How Much Do You Believe, In Regression To The Mean?

[sveumrules]

We're sixty games into the schedule now and there are no two ways about it, the offense has been bad. Like 87 wRC+ and 4.02 R/G (both 25th in MLB) bad. But as we all know, something happening for sixty games is no guarantee it is going to continue happening that exact same way for the next one hundred plus games.

Projections try to suss out some of that still early season small sample noise, and as we know they do a better job of predicting future results than just looking at actual (or even pythagorean) results to date this early on. Want some mild, un-biased optimism? The computers at FanGraphs think that rest of season the team projects to score 4.47 R/G which would be 22nd, BPro thinks we have something like the 19th/20th best offense based on their DC RS team metric. They both seem to think the Brewers have underperformed their talent level to this point and better days are ahead.

I was curious if looking at recent history might offer some additional insight so using the custom date tool on the FanGraphs leaderboards I went back over the last four full seasons and found thirty teams (a whole league!!) who had team position player wRC+ ranging from 80 to 92 over their first sixty-ish games or so, and then adjusted the dates to see how they performed over their remaining 100 some games to close the season. The results, with format of (60 game wRC+ | rest of season wRC+ | difference)...

2018
TEX (80 | 95 | +15) COL (84 | 101 | +17) BAL (85 | 89 | +4) ARI (85 | 99 | +14) MIA (86 | 90 | +4) SDP (87 | 92 | +5)

2019
BAL (81 | 93 | +12) CLE (81 | 107 | +26) CIN (85 | 96 | +11) KCR (88 | 82 | -6) CHW (88 | 94 | +6) SDP (90 | 97 | +7) PIT (91 | 99 | +8)

2021
PIT (81 | 91 | +10) SEA (83 | 100 | +17) DET (84 | 101 | +17) TEX (85 | 83 | -2) MIL (86 | 105 | +19) KCR (87 | 90 | +3) CLE (87 | 97 | +10) MIA (89 | 90 | +1)
ARI (91 | 88 | -3)

2022
PIT (83 | 85 | +2) CIN (87 | 83 | -4) COL (88 | 85 | -3) CHW (88 | 104 | +16) KCR (88 | 95 | +7) ARI (90 | 04 | +4) BAL (90 | 103 | +13) TEX (92 | 101 | +9)

Put if all together and out of the thirty teams...twelve (40%) were who we thought they were with their wRC+ getting worse or only improving marginally over the rest of the season, five (16.7%) saw an increase between five and nine percent, and the remaining thirteen (43.3%) saw an increase of over ten percent. The average across all thirty teams was an increase of about 7.6%.

Obviously contextual factors are unique to each team, with some of the larger increases being driven by player acquisition (2021 MIL with Adames) or prospect ascension (2022 BAL with Adley/Gunnar), while most of the teams who saw a decrease or little to no change likely were also selling off players at the deadline and/or giving younger players a shot down the stretch.

Either way, whether it is the computerized projections or a smattering of recent history, it appears as though the Brewers offense should see some degree of improvement as the season goes along. Will it be enough to win a mediocre NLC or maybe get hot and make some noise in the playoffs? I guess that all depends on how much one believes, in regression to the mean. 

[brewers888]

If I believed these hitters were any good I would expect better results but I don't. If we are talking about regression don't we have to discuss the high leverage relievers who have pitched unbelievably well and are due for some negative results. Overall I see this as a mediocre team that should win a division where they will probably be the only team over .500 but not nearly good enough to beat the very good teams like Atlanta and LA.

If anything the lousy division means we will likely hold on to Burnes, Woodruff, and Adames until they become free agents after the 24 season and then really hit the skids with very little starting pitching on the way and a cheap owner.

[brewers888]

I suppose Urias could give the offense a boost as well as getting more from Contreras and Willy but other than that I don't see much upside unless the Wiemer of the last week or so is legit which I question.

