THE BOOK--Playing The Percentages In Baseball

On the radio this AM:

“Giants drop their opening series to the Dodgers and are now 1-3.  But don’t worry, there are lots of good teams that have losing records right now.”

Great!

We could get into a lot of arguments around the semantics of “pace.” Let’s flip it to RBI instead (since people actually have made it to 162 RBI).  How many RBI are needed through n games to be “on pace” for 162 RBI in a season?  I personally think it would be strange to answer anything other than n.

Now I fully understand that an observed n RBI in n games does not mean one should be projected 162 in 162, but I still think it’s fine to say that one is “on pace” for it.  I think the real issue is convincing people that the pace is unlikely to be maintained (which can lead to discussions of sample size, regression to the mean, Bayesian probability, etc. if needed).

The problem with saying that someone is “on pace for” 162 RBI, say after 30 games, is that there have been DOZENS, if not hundreds of players “on pace for” 162 RBI, and yet none of them eventually made it.

What’s the point of saying that someone is “on pace for” something if the chance of that thing of actually happened, based on actual historical precedent, is zero percent?

You may as well someone say is on pace for a 1hr 40min marathon after the first mile.

You can say he’s hitting “at a rate of”.  But, no, not “on pace for”.

My kid is “on pace for” getting A+ throughout high school and college, even though he’s in elementary school.

Somehow, we apply to baseball what we would never apply to anything else.

The “on pace” also happens in the draft.  Pierre Turgeon is “on pace to” become the next Gretzky.  Lindros is “on pace to” become the next Mario.  Crosby is “on pace to” become the next Gretzky.  Spezza, Daigle… the list is endless.

And they say this BEFORE they ever play one NHL game.

They don’t say “on pace to”, but they may as well.  They say “could be the next...”.  Well, there have been a dozen “could be the next...”.

Why don’t they say: “Crosby is the next who could be the next...”

We did it on Thursday with the brewers opener and back-to-back homeruns to open the game.

“Quick… somebody… two homeruns and no outs so far, the brewers are on pace for… how many homeruns this season… what’s two divided by zero?”

So to get into those semantic arguments mickey seems want to avoid, we de-power this negative language, “on pace,” by mocking it. Works in my little baseball circle, which is good enough for me.

"On pace to” is a perfectly reasonable comment, that doesn’t have to mean “Projected to finish with”.

I hate to defend it....but sometimes people use ‘on pace to’ as a way to put the season to date into context instead of predicting the future. For example, everyone understands what 30 home runs, 100 RBI, 100 runs, whatever means in the context of a 162 game schedule. But sometimes it’s not as easy to understand level of performance 38 rbi in 68 games represents.

For that purpose, I think it’s helpful (for the less mathematically disadvantaged anyway.)

I think this is a battle not worth fighting--the use of “pace” as popularly used is not technically incorrect, and no one really thinks that players are going to keep current outlandish paces.

This is, however, a good way to introduce someone to regression to the mean.  Ask someone why they don’t expect Heyward to hit 162 HR/season.

btw, there have been 20 162+ RBI seasons; Manny’s 1999 (165) is the only one after the 1930s.

#8, I think the problem is that the “on pace” statements don’t properly establish context.  They have a very specific meaning (the player’s current rate for the season extrapolated to 162 games), and that meaning is completely separate from our understanding about the context of the full-season numbers.  It is when people try to use the context of the full-season pace to infer something about the current numbers that an issue occurs.

If you want to talk about someone with 10 HR through a third of the season, using a 30 HR full season is not the right context.  Everyone understands the meaning of a 30 HR season, but that meaning is not the same as a 10 HR third of a season.  The latter is more common.  Using a fan’s knowledge of the meaning of a 30 HR season can lead him/her to infer that that is the context of a 10 HR third of a season when really it isn’t.

The proper context would be what Tango is talking about, where hitting 10 HR in a third of a season is comparable in context to maybe 25 HR in a full season (or whatever the number ends up being).  That’s where you see how rare/impressive a partial-season performance is in the context of full-season numbers that are easily understood.

