THE BOOK--Playing The Percentages In Baseball

Just an aside for a minute; is this blog popping up in a news aggregator time stamped 5 hours in the future for anyone else? This is a 4:30PM story that is time stamped 9:30PM, thus it says on top of my feed for the next 5 hours.

Google Reader is showing as 4:57pm.

Zulu time?

Looking to be educated here:

Does +1 WAR in a 16 team league equal +1 WAR in a 30 team league?  Does the size of the league (and thus the number of teams you have to beat, and the expected win total of the top team in the league) have any impact on teh value of wins?  Is that accounted for somehow in WAR?

The ‘Atom’ option is correct, I was using the ‘RSS 2.0’ feed and was having the issue. Thanks.

Baseball definitely views players of the past differently from other sports, but I think there’s probably some justification for that. Baseball is a significantly older sport than the other three major North American sports, and has been a mature professional sport for much longer.

Roughly speaking, I’d peg the development of each sport as follows, based on stable rules and league structure, ability to attract top athletes, similarity in style to today’s game, and the general opinion of the players of the era.

Baseball: Developing from 1850-1920, modern from 1920 on.
Hockey: Developing from 1880-1950, modern from 1950 on.
Football: Developing from 1890-1960, modern from 1960 on.
Basketball: Developing from 1890-1960, modern from 1960 on.

Then it’s reasonable that baseball goes farther back in history when listing the greatest players, as it has a “modern era” going back to 1920, unlike the other sports.

Regarding your point on Gretzky, it sounds as if you consider him to be easily the greatest hockey player ever. I can see that on a career level, but do you consider his peak/prime to be better than Bobby Orr’s?

I should add that I agree somewhat with your overall point. It’s not uncommon to see someone put together an all-time baseball team with a majority of players that starred before WWII, and it doesn’t seem plausible that the greatest players were all from that time.

Baseball “all-time” debates on the internet are just VORP counting contests.  This has long been a pet peeve of mine.

Saying Lefty Grove was better than Randy Johnson because he had 5% more WAR is absurd.  Independent of modern technological advances, Lefty Grove’s major leagues weren’t close to as strong as Randy Johnson’s AAA.

If you want to determine who’s the greatest, you can’t just measure everyone’s accomplishment against their peers.  You need to make some adjustment for the quality of their peers.

I suppose one could argue that the best player in NBA history is Bill Russell, but that is a borderline insane argument, when one watches old tape. 
Today’s big men, are bigger (somewhat), significantly stronger, and generally faster.  Part of this is things like strength training being much more important than it used to be - which isn’t his fault but it does make todays players better.  Couple this with that the players are generally faster and just overall better, I don’t think Bill Russell would be an all star in today’s NBA (Then again, considering how poor the centers are in the east.....)

JB H, to the extent that I talk about “greatest” ever, I always start from the premise that a major league win in 1876 is worth the same as a major league win in 2009.  I do this fully recognizing that few people share this perspective and that it only addresses a very limited question.  So I have no problem with the conclusion that Grove is the top left-hander and it is in fact my conclusion as well.

However, I agree that the glorification of the past in baseball is tiresome.  When I rank Grove #1, I’m not making any claim beyond “he contributed more wins for his team in the environment in which he played than did anyone else”.  I don’t disagree with the conclusion that Johnson was a superior athlete or however you’d like to put it.

Just to defend Grove a bit, part of his argument is that he really should have had 110 or 120 WAR, as he had five great seasons in Baltimore before being sold to the A’s.  If you take the stance of adjusting for the conditions of the time (beyond the pure baseball factors like scoring context, park effects, etc.), then I think it is only fair to grant an allowance to Grove for the Baltimore years, as his being there and not in the majors was a unique product of the time in which he pitched and would never happen under today’s conditions (Evan Longoria and others held back for a couple months notwithstanding, nor Ryan Howard).

So in Grove’s case, there’s a little more there than a 6 WAR gap, which is not to say that it is outlandish to still put RJ ahead.

"Then how can you still claim to have Lefty Grove as the best LH ever?”

Because I’m ranking the player relative to his peers?

I am sick of these arguments.  Nobody believes if you transported Grove to 2008 - literally yanked in time with no training, etc - he would do as well.

That is never what the all time great arguments are about.

It’s all context!

@Patriot (and others),

Exactly.  That’s how I view the question as well.  Who was the most valuable in their time period - and potentially some reasonable adjustments based on conditions we can account for.

