His ERA was over 2 full runs higher in 1998 before the trade compared to 1997 but, sure, his stats were right in line.

No one else does? LMFAO. What'd you do, filter out everyone's posts but mine? This entire thread has been others posting their opinions, them me providing my opinions, others shitting on me for having them, and then me throwing shit back.

And what about this very post of yours? Not a dickish post?

I tried to find an article about Randy throwing games and faking injuries to get out of Seattle. I couldn't even find one. Sure seems like you guys are just making shit up. I did, however, find articles about him having back surgery. Twice.

His K/9 was 12.0 while he was in Seattle in 1998. It was 12.3 the year prior. His BB/9 was 3.4. It was 3.3 the year prior. And while, yes, his ERA was 4.33, his FIP was 3.35, almost a full run lower, and is also right in line with his FIP from the two seasons prior to that. Which means a full run per game of that 4.33 ERA was due to circumstances outside of his control.

I'd wager good money that this rumor about him throwing games and faking injuries (if it even was an actual rumor) came about because people who don't understand variance and sample sizes looked at the borderline irrelevant discrepancies between his ERA in Seattle and Houston that season and just pulled that explanation out of their ass because that's how stupid people attempt to explain away variance.

I'll say it again. Stop looking at wins, and stop looking at ERA if you want to evaluate pitching performance. I understand that this may be a difficult habit to break because it's been pounded into your heads for decades, but all it's doing is confusing you, whether you recognize it or not.

Your second point is valid. My apologies.

Randy also posted a 0.337 BABIP during his time with Seattle in 1998 before being traded, which was almost the highest BABIP he posted in his entire career. This means that the ever so slightly elevated numbers he posted that season in Seattle were entirely explainable simply by bad luck. Nothing about his statistics from 1998 are indicative of him throwing games or pitching worse than he was capable of during his stint with Seattle. This is not just my opinion or me trying to say something controversial. It's a simple fact. If you disagree, you simply don't understand how statistics works with sample sizes, variance/luck, and confidence intervals.

This is the type of nonsense I'm talking about. The fact that he posted an ERA of 2 runs higher than he did the year before in Seattle at the trade deadline literally means absolutely nothing. It tells you nothing about how well he was pitching without also looking at his other stats like FIP, BABIP, K/9 and BB/9. The data clearly shows that he simply got unlucky over the course of the first 2/3s of the season but was still every bit as dominant with the factors that were within his control.

Beware of Stats.

1. that do not always mean anything
2. Different people value different stats differently
3. Some people use the Same Stats and read it differently to make their points

Also as Wilt Chamberlain once said, "Nobody roots for Goliath"

Or more colloquially…

https://www.goodreads.com/quotes/tag/statistics

I thought a pitcher's BABIP could be indicative of his ability to induce weak contact and therefore having a higher one than typical could indicate he was not pitching as well as before and not just random bad luck.

Anyhow I guess his bad luck just disappeared the day he was sent to Houston and his BIP then dropped by over .3 for the rest of the season. Just regression to the mean, inconsequential.

Those quotes are hilarious..............and at the same time, pretty much spot on!

So true, so true.

63% of all statistics are fake.

One more famous one that didn’t make the first page cut:

“Statistics are like a bikini. What they reveal is suggestive, but what they conceal is vital”

Aaron Levenstein, economist, November 1951

As this thread shows all too well, raw data may be objective, but how one presents and interprets it is anything but.

Yeah Peter, I vaguely seem to remember someone on here posting about how they used that same statistic to declare that Spahn was just lucky and not so great a pitcher. Good thing his luck didn't last for more than 363 wins or 20 or so successful winning seasons. Otherwise, people might actually begin to question the infallibility of such a statistical measure and the thoroughly conclusive answer that can be gleaned and ultimately drawn from it in regards to something as readily and objectively measurable as the level of Spahn's luck. But of course, we all know that a loss of faith in statistical infallibility would never happen, especially when such a conclusive and non-arguable determination of Spahn's luck is drawn from such totally unbiased and all-inclusive statistical and intangible data by qualified data scientists in their unrivaled capacity and knowledge, the results of which they so unselfishly bestow upon and honor us with. For without them, we, the unknowing minions, would truly be in the dark. But fortunately for us, we have our beloved statisticians and data scientists to instead smile down upon us with their sage and benevolent wisdom, and lead us unto the light! Amen.

