Sabermetrics, My Take

[weams]

Here is a good (or at least fun) use of sabermetrics, from our very own team. Ubaldo is obviously a lot better this year than last. Is this a breakout, luck, or something in between?

FiP and xFip, when compared to ERA, are usually fairly easy ways to spot luck, or at least a regression candidate. Ubaldo's are right in line his ERA this season, so maybe we have a true breakout?

2015 is Ubaldo's career best Str%. That?s the good news. And on the surface, his strikeout rate is up, as are his ground balls. But his swinging strike rate sits at a career low and his strikeouts are currently boosted by a career high looking strike rate, which is at a level that won?t be sustained. It?s true some of those looking strikes may become swinging strikes, but some of them will also be put in play, eliminating the possibility for a strikeout.

So, while I think a lot of Ubaldo's gains are here to stay (well, for 2015 at least) I expect his K rate to drop, and his ERA+Fip to rise a bit.

Am I right? We'll see.

I think Ubaldo was always this guy. Right? Very inconsistent. Real talented. It's just we have good guys to help him along when he gets out of whack.

[Frobby]

Is it "sabermetrics" to say that Paredes wasn't a very good hitter over 6-7 years in the minors, or in his previous major league stints, and therefore it is not very likely that he will be a good major league hitter in the future even though he's been good for his last 100 at bats? That's not exactly calculus. And sabermetrics certainly wouldn't say Paredes' chances of being a good hitter in the future are zero, just that they aren't very high. Hopefully he's the exceptional case, but even if he is, it doesn't disprove anything.

[CA-ORIOLE]

Is it "sabermetrics" to say that Paredes wasn't a very good hitter over 6-7 years in the minors, or in his previous major league stints, and therefore it is not very likely that he will be a good major league hitter in the future even though he's been good for his last 100 at bats? That's not exactly calculus. And sabermetrics certainly wouldn't say Paredes' chances of being a good hitter in the future are zero, just that they aren't very high. Hopefully he's the exceptional case, but even if he is, it doesn't disprove anything.

Paredes wasn't exactly a poor hitter in the minors. Batting average is routinely dismissed in islolation, but it's probably one of the better tools to project major league performance. You add power, even if that power may not seam readily evident by the statistics available to us, then that player has some potential. Brandon Wood was a great minor league hitter, but I'm sure scouting along with statistical analysis can be used to project a Paredes more comparably against a Brandon Wood these days.

As stotle has verified, the process of scouting itself is well ingrained with statistical methodology in many aspects. As he also stated, field scouts opinions are one of many data inputs in evaluating minor league players. We can guess what those other data inputs are aand how much they are weighted. My guess is they have a lot of data. I would say the calculus is a bit more refined than Paredes chances may not be zero.

That said, I have no idea where Paredes might eventually settle in. If I had to guess it'd be a high 700 OPS guy that may need to be platooned against LHP.

[TAMC]

Down home we prefer "Figures don't lie, but liars figure." I am usually an interested observer to these discussions, but as a person who has used models and statistics routinely over a career lasting much longer than I like to think about, I have found that when considering complex systems (and humans are about as complex a system as one will ever encounter), both observation and modeling/statistics are necessary. Statistics, as used in this thread, is a way of characterizing a large sample of data involving complex interactions in a consistent and unbiased way to extract as much information as possible from the data. Values so derived are perhaps uncertain, but if the underlying data is valid, the values derived cannot be viewed as incorrect. Similarly, human brains can process a huge number of variables and condense them into a value judgement. This comes with a level of uncertainty that is much more difficult to quantify than for statistical analysis. My professional experience has been that when the two methods are in close alignment, you can be pretty certain of the result. When they do not align, a smart person tries to understand why they don't align to avoid being mislead by one or the other. It is seldom a positive to say one is correct and the other incorrect until you know why they disagree. I'll go back to sleep now.

[CA-ORIOLE]

Down home we prefer "Figures don't lie, but liars figure." I am usually an interested observer to these discussions, but as a person who has used models and statistics routinely over a career lasting much longer than I like to think about, I have found that when considering complex systems (and humans are about as complex a system as one will ever encounter), both observation and modeling/statistics are necessary. Statistics, as used in this thread, is a way of characterizing a large sample of data involving complex interactions in a consistent and unbiased way to extract as much information as possible from the data. Values so derived are perhaps uncertain, but if the underlying data is valid, the values derived cannot be viewed as incorrect. Similarly, human brains can process a huge number of variables and condense them into a value judgement. This comes with a level of uncertainty that is much more difficult to quantify than for statistical analysis. My professional experience has been that when the two methods are in close alignment, you can be pretty certain of the result. When they do not align, a smart person tries to understand why they don't align to avoid being mislead by one or the other. It is seldom a positive to say one is correct and the other incorrect until you know why they disagree. I'll go back to sleep now.