While everything Tom said is true about the advances in physical performance, it pretty much reduces the question of best player to who’s the best player currently playing (or I suppose recently retired).  Which to me, is a very boring question.

And while I didn’t mention Grove’s time in the minors since my commentary wasn’t really the goal of the piece, it was definitely in my mind.

Craig - To your point on Russell, if Shaq had played in Russell’s era, or worse, George Mikan’s era, dude would’ve probably averaged 50 points per game.

What’s strange to me is that people consider players (aside from guys like Ruth, who clearly are all-time greats) from pre-1947 (and really, even though Jackie Robison played in 1947, real integration didn’t occur till later). Playing in a league that chooses to keep out a number of great players simply can’t be compared to a league that allows everybody.

"if Shaq had played in Russell’s era, or worse, George Mikan’s era, dude would’ve probably averaged 50 points per game.”

He did.  Went by the name Wilt back then.

Greg, my WAR does not make an adjustment for number of teams in a league. 

If MLB suddenly expanded to 75 teams, the quality of opponent would drop, and as I calculate it, Pujols might go from a +10 player to a +15 player or something.  It is fortunate that through the 20th century, baseball has expanded slowly, and the number of teams has expanded in the same direction as the increase in the talent pool.  Which one has increased faster is up for debate, but at least they have both moved in the same direction, which makes using a constant WAR somewhat reasonable.

It’s always good to keep in mind what any measure is measuring.  By my estimation, having Lefty Grove meant just slightly more wins to his teams than having Randy Johnson 70 years later.  That’s all I am attempting to measure.  Which was greater when they played in seasons that far apart, WAR cannot say.

As far as I know, nobody’s traveled in time yet so it’s up to your imagination to answer and can’t be proven.  But why do we assume, for comparison’s sake, that we plunk the Unit into the 1930’s and watch helpless batters who’ve never seen a 6’10 pitcher before?  Or that we shift Lefty into the 1990’s, where the fastball great for his time might only be a league average pitch?

We could also look at it as transporting Randy back in time, but having him grow up in the 1910’s, and Left growing up in the 70’s.  Maybe Randy only grows to 6’6 with the nutrition of the day, and Lefty through modern training winds up throwing 95-100.

Take all the players from the 1920’s and 30’s, have them grow up in modern times, and I think they would be just as good as modern players, if the number of teams adjusted to the smaller population of players.  And Lefty would rise to the top of that group, just as he did in his own time.  But I think his WAR would be lower than Johnson’s in 1989-2009, because Josh Gibson would have taken him deep a few times, and ol’ Satch would have lowered the league ERA he’s compared to.

Question for Alan Nathan or anyone up on the physics:

If we know a player threw a baseball X number of feet, can we make these assumptions:

1. He threw it at an optimal angel for distance
2. He threw it with optimal backspin
3. Assume wind was 25 MPH, or whatever you want to put in

And calculate who hard the ball had to be thrown from his hand?

This site has some interesting information on pitcher speed:
http://www.baseball-almanac.com/articles/fastest-pitcher-in-baseball.shtml

In 1917, Walter Johnson, Christy Matthewson, and Joe Wood had their pitches timed by a machine.

The results, in feet per second:
WJ 134
CM 127
JW 124

Converting to miles per hour I get:
91
87
85

So the fastest pitchers then, with a big uncertainty as to the accuracy of the measure, ranged from an average 2009 fastball to a Barry Zito/Livan Hernadez type.

But the ball loses about 8 MPH from the pitcher’s hand to home plate, and it sounds like the machine is measuring them towards the end, not out of the hand.  Of course, in a test they probably weren’t throwing from a full 60 feet.  It seems reasonable that Johnson was throwing 95 and the other two right around 90.

Rally - The Wilt comparison is kind of my point. Does anybody think Wilt would average 50 in today’s game?

It is fortunate that through the 20th century, baseball has expanded slowly, and the number of teams has expanded in the same direction as the increase in the talent pool.  Which one has increased faster is up for debate

It’s not even close.  If you were the 50th best 18 year old baseball player in 1920, following your dream meant working for nothing and hoping to defy long odds to eventually make a middle class salary for a handful of years.

Today baseball gives six figure bonuses or college scholarships to hundreds of amateurs every year.

