Analyzing Analytics

[TJseven7]

I'm of the belief analytics are still a work in progress where filters are being added/removed and tested in the attempts to find the proper weight for each factor.

 

When the question was proposed.... what is wrong with the offense I did some digging. Sample size is small and much of it might be an anomaly. I'm not capable of doing the analytics they are but I found this interesting and possibly telling.

 

Since the ASB the team is 10-18

Since the ASB in games where the team had 2bb or less 2-12

Since the ASB in games where the team had 3bb or more 8-6

 

Sample size... i know

 

Since the season started the brewers are 38-28 in games with 3bb or more (30-22 pre asb)

Since the season started the brewers are 22-31 in games 2bb or less (20-19 pre asb)

 

I can hear it... "yeah idiots bbs are good."

 

Since asb 4.28 runs in games with 3bb or more

Since asb 2.64 runs in games 2bb or less

 

Ks since asb are also up .8 per game.... when I ran this 3 games ago it was up 1.1 per game. In the k decrease 6 runs per game average.

 

Bbs had also dropped after the asb... they are currently down .2 bb per game... they were down .4bb per game 3 games ago... in the bb increase 6 runs per game.

 

Again... team is 10 over 500 in 3bb or more games... and it dropped from 3.3 average to a post asb average of 2.9... corrected to 3.1 the last 3 where the team averaged 6 rpg.

 

First thing I suppose. The 3bb line is not a false indicator. The increase in Ks and drop in BBs was enough to push the offense off a statstical cliff.

 

Second thing I suppose. Analytics once argued Ks are not that bad of a thing. I don't believe that has changed for individuals. But I believe they've identified a threshhold effect where Ks choke an offense as a whole.

 

1 indicator: Houston... offseason of beltran aoki mccann do not strike out. Mccann is a very good catcher but the others are mediocre. But what they do is raise walks and lower Ks.

 

2 indicator: Draft... 2nd base Huira elite contact hitter. Expected low K rate. Lutz very natural contact tool for a plus power profile.

 

3 indicator: Sogard preseason... low K batter.

 

4 indicator: Why Kinsler? Cuts 2nd base Ks in half (almost shaves 60% off villar). Granderson? Ks 40% less than Broxton. None of them are all that good but their big plus is better bb far lower K. Walker? Better than kinsler and still shaves 40% off Ks.

 

You can call this nonsense and sample size but for some reason houston and now a competative MKE are actively looking to shave Ks and add walks to their power based lineups. Trying to stay away from a saturation point I'd suppose.

 

With high k power in santana, 1st (thames and aguilar), shaw, that you cant sacrifice. Braun Arcia fit in just fine with moderate K rates (braun lately rusing his contact and plate duscipline as if told they need obp over power from him) and Pina is the ideal catcher. That leaves 2 spots where bbs need to rise and Ks need to be cleaved out.

 

If that's the case... villar is done in MKE and Broxton is next. 2b and CF who do not K and have a healthy difference in avg and obp are the target.

[Brewer Fever]

It's pretty simple, when they're not hitting home runs they have no way to generate offense. Especially with all those K's. Without HRs they need 3 hits to generate a run, something they just aren't good at. Very few productive outs

[Boomer5]

People used analytics as the be all, end all.

 

K's do matter. "Clutch" is a word that analytics people hate but it absolutely does matter. Batting average matters.

 

Good analysis on why just number crunchers aren't seeing the whole picture.

[Ennder]

People used analytics as the be all, end all.

 

K's do matter. "Clutch" is a word that analytics people hate but it absolutely does matter. Batting average matters.

 

Good analysis on why just number crunchers aren't seeing the whole picture.

 

 

Nobody has ever said Ks don't matter, a K as a result does not significantly matter compared to an out as a result. That is a completely different argument. People also have never said clutch isn't good, clutch just isn't predictive. Because you were clutch one time does not mean you are more likely to be clutch the next time.

 

The big issue with analytics is people not understand what they are saying and misquoting them.

