn't just the Orioles they get wrong.  They missed big time on the Rays.

I didn't understand their love for the Rays at all.  Even now, they have us scoring only 5 more runs than the Rays, who we outscored by 72 runs last year.    I look at their lineup, and the stadium they play in, and it's inconceivable to me that they'll do that.   

Look, I'm not saying PECOTA is a perfect system.    I'm just saying that it should be judged on how it does overall, not how it does with one or two particular teams.     It's all just for fun, anyway -- it's not like we suffer some adverse consequence because PECOTA perpetually underrates the Orioles.    If anything, it probably helps keep the betting odds healthy, to the advantage of Orioles fans who like to put a little money on their team.    

[Tx Oriole]

On 2/7/2017 at 8:40 AM, Aristotelian said:

I don't see anything hilarious about that. The margin between a 70 win team and a 90 win team is pretty small. The Orioles have lately overperformed. It is possible that given enough chances their model will eventually predict us correctly. I had us as a 70 win team last year and was pleasantly surprised, and we have pretty much tread water this year. I would probably be more surprised by 90 wins than I would 70 wins this year. That said, they must really hate our SPs. They are basically predicting Tillman and Gausman to both have career-worst seasons, with Bundy pitching like a #5 and both Ubaldo and Miley below their norms. At least one of those guys has to step up, right?

Is Ubaldo even on the team? I thought 2016 was his last year with the O's. 

[Number5]

1 minute ago, Frobby said:

I didn't understand their love for the Rays at all.  Even now, they have us scoring only 5 more runs than the Rays, who we outscored by 72 runs last year.    I look at their lineup, and the stadium they play in, and it's inconceivable to me that they'll do that.   

Look, I'm not saying PECOTA is a perfect system.    I'm just saying that it should be judged on how it does overall, not how it does with one or two particular teams.     It's all just for fun, anyway -- it's not like we suffer some adverse consequence because PECOTA perpetually underrates the Orioles.    If anything, it probably helps keep the betting odds healthy, to the advantage of Orioles fans who like to put a little money on their team.    

Again, massively underrating the Orioles and massively overrating the Rays is not  a matter of doing well overall.  The two massive errors do not cancel each other out.  They missed on the Orioles by nearly twenty games and they missed on the Rays by more than twenty games.  They missed by forty games on those two teams, not zero.

[weams]

53 minutes ago, Frobby said:

Here's a good article on PECOTA and the Orioles.    It notes that PECOTA correctly predicted 8 of the 10 playoff teams last year.     There's just something about our team that doesn't fit their model.   http://www.camdenchat.com/2017/2/7/14533204/pecota-projected-standings-2017-orioles

Homers and Buck and Bullpen Aces.