That’s not to mention that every kid today is constantly exposed to baseball and has access to organized leagues, whereas Lefty Grove grew up before baseball was on the radio or little league existed.

There are many more people competing for each major league job (and run/win) today than ever before.

Of course not, but that may not have anything to do with level of competition, but game pace.

In his 50 point year, Wilt’s team scored 125 points (and allowed 122).  Wilt played every minute, and scored about 40% of his team’s points.

For the 1999-2000 Lakers, Shaq played 3163 minutes and scored 29.7 points per game.  Prorating the minutes, Shaq scored about 35.5% of the team points when he was on the court.

Not that much difference context-wise.  I have no doubt Wilt would have scored 25-30 per game in today’s NBA.  He was a very skilled offensive player outside of shooting free throws, and would have a size/strength advantage on every current NBA center except for Shaq.

I don’t think organized leagues is any advantage.  I’m sure kids back at the turn of the century played a lot more baseball than today’s little leaguers play.

If we know a player threw a baseball X number of feet, can we make these assumptions:

1. He threw it at an optimal angel for distance
2. He threw it with optimal backspin
3. Assume wind was 25 MPH, or whatever you want to put in

And calculate who hard the ball had to be thrown from his hand?

Yes, if you add
4. Know the air density (altitude, temperature)

It’s also worth noting that assumption #2 may or may not be reasonable.  The optimal backspin is on the order of 3000 rpm, which is near the upper limit of a pitcher’s ability to apply backspin to a pitch, according to our PITCHf/x estimates (I think previous estimates in literature are for lower spin rates).  I’m not sure how an fielder’s ability to apply backspin to a throw compares to a pitcher’s ability to apply backspin to a pitch, but I wouldn’t think the fielder could do much better than the pitcher.

This is the average weight of pitchers, by decade (weighted by IP):

decade avgWeight
1870 140
1880 147
1890 159
1900 167

1910 180
1920 179
1930 184
1940 186
1950 188

1960 194
1970 195
1980 198
1990 200
2000 199

Through the 1900s, pitchers were very very small.

A jump occurred from 1908 (weight of 172) and 1910 (weight of 179).

From the 1910s through the 1950s, they were on the small-side by today’s standards.

Starting in the 1960s, pitchers started getting bigger.

***

Yes, when you do the timeline adjustment, you have to decide whether you pluck them from their peak to today (say Jesse Owens, 1936 to 2008). 

Or, whether you pluck them at birth, and give them the benefit of growth in today’s conditions.

Or, you simply compare to players of their time period, which really means you are saying it is impossible to compare players of then to now.  That is, if Jesse Owens runs 0.50 seconds faster than his competition, and Bolt runs 0.30 seconds faster than his competition, you will “rank” Owens higher.  But, this is really meaningless, since the competition level itself is incomparable.

This would be no different than comparing Sadaharu Oh to his competition, and then comparing him to Hank Aaron or Babe Ruth.  And if Oh stood above his competition more, then he’s the best hitter.

As you can see, it’s rather nonsensical to do this, because then why not compare college hitters to their peers, and then including those rankings with the pros?

Your going to object and say “it’s not the highest level of competition”, but if your league doesn’t include Josh Gibson or Paige or Oh, then your league doesn’t have the highest level of competition either.  And if you say the “highest possible at the time”, then for Josh Gibson, he gets to count against his black peers, since that’s all it was possible.

All to say that you need to define exactly what it is that you are comparing before drawing your list.

Mike,

I would not assume optimal backspin if I picked a player at random, saw how har he threw, and tried to estimate initial velocity from it.

But for guys like Glen Gorbous, who I believe holds the record throwing a ball 446 feet, or the other players who held the record at one time or came close, I think it’s reasonable to assume they had everything working right that could be controlled.