[Jake McKibbin]

6 minutes ago, brewers888 said:

If I believed these hitters were any good I would expect better results but I don't. If we are talking about regression don't we have to discuss the high leverage relievers who have pitched unbelievably well and are due for some negative results. Overall I see this as a mediocre team that should win a division where they will probably be the only team over .500 but not nearly good enough to beat the very good teams like Atlanta and LA.

If anything the lousy division means we will likely hold on to Burnes, Woodruff, and Adames until they become free agents after the 24 season and then really hit the skids with very little starting pitching on the way and a cheap owner.

Carlos Rodriguez and Misiorowski would argue this, plus Gasser was lights out last start in Triple A

There's some shoots of pitching coming through for sure

An owner with a budget, absolutely. And it's fair to say that using younger guys coming through has worked well.

The hitters (such as Adames) are underperforming massively compared to career numbers, so I think that's fair enough in some regards

 

[brewers888]

3 minutes ago, Jake McKibbin said:

Carlos Rodriguez and Misiorowski would argue this, plus Gasser was lights out last start in Triple A

There's some shoots of pitching coming through for sure

An owner with a budget, absolutely. And it's fair to say that using younger guys coming through has worked well.

The hitters (such as Adames) are underperforming massively compared to career numbers, so I think that's fair enough in some regards

 

The problem with the offense is that most of these hitters just aren't any good. Sure you can say that Adames has underperformed and it can be countered with Miller overperforming.

[owbc]

As a whole for a team like Milwaukee I think they will regress to the mean, and that comes mostly from personnel decisions -- swapping out under-performers with other options and potentially targeting upgrades via trades. I would be more concerned if the offense was ranked #15 because it's much, much harder to go from 15th to 10th than it is to go from 25th to 20th or even 15th. Those nice statistics clearly back up that the odds are at least 50/50 that things will improve. 

For individuals, it comes down to what their track record is. Adames is obviously the most likely candidate to regress closer to his career averages, but players can and do have down years that have little explanation. Getting Urias back will be huge since that will get Turang and his 48 OPS+ out of the lineup. 

I think you ride Wiemer in CF, he's been below-average but not terribly so and there's a ton of upside there. I get why they have been riding Turang as well but he's gotta put up at least a 75 OPS+. 

The rotation has its own set of problems...I have lower confidence that we'll be able to get considerable improvement there. 

[Jake McKibbin]

6 minutes ago, brewers888 said:

The problem with the offense is that most of these hitters just aren't any good. Sure you can say that Adames has underperformed and it can be countered with Miller overperforming.

As mentioned by @owbcit's important to realise some of these are rookie struggles too, Turang has been a black hole, Wiemer was awful in May (though he may be the Wiemer we hoped for now) and there's a settling in period.

Would a Adames/Urias/Miller/Anderson infield be better right now, of course, thats where we need to recognise the injuries (that's with a Rowdy DH) and the toll they've taken.

So can I se them improving, undoubtedly. And with a full strength pitching lineup, this team may only need mediocre bats in the first place

[Brewcrew82]

2 hours ago, owbc said:

The rotation has its own set of problems...I have lower confidence that we'll be able to get considerable improvement there. 

That’s interesting because I actually have high confidence we’ll get meaningful improvement in the rotation when Woody and Miley get healthy and Burnes and Peralta pitch closer to their abilities.

[owbc]

1 hour ago, Brewcrew82 said:

That’s interesting because I actually have high confidence we’ll get meaningful improvement in the rotation when Woody and Miley get healthy and Burnes and Peralta pitch closer to their abilities.

I totally forgot Woody is coming back. That will be tremendous. My thinking was that both Miley and Houser pegged as likely to regress downward, but on further review Houser is performing similar to his 2021 numbers and Miley's numbers look sustainable as well. Miley was a tremendous steal, I have no idea why nobody offered him twice the salary that he's making. 

So on second thought I agree with you.