We do that mocking thing too.  Like “game 2 is the most important game in a 7 game series”.  Then “game 3 is the most important game in a 7 game series”.  Then “game 4 is the most...” and so on.

Early on pace numbers are not meant to be predictions, they are meant to demonstrate how hot a batter is hitting. It’s just a bit of fun.

@#10:

I didn’t mean to imply that a player who hits 10 HR in 1/3 of a season is as VALUABLE as a player who hits 30 HR in a full season. It’s just an easy and accurate way of describing a rate using a unit (unit in this case being a full season) that most people can relate to.

In my experience, when folks do it after an extremely small sample size, it’s always been in jest anyway.

Agree with those of you who think that Tango is misunderstanding the usage here.  “On pace for” is not meant to be a prediction.

I’m not talking about value either.  I disagree that it is an easy and accurate way to relate to partial season numbers because the relations that we assign to full season numbers are not accurate for partial season numbers at the same rate.

This thread is on pace for about 300 responses by the end of the week.

I agree with what many are saying in this thread.

When I see a bunch of statistics through 40 games, say, it is hard to put that into context.  So I mentally multiply all numbers by 4 to get a sense of whether the numbers are good/bad etc.  It is not a way to “project” their stats over the course of a season, but to gauge their progress so far in the season.  It’s like “Oh, 5 homers through 40 games is only about 20 in a season so he hasn’t been doing too well so far.” I don’t put anything into it further than that.

It’s just a misapplication of the term.  It works fine in a marathon to say someone’s on pace for 2 hours 10 minutes or something, because the “results” in a long distance run flow smoothly from start to finish.  For home runs, the results don’t flow, they come in discrete arrivals, one at a time.

On the other hand, if someone sprinted the first mile of a marathon, you might say “with that 4 minute mile, he’s on pace for a 1:48 time”, but everyone would know that such a pace couldn’t be sustained.

I think the same goes for home runs: no one who uses the “on pace for” lead in (and I forget, did anyone actually say this or was this a pre-emptive grumble?) really thinks Ian Kinsler is going to hit 162 homers, it’s just a clever way to highlight that someone has had a hot first few games…

"But his actual pace is to do whatever it is other people who started with 3 HR in 3 games actually did.”

Nelson Cruz now has hit a HR in each of the first 4 games.  Of all the other people who have done this in MLB history, 50% of them have ended the season with 70 home runs and such players have averaged 44 home runs.  Therefore I conclude Nelson Cruz “on pace” to hit 44 and is a coin flip to hit 70

Mickey, I was just about to post that!  Seems very appropriate to Tango’s point.

I thought it was amusing too, that the Sportscenter anchors said “McGwire and Mays, elite company!!!!!” and yet, Mays only hit 14 more the rest of that season.

That’s fantastic.  I think it would be great if they actually did show that, that for the season, they showed one with 70 and one with 18.  Did they?

"What’s the point of saying that someone is “on pace for” something if the chance of that thing of actually happened, based on actual historical precedent, is zero percent?”

Because it’s whimsical fantasy? ... and that can sometimes be fun?

Isn’t it accurate though? If Kinsler keeps up his current “pace”, he will hit 162 HR. We just know the likelihood of that is zero, and the reality is implied. That’s why it’s a fantastical, almost jestful, “on pace” comment.

“You may as well someone say is on pace for a 1hr 40min marathon after the first mile. “

Aren’t they? But, you’re right, you might as well say it ... it’s a similar thing. The “pace” comments at this point are all fantasy.

What Mickey did is more valuable, more interesting (to some, including me, more informative ... but a lot, lot, lot less fun.

April is all about “fantastical pacings/projections”. Don;t even get me started on how many games the Orioles are going to win, because right now they’re on pace to win ...

If it were a whim, why are we bombarded with it, and why do fans panic so much?  No, it’s not sold as a whim.  It’s sold as something real.