(Shoot! Did I just cross a line and post something religious? If so, my apologies, no offense meant to anyone............well, almost! :rolleyes:)

Oooooo, I like that one.

Trying to keep it in a baseball vein, here's one I think would have been a classic Yogiism:

Statistics can always tell you everything you want to know.........about half the time!

Again, so true.

Which is fine, as long as the interpretative part isn't presented as infallible objective truth. PS if you ever want to see people spin data (one could in some cases say manipulate), look at some clinical trial results sometime, including for some of our favorite drugs.

"ever so slightly elevated numbers", LOL.

If 2/3 of a season is too small a sample size to make any meaningful observations that exclude good or bad luck, as apparently is the case for Randy Johnson 1998 and his slightly elevated first 2/3 numbers, why is a full season really that much better? Maybe we should just junk the Cy Young award, since it's just rewarding randomness.

Yes, it did. This happens all the time. He even got lucky in Houston. His HR% was half of what it normally is for him during those 10 starts. Anything can happen in 11 starts. It's just way too small of a sample size. Hell, nearly anything can happen even over the course of a full season too, let alone 11 games.

A pitcher's BABIP is almost entirely outside of his control. There are some who suggest that they may be able to exercise some minuscule amount of control over it, to the tune of a few points, but that's not an easy sell even at that. Either way, large fluctuations above and below the league average BABIP is indicative of a pitcher having gotten either lucky or unlucky that season. Just go look up your favorite 10 pitchers and look at their best and worst seasons with respect to their ERAs and WHIPs. You'll usually find that those were usually just seasons where every bounce or wind gust went their way (or failed to when their numbers were "bad"). Especially when there is a discrepancy between their ERA and their FIP.

If I want to know how well a pitcher performed, I look at the stats that are within their control.

Your logic would mean there's no such thing as a great pitcher who pitches to contact, or who isn't a dominant strikeout pitcher, because once a batter puts a ball in play it's all just dumb luck. That just does not square with experience. Did you ever watch Greg Maddux pitch?

Put another way, putting a ball in play on a pitch that was a hanging curve or a fastball with no movement down the middle is just not the same as doing so on a wicked slider two inches off the plate. A great pitcher can throw more pitches that are difficult to make solid contact with and thus your chances of getting a hit off him on a ball you put into play is not just random or some stat that will eventually hit the mean.

Absolutely. I have no problem with using statistics, they are a vital and necessary tool in helping to find solutions, trends, and so on, over an infinite number of situations and questions. And the people who have learned and mastered this mathematical science can come up with and discover some truly amazing things. But for the most part, you and I both know statistics are rarely, if ever, going to be 100% accurate and/or predictive in their results. And as you pointed out, can be (and many times are) able to be manipulated. The problem isn't in the statistics themselves, its with the people who ignore the shortcomings of relying solely on statistics without taking into account inherent weaknesses and bias in the data they're using, or that manipulate it, knowingly or unknowingly, for their own purposes or ends.

And speaking of manipulating data, reminds me of an old accountants joke. Owner of a company needs a new accountant, so he puts an ad in the paper (I did say this was an OLD joke). Later that week, he starts having people come in for interviews. And at the end of every interview as the applicants get up to leave, he always asks them one last quick question. "What's 2 + 2 equal?" And invariably they all they all give him the correct answer of 4. So he shakes their hands, thanks them and says he'll be in touch, and they part company. Now its Friday, and the owner has been at these interviews all day, and still hasn't found an applicant he really likes for the accountant's job. He's tired, but has one has last interview for the day. So the applicant comes in, sits down, and they start. Interview goes okay, like pretty much all the other earlier ones. And as they wrap it up and the applicant starts to get up to go, the owner asks his same final question. "By the way, what's 2 + 2 equal?" To which the applicant quickly replies, "What do you want it to be?" And as he then goes to shake the applicant's hand, he smiles and asks one more question. "When can you start?"

Careful there Peter, you're using too much logic and sense for some people!

One of pharma's favorite spins is relative vs absolute risk reduction. If I told you I had a pill that could cut your risk of serious adverse event X by 50 percent, you might be like, wow that's impressive. If I told you it could cut your risk from .001 to .0005, you might be, well is it worth the risks and side effects? Same data, different look.