[JosephC]

There are many evaluations of hitting in clutch situations, but as ennder states results for a certain hitter are generally mixed and not predictive of future outcomes. Some argue that it's generally just due to small sample sizes. For example, a hitter probably only sees 60-80 clutch plate appearances a season. If he goes 12-60 he's a .200 hitter in those situations and doesn't look good, but if he goes 18-60 then he's a .300 hitter in those situations and looks pretty good. One year he ends up with a 12-60 season and the next he has an 18-60 season? How much of that difference was due to good or bad luck? Is he a clutch hitter or isn't he?

 

Take Braun in his prime years for example. Following are clutch factors as calculated by fangraphs.

2007 = -1.63

2008 = +0.83

2009 = +0.49

2010 = -0.32

2011 = +0.17

2012 = -0.87

Was Braun a clutch hitter or not during these seasons?

 

Or take Albert Pujols during his 17 year career. He had mostly negative seasons during his early years and more positive seasons later in his career. But to take just a 5 year section:

2004 = -0.33

2005 = -1.64

2006 = +3.26

2007 = +0.38

2008 = -0.17

So was Pujols clutch during those seasons? Average stat line during those years was .335/.443/.631/1.064 with 41 home runs and 119 RBIs. Over his 17 years his total clutch factor +1.71, so +0.1 per season. I'm not an expert and wouldn't claim to know how much extra run generation can be attributed a clutch factor of +0.1 per season, but I'm guessing it's not significant. And when looking at year to year variability, his worst clutch season was 2005 and his best clutch season was 2006...back to back seasons. Was that due to him being a poor hitter in 2005 and was much improved in 2006? The overall statistics say no. Pujols was a great hitter during those seasons and any variability in clutch situations during those seasons would more likely be attributed to "circumstance" or "luck" or whatever term you wish to apply.

[Lathund]

With regards to the BB point: One thing that "analytics" does show, one of the early finds of the sabermetric era really, was just the value of walks. Or to be more specific, the value of getting on base and not recording outs, which is the main value of a single. So walks do matter, and it's reflected in stats like wOBA and wRC+, which are essentially OPS, but where the constituent parts of OPS (Walks, HPB, singles, doubles, triples, HRs) are weighted. One of the issues with OPS is that 1pt of slugging and 1pt of OBP count the same, when they really shouldn't.

 

So to move from BBs to Ks. Of course Ks are bad. Especially in certain situations. But that's missing the tradeoff, and why the K% is high. A player who takes a lot of walks will also, as a consequence of that (Unless you're Joey Votto) take more called third strikes than others. Likewise, a player who swings for the fences will have more swinging third strikes. So as the poster quoted below points out, K's aren't all that different from other outs; you lose value in not being able to advance runners. But you gain value in more walks, more HRs and, I supopse, in running up pitch counts.

 

So if the high K% are a result of a specific approach that gets you other advantages, they're not as bad as they seem. But striking out a lot due to simply being bad, well, that's never a good thing. The Brewers are a bit of both currently. There seems to be a Patience/Power approach. But we're also in a situation where the team isn't necessarily filled with the players, or type of players, the front office want in the long term. When you're taking flyers on waiver claims, cheap trades, cheap FAs and such, you can't be too picky. You see a player that you suspect is undervalued and where you see upside, and take a chance. That will include guys who strike out a ton but where the positives have been missing; Keon Broxton might be a shining example of that. These types of players will have warts, that's why they're available in the first place.

 

 

People used analytics as the be all, end all.

 

K's do matter. "Clutch" is a word that analytics people hate but it absolutely does matter. Batting average matters.

 

Good analysis on why just number crunchers aren't seeing the whole picture.

 

 

Nobody has ever said Ks don't matter, a K as a result does not significantly matter compared to an out as a result. That is a completely different argument. People also have never said clutch isn't good, clutch just isn't predictive. Because you were clutch one time does not mean you are more likely to be clutch the next time.

 

The big issue with analytics is people not understand what they are saying and misquoting them.

 

Very well put!