***

For the 4HR in first 4 days thing: to simply highlight it as something that was only done twice is a great thing.  And, to show that the end-of-year HR was 18 and 70 is fantastic.  Right there, it shows it’s trivial, and it shows how much range you can expect.

Tango/21, they did show it.  The graphic said:

Cruz ?
McGwire 70
Mays 18

But the only thing they said aloud were the names of McGwire and Mays.

How fortunate though… I thought it was an interesting graphic for one reason, and you expanded my understanding of the possibilities of such an analysis.  Great stuff!

Albert Pujols is on pace to hit into 84 double plays and hit 400 ground balls this year.

We’re through 4 games, how much on pace (serious) stuff can we really be exposed to? Boston is going to win 40 games? Baltimore is going to win 162?
Texas is going to hit 500+ HRs? Kyle Drabek is going to win 40 games?

I guess I never viewed it as serious, until we start reaching the ASB and people start “doubling the half’s stats to project the season total”.

I agree with you in regards to how interesting the regression and variance issue is. I just don;t think many folks put much into the “pace” comments ... especially in the first week.

Now at the end of may, some of the pace comments might be arriving at the “taken too seriously” stage.

At this time of year, I always enjoy the ERAs of infinity, 108.00, 81.00, and the like.

I can’t agree with Tango when he says, “You can say he’s hitting ‘at a rate of’. But, no, not ‘on pace for’.”

Merriam-Webster’s first definition of ‘pace’ is ‘rate’: http://www.merriam-webster.com/dictionary/pace

"at a pace of”
and
“on pace for”

is not the same thing.

Perfect!

Tango, look at the example in the definition. They use “on pace to”.

"on pace to set a record”

“rate of movement; especially : an established rate of locomotion “

It seems to me that the dictionary definition shows that “pace” is a subset of “rate”.  While rate can mean alot, pace is more specific.  It’s an establish rate, it’s a rate that leads to something in particular.

Rate is more generic, and pace is more specific.

In any case, if you think the speaker means “pace” as a “generic rate”, then I am wrong.  In my view, the speaker means “pace” as something more specific, if not predictive, than at least highly suggestive.

I think Tango is suffering from Chinese jibberish syndrome here.

He thinks the MSM is saying they think Kinsler would hit 162 home runs when really none of them would actually say that. They all really do understand that he’d come up well short.

No, I’m not saying they think that.  But they say it in such a way as to make it a big story, when really there is ZERO story.

“Kinsler has 2 HR in 2 games”

-- “Yeah, that’s good right?”

“Did you know that if you take two, divide it by two, and then multiply it by 162 you get 162?”

-- “Really?”

“Right. So, we can say that he’s hitting HR at a pace of 162 HR for the season”.

-- “Really?”

“Yes.  And, if you figure Kinsler has 10 more years, we can multiply that number by 10, and get… let’s see… 1620.  So, he’s hitting HR at a pace of 1620 for the rest of his career.”

-- “Really?”

“That would put him ahead of Hank Aaron and Barry Bonds… combined!”

-- “Can you stop now?”

***

If it was whimsical, it would be one thing.  But these “pace” gasbags annoy me.

This “on pace for” writing/reporting line reminds me of all the stories in spring training about some player coming to camp in great shape. Whether such a story is being reported as news or used by a blog to critique reporters, I think it tells me what what I need to know about the writer/reporter: S/he would rather beat a dead horse than try to find something informative to cover. For me, such works are usually a strong signal that I should look elsewhere for content to read or watch.

I like that analogy.  They may as well say “because he’s in the best shape of his life, he’s on pace to have one of his best seasons”.  That is the implication.

"But these “pace” gasbags annoy me. “

Heh.

I think it’s hilarious/awesome/ridiculous that such a thing would get to you.

Now, every time I feel the need to aggravate someone or just try and piss in one’s oatmeal ... I know I can always come here and make some comments about a player being “on pace” for something historically great.

