The Twins since April 15th

[DaveW]

Some numbers to look at since April 15th (Twins were 1-6 prior and had a negative 29 run differential)

 

 

Twins record: 48-34

Run differential: +52

 

Kyle Gibson: 2.53 ERA

 

[drjim]

But what are the base runs?

[DaveW]

 

But what are the base runs?

Haha, I assume you are being sarcastic?

I'm not really sold on "base runs" at all, according to base runs the Twins should be 10 games below .500 and the 2nd worst team in the entire American league (worse than the dumpster fires that are the Red Sox and Mariners). Now, have the Twins have some luck? Sure, but I have a hard time believing that it has been so much luck that it has taken the 2nd worst team in the league and made them the 2nd best (record wise)

 

 

[drjim]

 

Haha, I assume you are being sarcastic?

I'm not really sold on "base runs" at all, according to base runs the Twins should be 10 games below .500 and the 2nd worst team in the entire American league (worse than the dumpster fires that are the Red Sox and Mariners). Now, have the Twins have some luck? Sure, but I have a hard time believing that it has been so much luck that it has taken the 2nd worst team in the league and made them the 2nd best (record wise)

 

I think it had some validity earlier in the season but the Twins are a better team right now, maybe significantly better. And the wins are banked. In my mind they are a bullpen arm or two away from really solid up and down the roster.

 

(And yes, I was being sarcastic about base runs).

[DaveW]

 

I think it had some validity earlier in the season but the Twins are a better team right now, maybe significantly better. And the wins are banked. In my mind they are a bullpen arm or two away from really solid up and down the roster.

 

(And yes, I was being sarcastic about base runs).

Agreed, it's not like they are winning 4-3 games with 4 runs off of 5 well timed hits.

 

Also, I think Sano at DH instead of Nunez/Vargas/Escobar/etc honestly may be worth 3-4 wins the rest of the season.

[Otto von Ballpark]

 

I'm not really sold on "base runs" at all, according to base runs the Twins should be 10 games below .500 and the 2nd worst team in the entire American league (worse than the dumpster fires that are the Red Sox and Mariners). Now, have the Twins have some luck? Sure, but I have a hard time believing that it has been so much luck that it has taken the 2nd worst team in the league and made them the 2nd best (record wise)

In regards to Base Runs:

 

The Twins rank 12th in the league in wRC+, and 11th in baserunning runs at Fangraphs, but somehow we rank 7th in runs per game.

 

On the pitching side, we rank 12th in FIP and 13th in xFIP, but 8th in runs allowed per game.

 

We also rank 8th in the league in pythag win percentage.

 

This is not a team you would expect to have the 2nd best record in the league!

 

And Base Runs isn't saying it's all "luck" either -- it could be just a few games luck, plus a few games skewed by early season blowouts, etc.  We're actually 2 games up on our Pythag, which is in turn 7 games up on our Base Runs record.

 

Also, it's not necessarily evenly distributed -- I think the two points in the season when we've really appeared to be a Base Runs outlier was after our 20-7 May and right now on a 6-1 run -- and it wouldn't surprise me if the most extreme Base Runs / Pythag outliers occurred on the heels of such winning (or losing) streaks.  Basically by definition, it's hard to fully support such such extreme records with underlying peripherals and run distributions.

 

Finally, I wouldn't read too much into our Base Runs record relative to the league ("2nd worst") as measures like Base Runs, Pythag, and projections tend to bunch teams together (and the AL is rather bunched up this year anyway).  With our Pythag luck, we would still be right around .500 and in the thick of things even with much more modest Base Runs "luck".

[drivlikejehu]

The ugly start still counts. It certainly drags down the Twins' BaseRuns, which has been steadily rising ever since.

 

It's important to keep in mind that BaseRuns does not predict anything. It only looks back and provides an expected record based on the individual hitting and pitching events that have already occurred.

 

Having said that, the expected BaseRuns winning % is now pretty close to the Fangraphs projected rest-of-season numbers. Extrapolating BaseRuns (which, again, it is not intended for) results in 82 wins, whereas Fangraphs projects 83 wins.

[Otto von Ballpark]

 

Having said that, the expected BaseRuns winning % is now pretty close to the Fangraphs projected rest-of-season numbers.

I think that's been the case for awhile now (BaseRuns winning % close to projected rest-of-season winning %, which in turn is pretty much just the preseason projected winning percentage over the remaining games).  At least, that was the case the previous time I looked at BaseRuns, after May.

[kydoty]

So what you're saying is that they were all bucking under the stress of trying to finish their taxes.

[drivlikejehu]

 

I think that's been the case for awhile now (BaseRuns winning % close to projected rest-of-season winning %, which in turn is pretty much just the preseason projected winning percentage over the remaining games).  At least, that was the case the previous time I looked at BaseRuns, after May.

 

The discrepancy was much larger pretty recently. The past couple series have increased the BaseRuns win% by a fair amount. The projections have exceeded BaseRuns all year, but if the Twins continue to play well that will flip.

[drjim]

I also noticed that Fangraphs now has the Twins as second most likely team in the division to make the playoffs for the first time this season.

 

So something is obviously moving, or the wins banked are starting to become that much more important.

[Brock Beauchamp]

BaseRuns has merit. Over the course of a season, a team's luck is probably going to balance out and BaseRuns is representative of what *should* have happened, not what actually happened.

 

Not unlike FIP or xFIP, really.

 

The problem with these metrics is that a lucky team doesn't automatically become unlucky just so their FIP/xFIP/BaseRuns balance at the end of the season.

 

The Twins were incredibly lucky in late April and all of May. That shot them out to a (going from memory) +8 BaseRuns record, which is insanely high. Well, guess what? Those lucky wins are in the bank. At that point, the BaseRuns record becomes kinda irrelevant because it doesn't factor in games that already happened and project from there.