Isn't that the whole idea behind good knuckleball (or sinkerball) pitchers? Don't let batters get good contact on balls so they hit mostly pop us, weak grounders, or foul balls. But that isn't really cool anymore so no one really tries mastering it. Nowadays seems like everyone wants the studs throwing 100 MPH that strike everyone out, so are more and more the kinds of pitchers you see coming up.

I also question some "oh it's too small a sample size" arguments. Those always seem to me to reflect cherry-picking, to dismiss inconvenient stats that don't fit the theory. We used to see that argument all the time here to rebut the theory that Kershaw was not a good post-season pitcher; his lousy performances were just random events and couldn't possibly reflect that he wilted under pressure. Of course after a full season worth of postseason outings there's still a huge disparity so maybe that argument has been retired.

Of course when the stats do fit the theory, we don't see the sample size argument so much.

I've watched enough baseball games to be pretty damn sure there are pitchers who are harder to hit solidly whether you put the ball into play off them or not, but maybe it's all an illusion. Incidentally Maddux' BABIP for his career was 9 points below the league average, and if you exclude his end of career seasons probably a bit better than that. That seems significant to me but I don't know. Kershaw -- 23 points below.

And therein you have so elegantly described the beauty, and the curse, of statistics, in one short, sweet, simple statement. Kudos.

Statistics don't lie. People misusing them to support their narratives or stripping them of proper context do.

Laugh all you want, but the numbers that actually matter, the ones that aren't subject to luck were nearly identical to his prior years (12.0 K/9 vs 12.3, 3.4 BB/9 vs 3.3, 3.35 FIP vs 2.82 and 3.42). If that's not "ever so slightly elevated", then I don't know what is.

But you can keep looking at ERA if you want to. The imbeciles always do.

Those are the product of the same set of games you dismissed as being too small a sample size to be meaningful, so why are they now meaningful when they favor your argument?

I realize people think that, but the fact of the matter is that yes, it is dumb luck. The idea that Greg Maddux was better able to control the fate of the ball after contact than his peers is not supported by the data, though often claimed. His BABIP numbers across the course of his career are precisely in line with league average BABIP numbers during that era. If you don't believe me, I can plot it for you.

No they are 9 points lower, I already posted that.

If your thesis is that Greg Maddux' career (after all he was not a dominant strikeout pitcher with 6 K/9) was jut the result of dumb luck, you have pretty much disqualified yourself as knowing anything about baseball, however good you are with data.

If it gets awarded to whoever has the best W/L record and the best ERA, then yes, it is borderline meaningless. As has been mentioned earlier in this thread, the award has gone to countless pitchers over the years who clearly did not deserve it.

But it wouldn't be that difficult to make it meaningful. They just need to look at the right statistics. They finally figured this out for offensive players. Maybe they'll come around sooner or later with pitchers too? Who knows.

Sample size is just as small for all statistics, no?

Because the sample sizes needed for those statistics is remarkably smaller than it is for ERA, and thus their corresponding confidence intervals are much narrower. This is because they are not nearly as subject to luck as ERA is. It takes years for ERA to converge. It takes months for K/9, BB/9, and even FIP.

Oh, really?

And by any chance could the fact that pitchers end up facing the best teams during the playoffs and WS possibly be a factor in their not-so-stellar playoff performances also? Or how about those pitchers where age, injuries, the long regular season, and other factors would then catch up to them at playoff time? Corey Kluber may be a great example of that particular set of circumstances. A fantastic pitcher and Cy Young winner, Corey was always focused and ready to go. But as he'd head into the postseasons, it always seemed like he'd run out of gas or nagging regular season injuries finally caught up to him. Still, when he was right and rested, he could hang with the best of them.

The one other thing you said that I really like is in reference to the possibility that someone like Kershaw could also possibly wilt under post season pressure. Now I'm not saying he does, but I think ALL people realize and recognize that stress can and does affect every single one of us, and does so differently depending on the unique set of facts and circumstances a person is presented with. That is a prime example of one of those human elements I keep harping on about that statisticians can't possibly ever effectively measure and fully account for in their equations, formulas, and algorithms. They may try to tell that they in fact do have such variables and factors accounted for, until you ask them to show you and prove it to you, and then you get the excuses about how you wouldn't understand, or it would take too long, and so on.

Now I actually don't doubt that statisticians may in fact try to account for human variables like luck, stress, heart, competitiveness, or whatever, but since there is no real way to effectively measure and quantify such human variables in their work, the things they do to account for them are at best, SWAGs, and at worst, WAGs. And a guess is basically no more than someone's opinion, maybe an educated one, but still an opinion, no more, no less. Not ever saying a statisticians work is bad, just that they should be honest and admit what it actually is, their (very) educated guess more often than not.