Did you know the Rangers are on pace for almost 500 HRs? Isn’t that awesome?

They annoy me like mosquitos.  I try to turn off the volume when I can.

I couldn’t find a place to post this, but I thought all readers of this blog would appreciate this note on Bonferoni corrections

http://xkcd.com/882/

Seems like a minor, even trivial thing to get annoyed about, considering they’re probably not being 100% serious.  However, I do agree that there is no need to say such things.  The problem is context which Kincaid mentioned above.

For example, if what the Rangers did to the Red Sox happened in the middle of the season, nobody would think that the Rangers suddenly had an increase in ability and then extrapolate their season HR total using this new level for the rest of their games.  But for some reason, at the beginning of the season, all of the context of last year, the year before that, even entire baseball history gets washed away, so that everyone gets a clean slate.  Nobody should buy that.

Although, I do have Teixeira on my fantasy team.  So I guess I should expect him to put up a monster year now that the “notorious slow starter” is off to a great start. 

Bobzilla, at the risk of stating the obvious, you should expect Tex to have a better season for HR than you did a week ago.  Take whatever you would have projected for a 157 game season (which ought to be 97% of a 162-game season projection), and then add 4.  It’s bigger than whatever you expected from a 162-game season before Opening Day, by something in the neighborhood of 10%.

You can’t do this with things like running races, where an outstanding early split is often purchased with energy that won’t be available later, but Tex’s 4 HR’s are in the book, and he’s got 157 more games (less injuries and such) to add to them.

I get all of that.  A hot start will increase a player’s projection in the manner that you talked about.  My example of Teixeira is a little different than what Tango was initially talking about too.  But there are people (I know some) who think of him as a slow starter, and consequently believe that if he could get off to a good start that he would put up a great(er) season because they hold what he’s done in the 2nd half constant.

Right.  Some guys are notorious slow starters… until they aren’t, at which point someone explains in retrospect what was different.  When they tell that story, they may be right, or it may be noise, there are just too many moving parts to know…

I’m not sure, though, that I would want my MSM sports reporters to totally surrender, stop looking for cause & effect explanations and just robotically chirp “sample size, regression to the mean” for the first 3 months of every season.  That might be more annoying than the “pace” stories…

I agree that there could be something about a specific player that causes him to be a slow starter, even for a specific period of time and not their entire career.  But it could just be random variation.  You’re right though, that I don’t know.  And I don’t pretend to know either.  But I’d rather label someone a slow starter, which kind of suggests some sort of failing in them.  Yes, Texieira has had some slow starts, and if it’s possible, we should try to find out why.

*rather not

Another thing is when they flash their current-season numbers (and completely ludicrous in the playoffs).  If it were me, I’d show the numbers for the last 12 months.

Everyone is looking for a “story” in the first two months, and of course, you will rarely get that story.  You basically end up with about 200 stories by June 1, of which 20 will eventually actually end up being stories by the end of the year.

But, you show stats for the last 12 months, and then real stories will happen eventually.  Fangraphs has a great tool, where you can see the actual last calendar year as I’m talking about:

http://www.fangraphs.com/leaders.aspx?pos=all&stats=pit&lg=all&qual=y&type=8&season=2011&month=10&season1=2011&ind=0

We see Jered Weaver is not 2-0, but 15-12 (and he’ll go down to 14-12 when you see the above link by tomorrow, as his April 5, 2010 start disappears).

(That the two seasons started one week apart means we have more than 162 games to contend with.  You can of course adjust it so it’s the last 358 days, or last 162 games, if it bothers you.)

I know, sometimes that means we’re going to miss the rookie seasons like Heyward and Strasburg as they play catchup, or we’ll miss breakout seasons like Cliff Lee, and so on.  But against that is all the non-stories sold as stories.

Sometimes the commentary walks the knife edge between an acceptable reporting of facts and unwarranted projection. 

If someone says “Tex has been a slow starter”, that is a fact which is describing his relatively poor performance early in the season, and this observation should not send anyone’s blood pressure through the roof. 

