Regression and the trade dealine

[chaderic20]

There was a good article on ESPN today about regression, particularly as it pertains to players traded at the deadline.  It's definitely worth going to read the whole thing, but here's the TL,DR:

• First, regression does not mean getting worse.  Regression means a player's performance will tend to move towards their established track record.  Therefore, a player can have either negative regression, meaning they've been performing above their track record, and are likely to get worse, or a player can actually have positive regression, meaning they've been performing below their track record and are likely to get better.
• Second, GM's have a pretty bad overall success rate (44%) on betting against regression at the deadline.  Breaking it down though, they actually have a 62% success rate when betting on positive regression and only a 24% success rate when betting against negative regression. In other words, GMs do much better trading for an underperforming player than an overperforming player.
• Third, and Twins-specific, he notes that Escobar is a prime candidate for negative regression, so buyer beware, but Dozier is a prime candidate for positive regression, so he's a good buy-low candidate for other teams.
• Finally, looking at teams' performances based on WAR and expected win/loss, he notes that the Twins have the highest expected positive regression for the second half at +2.84 wins, and the Braves have the highest expected negative regression at -6.41 wins (yikes!).

[Vanimal46]

Shhhh. Don't let others here know that Escobar is a prime candidate for negative regression and Dozier for positive regression...

 

Interesting that they're predicting a crash to Earth for Atlanta. They do have a ton of young players to rely on I guess.

[chaderic20]

The author does note that Atlanta's projections could be due to the lack of track record to base expectations on for all their young players.

[Han Joelo]

This makes good sense.  With Dozier's track record, hopefully the Twins don't sell low...I think the second half could be wild.  Or maybe just hope.  As in epic collapses, epic comebacks.  Most (all?) national writers have bestowed the division to Cleveland, one writer on the Ringer noting that their 7.5 game lead will soon be in double digits.  Hubris, I say.  I feel like every single team has at least a Death Star (porthole) sized flaw...except maybe Houston.

[Doomtints]

I don't think this is a particularly great article as nothing is quantifiable. 

 

24% negative regression ... OK ... does this mean a player was at 180 OPS+ before the trade and 150 OPS+ after? Every team will take that.

 

62% positive regression ... same thing ... if a player is at 98 OPS+ before but 102 OPS+ after, so what?

 

What I'd really like to see is quantifiable numbers regarding the value of the player compared to the league. If Joe Nathan gets traded in his prime and goes from 316 ERA+ to 250 ERA+, he's still one of the top two closers in the game. His new team would be thrilled to have him. And this is how every team will gauge the success or failure of a trade -- what did the player do for us?

The Twins are on pace for +2.84 WAR? Again, so what? Are they going to pull out the final W-L record for the Twins at the end of the year and say "Everyone else expected them to get 3 wins less than this." No, they won't.

[chaderic20]

The article points out the exact flaws you mention (a player may have technically negatively regressed, but was still a great player, therefore it may have still been a good trade).  But the point isn't to evaluate individual trades, it's to look at the broader picture, namely, in general it's not a good idea to bet against regression, so when evaluating a potential trade target you should consider their long term track record at least much if not more than their current season stats.  You will always be able to find lots of individual exceptions to a generalized result, but that doesn't invalidate the overall result.

 

As for the team regression, it's just saying you can probably expect some positive regression for the Twins as a team.  Is an extra 2-3 wins gonna make a difference at the end of the season, and get us the division, probably not, but you never know.

[yarnivek1972]

An extra 2-3 wins at this point means nothing but a lower draft slot.

[old nurse]

 

An extra 2-3 wins at this point means nothing but a lower draft slot.

They will not get a top 5 draft pick as there are so many bad teams.  A mid round slot difference of 1 or 2  is a couple hundred thousand of slot money. They did just fine this year year with the money.

[old nurse]

 

 

This makes good sense.  With Dozier's track record, hopefully the Twins don't sell low...I think the second half could be wild.  Or maybe just hope.  As in epic collapses, epic comebacks.  Most (all?) national writers have bestowed the division to Cleveland, one writer on the Ringer noting that their 7.5 game lead will soon be in double digits.  Hubris, I say.  I feel like every single team has at least a Death Star (porthole) sized flaw...except maybe Houston.

What is selling low when the best part of the Familia trade was getting bonus pool money and the best part of the Machado trade was getting a potentially mediocre outfielder and a potential 5th starter?

[Otto von Ballpark]

 

They will not get a top 5 draft pick as there are so many bad teams

 

The Twins are only 4 games out of the 5th pick right now (Detroit). There are only 3 *really* bad teams by record so far this year, and then a bunch that are on pace for around 70 wins. Last year the 5th pick was a 68 win team.

 

If I had to bet, I'd say we clear that, but we've definitely got the potential to achieve top 5 status!

[old nurse]

 

The Twins are only 4 games out of the 5th pick right now (Detroit). There are only 3 *really* bad teams by record so far this year, and then a bunch that are on pace for around 70 wins. Last year the 5th pick was a 68 win team.

 

If I had to bet, I'd say we clear that, but we've definitely got the potential to achieve top 5 status!

They would also have to jump over 5 teams to do it. That is counting on the Mets, Miami, Cincinatti, Texas and Detroit to play better and Toronto and the Angels not to fold. Right now there are 7 teams that have the same or worse winning percentages as the Reds did last year.  Miami and the Reds could as easily regress to that number.  To get worse than Detroit that would mean that Detroit, who has been falling, would have to maintain their current seaso's pace and the Twins would have to play .350 baseball. To get to San Diego's level might be impossible as they traded their bullpen and currently the Twins would have to finish out at .320  Anything is possible except catching the Orioles and KC. We will see after the trade deadline what is left of some teams.

[Thrylos]

The issue with this is that they do not take into consideration the natural decline that comes with age/injury etc.

Tim Lincecum  and Joe Mauer will not regress to their mean performance.

 

Period.

 

Also, for players past their prime, like Dozier, is hard to tell how far behind the 8 ball they are at this point.

Opposite with players who are entering their prime like Escobar.  Why should his pre-prime numbers be the target of "regression.".

Performance is a bell curve.  It is important to: a. figure which part of the curve a particular player is and b. how wide is the peak.

It varies with each individual player and his circumstances.

Other than that, pretty entertaining