[Thurston Fluff]

Analytics will always be a work in progress. People are finding all kinds of new things to look at and measure. Add in new technology that can make measurements possible and it seems like a long time before we come close to covering all the bases (pun intended.) I think overall analytics are fine for judging major league talent. They do quite well judging minor league talent at the upper levels but not as well at lower levels. The problem is the lower the level a players is the more he faces subpar players. At lower levels I would be more interested in how they do against true prospects than I am in how they do against roster fodder. Numbers sometimes don't show that in large enough samples sizes to help. Eyes can tell you if they are overmatched when going 0-3 against better pitchers but a stat doesn't necessarily tell you that.

I don't think they work very well for judging ameteur talent or talent from international leagues. Things like exit velocity and spin rates may very well be changing that to some extent but I don't think it is enough to be trusted more than traditional scouting is at this point. Even then it has to be part of the evaluation. Just not the be all and end all.

[Thurston Fluff]

Analytics will always be a work in progress. People are finding all kinds of new things to look at and measure. Add in new technology that can make measurements possible and it seems like a long time before we come close to covering all the bases (pun intended.) I think overall analytics are fine for judging major league talent. They do quite well judging minor league talent at the upper levels but not as well at lower levels. The problem is the lower the level a players is the more he faces subpar players. At lower levels I would be more interested in how they do against true prospects than I am in how they do against roster fodder. Numbers sometimes don't show that in large enough samples sizes to help. Eyes can tell you if they are overmatched when going 0-3 against better pitchers but a stat doesn't necessarily tell you that.

I don't think they work very well for judging ameteur talent or talent from international leagues. Things like exit velocity and spin rates may very well be changing that to some extent but I don't think it is enough to be trusted more than traditional scouting is at this point. Even then it has to be part of the evaluation. Just not the be all and end all.

[Austin Tatious]

Good post, TJ. I enjoyed your analysis and found it interesting.

[Bombers]

You can call this nonsense and sample size but for some reason houston and now a competative MKE are actively looking to shave Ks and add walks to their power based lineups. Trying to stay away from a saturation point I'd suppose.

 

With high k power in santana, 1st (thames and aguilar), shaw, that you cant sacrifice. Braun Arcia fit in just fine with moderate K rates (braun lately rusing his contact and plate duscipline as if told they need obp over power from him) and Pina is the ideal catcher. That leaves 2 spots where bbs need to rise and Ks need to be cleaved out.

 

If that's the case... villar is done in MKE and Broxton is next. 2b and CF who do not K and have a healthy difference in avg and obp are the target.

Great post TJ. I agree with this 110%. I always felt years ago when we had Fielder, Braun, Weeks, Hardy, Hart and others we should have been finding more players who hit like Jerry Hairston Jr and Nyjer Morgan who seemed to have higher bb rates / lower K rates. If this is indeed their plan, I am in.

[trwi7]

I always felt years ago when we had Fielder, Braun, Weeks, Hardy, Hart and others we should have been finding more players who hit like Jerry Hairston Jr and Nyjer Morgan who seemed to have higher bb rates / lower K rates. If this is indeed their plan, I am in.

 

You say you wanted to find hitters with higher walk rates and then listed players who had below average walk rates most of their careers.

[TJseven7]

Cain Yelich in... 230 total k, high bb/obp players. Looks like we are full on making our astros BBvK shift. Broxton alone strikes out close to what those 2 do combined. As it sits currently aguilar loses his ab and thames loses some. This lessens Ks.

 

2b... remains to be seen.

[adambr2]

Yeah, kind of. I am still unsure if the K rate itself is a major point of emphasis, or a side effect of other points of emphasis.

 

This is still a high K team. Broxton may be the king, but guys like Santana, Thames, Villar and Phillips aren't far behind.

 

I don't know why people see Yelich as a very strikeout averse player. I guess if you're comparing him to Broxton, certainly, but he'll be right there with Braun, Pina, and Arcia among starters at around 20% and then there's a big gap between them and the 30% group of Santana, Phillips, Villar and Thames, assuming Broxton isn't starting at the MLB level here, who is pretty much in a category his own in that department. Cain might be the only guy under 18-20% and he'll be close.

 

Don't know what else will change with the roster between now and opening day but this team will still strike out a ton. Just not a record breaking ton.