[Lathund]

I don't think this offense will be great, but yeah I do expect the offense to be somewhere close to league average going forward as opposed to what it currently is. There's simply more positive than negative regression to be expected. Can use both projections and things like statcast to aid. Using statcast (And setting aside the tiny sample sizes like Toro, Monasterio, Urias)  there is only really one big overperformer (In terms of wOBA vs xwOBA); Miller. And his xwOBA is still a slightly above average, and climbing, so he's due to regress, but it's not like the performance should crater. Yelich's statcast page actually looks really good, the best it has in years, and is underperforming by 30 points, Adames by 30 points, Contreras by 20. Taylor is also a huge underperformer (By 65 points) but the sad news there is that the xwOBA is still only .248. Oof. Now the Statcast metrics aren't anywhere near exact enough to say that that's what their performance will be going forward, just that they make good contact and haven't quite yet seen the results to match. The underlying performance might go up, or down, too. It is a reason for optimism though. 

Projection systems are adressed in the OP to some degree, but generally also expects more positive than negative regression. Miller again being the exception. 

Like I said I don't expect the offense at the plate to really really be better than average, but I do expect it to be average going forward. I think it's a very reasonable expectation both based on projections, advanced data, and personell changes. Voit has already been replaced. Winker will improve or be replaced. Taylor too. Turang will likely improve some, but if not will see less playing time. Yelich, Contreras, Adames should see some slight improvement overall, everything points to that. Urias is a better hitter than the guys playing time he'll replace. This will outweigh Millers likely regression. 

Elite defense and league average hitting is fine. It just needs to be accompanied by the pitching being reasonably healthy and matching its potential, which it really hasn't so far. Strikeout rates overall are down a ton, which is worrying; was something the team used to do really well. 

[Robocaller]

By definition, regression to the mean always happens, given a long-enough time frame.

The only time it doesn't happen is when the true value has changed.

[nate82]

The Brewers bats are rather bad.....

[brewerfan82]

23 minutes ago, nate82 said:

The Brewers bats are rather bad.....

So are the Blue Jays, Yankees, Rangers, Rays, Diamondbacks, and Padres even worse batters according to this?

[nate82]

8 minutes ago, brewerfan82 said:

So are the Blue Jays, Yankees, Rangers, Rays, Diamondbacks, and Padres even worse batters according to this?

Depends on how you read it but yes at least for the Yankees and Padres.  The Rays, Rangers and Dbacks probably have some regression coming at some point. 

[Oxy]

On 6/7/2023 at 4:42 PM, Robocaller said:

By definition, regression to the mean always happens, given a long-enough time frame.

The only time it doesn't happen is when the true value has changed.

Your second point is extremely important...Regression to the mean always happens with enough trials when the true value is known...i.e. rolling a fair dice or flipping a fair coin.

In most real world applications (including baseball), the true value is not known...and variables are constantly changing that affect that true value.

[BruisedCrew]

I believe in regression to the mean when it comes to statistics relating to truly random events.

But, I believe in it less when it comes to baseball projections, in part because the mean isn't a certainty and also because there are factors other than randomness that affect performance.

In baseball, I believe in players having bad seasons and players having breakout seasons that present a significant deviation from their previous performances. In the first part of the season it's hard to tell if a player is having one of those, or is just going through a hot or cold spell. 

Coming into the season I thought the Brewers offense would be close to league average because of the large number of question marks. Of course, they have been significantly worse than that. I'm having a hard time convincing myself that the rest of the season will be much different. This collection of players might just be showing who they really are at this stage of their careers.

[sveumrules]

On 6/6/2023 at 8:29 AM, sveumrules said:

We're sixty games into the schedule now and there are no two ways about it, the offense has been bad. Like 87 wRC+ and 4.02 R/G (both 25th in MLB) bad. But as we all know, something happening for sixty games is no guarantee it is going to continue happening that exact same way for the next one hundred plus.

In the 34 games since making this post the Brewers have posted a 94 wRC+ and 4.44 R/G, so the offense has been improving.

Hopefully a couple two tree trade deadline acquisitions and get that arrow pointing up at an even steeper angle.

[CheezWizHed]

I certainly hope the Brewers don't regress to the TLR led Cardinals!

Oh wait... the isn't the definition of "regress to the mean" ?