Is there a statistic that tracks the deer in the headlights look that (unfortunately, as I like him) Kershaw gets time after time after time in the post season?

Peter,

I'm with you. It's the interperative abilty of some people that seems to totally fail them. Oh, they can compile the data and create the formulas and all, but they still have to then determine and interpret the results and how truly meaningful they really are.

Example, there are a ton of different tax programs out there that are used by accountants, and some are definitely better (or worse) than others. Truthfully, most accountants/CPAs will tell they all have issues and could use lots of improvements to work better and more effectively. But the problem is these programs are created by programmers, not the accountants/CPAs that most often use them. And when we complain to programmers about issues, shortcomings, and errors in these programs they invariably give us their excu.....er, reasons, for not being able to really change anything because they are the programmers, we are not, and they know what they are doing so that is how it is. Hard to believe there could be so many variations in tax software out there when they are all supposedly trying do the same calculations across the board. One would think all tax programs should pretty much be exactly the same, basically this minus that times this rate = income taxes due, right? But its not, because programmers know programming, not taxes. And each different programmer puts their own unique thinking, biases, and such into the tax software product they create. And that's why their different tax software can end up being easier or harder to use than others, can do more or fewer things, and can even come up with completely different tax liability results.

Now go back and swap statisticians for programmers, and ask them to develop their formulas and equations to determine who the greatest lefty pitcher of all time is, instead of how to figure out what your income taxes will be next year. Want to guess how many statisticians will come up with different equations/formulas, along with different answers to the question, especially since each statistician will likely complete their assignment using their own definition of what "greatest" means, without ever asking what you or anyone else thought or wanted it to be?

I was told Maddux' BABIP "tracked precisely" the BABIP's of his era. I'd still like to know if -- given that in fact his career BABIP was 9 points lower than the average -- that is tracking precisely or not. Hard to explain away 23 years or whatever it was as a small sample size.

The other thing is, his BA against was 14 points lower than the average, even though he wasn't much of a strikeout pitcher. How does that happen?

Statistics don't lie Peter, so you've got to be right! LOL

Maybe a better question to ask some statisticians is what day of the week it is (or maybe what hour of the day), because that's how quickly their sample sizes and other statistical arguments seem to change in this discussion.

So a 20% jump in FIP is "ever so slightly"?

His home runs were up 40%. OPS was up 20%. Line drive% was up 33%. All that stuff says guys were hitting the ball hard off of him A LOT more than the previous year - and the rest of 1998 and the next four years.

Please specify the EXACT number of starts and/or innings to qualify as NOT a small sample size. Just for grins.

p.s. Name-calling reflects poorly on you. Do better.

Irrelevant man. Once the ball leaves the bat, it has nothing to do with the pitcher, it's just dumb luck and coincidence, or a function of other factors beyond the pitcher's control such as fielding. BABIP is binary -- either it's a hit or it's not -- and the rest is meaningless. Besides, every number you cite suffers from small sample size and confidence interval issues -- only numbers that support the thesis that his first half was just random are reliable.

All pitchers are essentially fungible beyond their ability to strike batters out. Maddux is the same as any other 6K/9 pitcher you can name. Johnson's K's stayed up, so all the rest is noise.

If it was all dumb luck, then BABIP would, over the course of long careers, all come out about the same when you factor in the defense behind the pitcher. There would not be pitchers who have incredibly successful and long careers, not giving up many runs, while being contact instead of strikeout pitchers. It doesn’t.

If it’s all dumb luck, how are contact pitchers often just as successful as strikeout pitchers? Maddox and Randy Johnson put together similar total careers. Johnson’s BABIP is league average, Maddux, like most hall of fame contact pitchers, is well below it. They achieved similar ERA’s and total careers via very different methods, in huge sample sizes.

I appreciate you clearing it up for me. It seems so simple when you explain it that way.

This is way off the subject. You guys are probably boring other readers. Let’s end it now. Koufax is the GOAT period. Good night

Maddux's success was the result of dumb luck!! 21 years of it, well, maybe 16 of which really defined him.

Koufax is the GOAT period. Another stunning Koufax argument. We’re going to hit 2,000 posts without a logical argument for Koufax actually being presented.