 

Right now, the Twins are +9 in BaseRuns. For all intents and purposes (again, going from memory here), they have been BaseRuns neutral for a month and a half. That means they're no longer lucky but because they banked so many wins early in the season, they're still within a few games of their high water mark of the season (which was +11 games over .500, IIRC).

 

It's almost a given the Twins will have a ridiculously high BaseRuns differential for the entire 2015 season but that doesn't matter. They don't need to "keep beating the odds" because they beat the odds so soundly in May. They just need to stay luck neutral for the rest of the season and they're a playoff contender, BaseRuns be damned.

 

Also, the luck neutral Twins of today are a better team than the lucky Twins of May. Sano, Hicks, Rosario, and Mauer are starting to produce while Kyle Gibson has mysteriously turned into a strikeout monster while inducing groundballs left and right. There's still some room for upward movement on this roster.

 

To summarize a post that's about ten times longer than I originally anticipated, it's July and the Twins' BaseRuns record is pretty much irrelevant. I'd like to see a second version of the metric that projected the rest of the season and removed all that "this should/shouldn't have happened" garbage because that's water under the bridge. It happened. It can't be changed.

 

My guess is that if BaseRuns took the Twins' 49 wins as fact and projected the rest of the season based on the last month of play (ie. using the current roster and not the one fielded in May), they'd put the Twins somewhere around 83-85 wins for the season (basically .500 ball from here on out).

[Otto von Ballpark]

 

The problem with these metrics is that a lucky team doesn't automatically become unlucky just so their FIP/xFIP/BaseRuns balance at the end of the season.

The metric doesn't say that at all, so I don't really see how it's a problem with the metric.  Fangraphs doesn't even say that -- their updated projections assume neutral luck for the rest of the season (more or less the preseason projections with some tweaks).  Nobody that I am aware of is simply assuming future bad luck to balance the ledger.

[Otto von Ballpark]

 

My guess is that if BaseRuns took the Twins' 49 wins as fact and projected the rest of the season based on the last month of play (ie. using the current roster and not the one fielded in May), they'd put the Twins somewhere around 83-85 wins for the season (basically .500 ball from here on out).

Fangraphs has us projected to 83 wins.  That's the 49 wins in the bank, plus .463 the rest of the season, which is more or less the same as their preseason .457 projection.

 

It's hard to move the projection needle quickly.

[Brock Beauchamp]

The metric doesn't say that at all, so I don't really see how it's a problem with the metric. Fangraphs doesn't even say that -- their updated projections assume neutral luck for the rest of the season (more or less the preseason projections with some tweaks). Nobody that I am aware of is simply assuming future bad luck to balance the ledger.

My point is that it becomes a counting stat that becomes less useful for projection as the season progresses. I'm not saying BaseRuns should be discarded, I'd simply like to see a projection version that is displayed alongside it. That would be more useful as the team banks lucky or unlucky wins along the way.

[Brock Beauchamp]

Fangraphs has us projected to 83 wins. That's the 49 wins in the bank, plus .463 the rest of the season, which is more or less the same as their preseason .457 projection.

 

It's hard to move the projection needle quickly.

Ah, there you go. I still think that's on the low side but it's an interesting projection.

 

Projecting on the fly is tough because it's hard to gauge regression or roster moves quickly. I'd like to see a metric more aggressively project those kinds of changes.

[drivlikejehu]

 

My point is that it becomes a counting stat that becomes less useful for projection as the season progresses. I'm not saying BaseRuns should be discarded, I'd simply like to see a projection version that is displayed alongside it. That would be more useful as the team banks lucky or unlucky wins along the way.

 

I'm not sure what a 'projection' version of BaseRuns would be. It's just a formula to estimate how many runs will be scored as a result of a set of batting outcomes. The Fangraphs projections don't need to really worry about linear weights too much because everything is being normalized - sequencing is not a consideration and so the projected pythag is fine.

 

Looking back, the actual pythag is pretty worthless because runs scored and allowed are heavily influenced by sequencing and other factors aside from the individual batting outcomes. BaseRuns essentially trues up runs scored and allowed to produce an accurate pythag record.

 

The Fangraphs projections don't require a true up like that - to the extent they have error, it is because the individual projections are just off, rather than anything to do with how the events will work together.

[Otto von Ballpark]

 

My point is that it becomes a counting stat that becomes less useful for projection as the season progresses. I'm not saying BaseRuns should be discarded, I'd simply like to see a projection version that is displayed alongside it. That would be more useful as the team banks lucky or unlucky wins along the way.

Any projection is essentially using BaseRuns, in that you wouldn't project a team to bunch together their hits and/or spread out their hits allowed.  BaseRuns just lets you look back and see what teams seemed to do that.

[Otto von Ballpark]

 

Ah, there you go. I still think that's on the low side but it's an interesting projection.

Projecting on the fly is tough because it's hard to gauge regression or roster moves quickly. I'd like to see a metric more aggressively project those kinds of changes.

Yeah, Fangraphs projections haven't changed much since opening day.  Every team's rest-of-season winning percentage projection is within 3.7 wins (prorated over 162 games) of their opening day projection.

 

But it's hard to project more aggressively than that at the team level.  For every guy that would benefit from a more aggressive projection based on recent results (Dozier, Hicks), there is probably another who would suffer (Suzuki, Danny Santana).  The Twins roster just hasn't changed that much to make a huge difference either way (keep in mind, their preseason projection already included time for May, Rosario, Sano, etc.)

 

Projections aren't supposed to be that sexy.  Results and *predictions* are the interesting part.