If, however, they say “Tex is a slow starter”, then they are suggesting that this past performance is a genuine trait of the player, which may or may not be true (default to “not true").  Commentary like this, absent some seriously in-depth and valid research, probably should make one’s lips pucker a bit.

It reminds me a bit of RBI’s.  It’s fashionable (and not without merit) to run down the importance of RBI’s these days, but some people tantrum at the mere mention of RBI’s, i.e. “Since last June, Delmon Young leads all ML hitters in RBI’s” ---> “You’re crazy , his WAR is -XX.X, his wOBA is .XXX, what’s the matter with you, blah, blah, blah...” It’s a statement of fact, akin to “Albert Pujols is leading the NL All Star voting for 1B”.

Citing a player’s RBI’s does not (necessarily) imply that a commentator thinks player X is (good, bad, great, mediocre...).  Sometimes they are just trying to tell the viewer/reader than Player X has driven in 100 runs this year; there are some fans who don’t obsessively follow baseball with their every waking moment… no need for a “throw-down"…

I would like to know if Young has led the league in RBIs over say the last 100 games.  Not because it tells me something about Young necessarily but that it tells me about his circumstances.

The story should be the circumstances, not the player.

Pat Tabler was something insanely awesome with bases loaded.  This tells us nothing at all about Tabler.  But, it’s a great story, if you treat Tabler as simply being circumstantial.

I don’t mind seeing current-season numbers, although at this point in the season they’re pretty much meaningless.  But at the season goes on, they start to mean more and more.  But I think you’re right about making too much of one month or half a season or whatever.  However, I don’t want to miss the legitimate stories as they happen, so I just have to take everything with a grain of salt until everything shakes out.

I had a look at the predictive power of the first 3 games of the season at http://blog.daniel-watkins.co.uk/red-sox-arent-doomed.  Needless to say, it’s minimal, but it was a fun little exercise.  Hopefully I’ll be flexing my saber/statistical muscles a little more this season than I have in the past (which was not at all).

The regression amount would be 69/(69+3), or 96%.

So, a team that say starts at 0-3 would be considered a true .480 team (absent any other information).  And I presume that teams that start off at 0-3 have a rest-of-season record of .480.

I can’t see your blog.  Is that close to what you got?

Also, because it’s so extreme (win% = .000), the regression process may not be the best way to do this.

Last year Ubaldo was the ultimate “on pace” guy ... And he ended up being a great example of “regression to the mean” ... As most do.

This year the whole AL East is an “on pace” story.

As an Orioles fan, my heart is wanting me to believe that with Buck and an improved lineup, we’ll have a chance to make the postseason this year.  Unfortunately, my head keeps telling me that it’s unlikely.  I hope my heart wins, but I’m not betting on it.

I don’t even want to read the article he linked:

http://bayesball.blogspot.com/2011/04/on-pace-for-162-win-season.html

Daniel/50: nicely done.

Tango/54: the article is Jon Heyman’s Daily Scoop, titled “Orioles not shocked by fast start, even if most of baseball is”.  The obvious conclusion is that readers of this blog are not “most of baseball”.

"But his actual pace is to do whatever it is other people who started with 3 HR in 3 games actually did.”

This leads to equally stupid claims.

“A team that starts 0-4 is on pace to lose 162 games.”
is as stupid as
“No team that’s started 0-4 has won the world series.”

And if you want to grumble about sample size, most journalists do actually start with “It’s still early, but...” in either case.

The 0-4 thing is, I won’t say stupid because that’s too harsh, let’s say “fatuous”, but not just because of sample size; it’s because (and I’m only guessing here, but confident I’m right) several World Series winners have at one time during their winning season lost four straight games.

The 0-4 distinction elevates the start of the season to a level of significance that is unwarranted and misleading.

So, while the “no team has ever...” statement might be true, it is also fatuous.  Let me also add that just because someone says something fatuous from time to time, that doesn’t make them stupid.