I generally agree with this with the exception that baseball performance and evaluation is a far more volatile and less predictable endeavor as compared to more standard business applications .

[Phantom]

I generally agree with this with the exception that baseball performance and evaluation is a far more volatile and less predictable endeavor as compared to more standard business applications .

Big data has revolutionized much more complex systems, like the stock market.

[CA-ORIOLE]

Big data has revolutionized much more complex systems, like the stock market.

Not sure it's a great example. Outside of extraordinary circumstances and insider trading/manipulation, the stock market (at least the broader market) is a very predictable model along with economic forecasts. Maybe the Stock market (and economics) has been revolutionized from the 1880's - 1920's (and later) but that has been because of strict control and regulatory effects.

I don't see any real parallel to the process of evaluating individual baseball player performance, particularly minor league player performance and development. Now I would agree that aggregate financial risk models in baseball have become far more precise.

[Herman]

So?

He is still in more suitable replacement than Markakis.

With that said, they have better options on the club.

So what are you trying to say that De Aza was a selfish guy? I suspect if that was the case, they wouldnt have brought him back.

No, De Aza and selflessness were two different point. I pointed out Buck's preference for selfless players b/c selflessness isn't something that you can plug into an equation.

I mentioned De Aza b/c the saber-metrics/stats projected him to be a suitable replacement for Markakis and he was just DFA. De Aza was also a low baseball IQ. Can baseball IQ get measured in an equation? Are you suggesting that the Orioles have a suitable replacement for Markakis on the team? For the record, I don't think that Markakis is anything special and I certainly don't think that he was worth the contract that the Braves gave him. The current platoon doesn't make the team any better.

[Can_of_corn]

No, De Aza and selflessness were two different point. I pointed out Buck's preference for selfless players b/c selflessness isn't something that you can plug into an equation.

I mentioned De Aza b/c the saber-metrics/stats projected him to be a suitable replacement for Markakis and he was just DFA. De Aza was also a low baseball IQ. Can baseball IQ get measured in an equation? Are you suggesting that the Orioles have a suitable replacement for Markakis on the team? For the record, I don't think that Markakis is anything special and I certainly don't think that he was worth the contract that the Braves gave him. The current platoon doesn't make the team any better.

They did?

Looking on Fangraphs they have him projected:

Zips - .9 WAR

Steamer- .1 WAR

Depth Charts- .1 WAR

Me, I didn't really want him back.

The stat guys I read thought the O's should replace him with Aoki, who got a 1/4 contract.

[CA-ORIOLE]

They did?

Looking on Fangraphs they have him projected:

Zips - .9 WAR

Steamer- .1 WAR

Depth Charts- .1 WAR

Me, I didn't really want him back.

The stat guys I read thought the O's should replace him with Aoki, who got a 1/4 contract.

There was nothing wrong with logic of projecting De Aza as suitable platoon player and corner outfielder. I don't recall too many people clamoring for Aoki at all prior to the season. I'm pretty sure zips projected him much higher earlier in the season than they do now, particularly that his fWAR was so much more favorable.

[weams]

There was nothing wrong with logic of projecting De Aza as suitable platoon player and corner outfielder. I don't recall too many people clamoring for Aoki at all prior to the season. I'm pretty sure zips projected him much higher earlier in the season than they do now, particularly that his fWAR was so much more favorable.

There is about a weeks worth of data different between Aoki and Snider right now. When the season is all done? Might be different.

[Can_of_corn]

There was nothing wrong with logic of projecting De Aza as suitable platoon player and corner outfielder. I don't recall too many people clamoring for Aoki at all prior to the season. I'm pretty sure zips projected him much higher earlier in the season than they do now, particularly that his fWAR was so much more favorable.

I read a number of pieces during the offseason comparing Aoki and Markakis with folks projecting similar production at a fraction of the costs in terms of money/years.

[CA-ORIOLE]

I read a number of pieces during the offseason comparing Aoki and Markakis with folks projecting similar production at a fraction of the costs in terms of money/years.

Fine, I don't recall it being much of an issue of discussion with stat people on here.

[Can_of_corn]

Fine, I don't recall it being much of an issue of discussion with stat people on here.

His name was mentioned as a candidate but you are right there wasn't a huge push for him. Sorry if I implied there was.

[CA-ORIOLE]

There is about a weeks worth of data different between Aoki and Snider right now. When the season is all done? Might be different.

Seems like Snider is driving the ball better. So maybe some more regular time and adjustments will help. Batting order generally isn't that big of a deal but I like him in the 2 spot where he may more regularly get a shot at the right side hole. SF seems to have a way of turning signings into gold. Remember how many people wanted Morse. How bad would that one look right now at 2/16.