I’m making posts here, and am including the link to make it easier for me to navigate:

http://www.baseballthinkfactory.org/files/newsstand/discussion/the_book_blog_tango_the_greatest_lhp_of_all_time_pitched_80_years_ago/

This is what I posted about a month ago. I think it shows how good Grove was. And I posted something else about how two separate and distinct 5 year periods of his are both among the best ever. I will have to repost that later

Using the Lee Sinins encyclopedia, here are the top 20 lefties in runs saved above average per 9 IP (adjusted for park effects). Minimum 2000 IP. Through 2008. Of course, it is not fielding independent

Lefty Grove 1.53
Randy Johnson 1.19
Hal Newhouser 0.93
Whitey Ford 0.91
Carl Hubbell 0.89
Lefty Gomez 0.87
Sandy Koufax 0.85
Noodles Hahn 0.84
Rube Waddell 0.77
Jimmy Key 0.73
Toad Ramsey 0.70
Andy Pettitte 0.66
Ron Guidry 0.62
Tom Glavine 0.62
Thornton Lee 0.62
Billy Pierce 0.61
Chuck Finley 0.57
Warren Spahn 0.55
Howie Pollet 0.51
Eddie Plank 0.51

Here are the top 25 lefties, 2000+ IP in strikeout to walk ratio relative to the league average. The 211 for Grove means that his ratio was 2.11 times the league average. His was 1.91 while the average was .9

STRIKEOUTS/WALKS RATE PLAYER LEAGUE
1 Lefty Grove 211 1.91 0.90
2 Carl Hubbell 206 2.31 1.12
3 Noodles Hahn 197 2.41 1.22
4 Rube Waddell 189 2.88 1.52
5 Ron Guidry 184 2.81 1.52
6 Toad Ramsey 180 2.26 1.25
7 Randy Johnson 179 3.27 1.82
8 Harvey Haddix 175 2.62 1.49
9 David Wells 172 3.06 1.78
10 John Candelaria 171 2.83 1.65
11 Greg Swindell 171 3.08 1.80
12 Ken Raffensberger 164 1.80 1.09
13 Dutch Leonard 162 1.75 1.08
14 Sandy Koufax 159 2.93 1.84
15 Jon Matlack 156 2.38 1.52
16 Hooks Wiltse 155 1.93 1.25
17 Mickey Lolich 153 2.58 1.68
18 Ed Morris 153 2.44 1.60
19 Rube Marquard 152 1.86 1.22
20 Jesse Tannehill 151 1.97 1.30
21 Hal Newhouser 148 1.44 0.97
22 Lefty Gomez 147 1.34 0.91
23 Herb Pennock 143 1.34 0.93
24 Eddie Plank 143 2.06 1.44
25 Fritz Peterson 142 2.38 1.68

Now for HRs allowed. The 200 for Rixey comes from 92 being half of 184 (he allowed 92 while the average guy would have given up 184). so 92/184 = .5 and 1/.5 = 2. Then 2*100 = 200. But Rixey pitched in a very tough HR park. Notice how no one else from the debate is here. So is it Grove hands down?

HOMERUNS RATE PLAYER LEAGUE
1 Eppa Rixey 200 92 184
2 Ed Morris 170 42 71
3 Eddie Plank 155 41 64
4 Danny Jackson 152 133 202
5 Dutch Ruether 151 54 81
6 Andy Pettitte 146 230 335
7 Lefty Grove 143 162 232
8 Tommy John 143 302 431
9 Al Leiter 141 198 280
10 Hippo Vaughn 141 39 55
11 Rube Benton 139 52 72
12 Sam McDowell 138 164 226
13 Mike Hampton 135 187 252
14 Noodles Hahn 135 28 38
15 Whitey Ford 133 228 303
16 Tom Glavine 132 356 469
17 Frank Killen 132 55 72
18 Hal Newhouser 131 137 180
19 Rube Waddell 131 37 49
20 Fritz Ostermueller 130 105 137
21 Doc White 128 33 42
22 Johnny Vander Meer 128 100 128
23 Charlie Leibrandt 127 172 219
24 Curt Simmons 126 255 321
25 Earl Hamilton 124 43 53

I know there are sports statisticians who measure things in terms of z-scores (ie, number of standard deviations above the mean).  You can then talk about a “1 in 5000 player” or a “1 in 1000 player” etc., a figure which is independent of the era of the player’s career.  (Here, the 5000 or 1000 would refer to the number of Major Leaguers, not people from the general population.)

Something like this should be applicable to comparing baseballers in different eras - the standard deviations were (I imagine) greater in 1930 than they were in 2000, so the 1930 player needs to be further away from the mean to be equal to the 2000 player.

This approach should also (I think) get around the problem of judging the expansion of the league against the rise in talent level.  But I’ll leave it for you guys, since you all know baseball much better than I do.

To define dominant pitcher= Using Adjusted ERA+ from baseball reference, we have the following pitchers who led their league 3 times in 5 years -
Johnson,Alexander, Mathewson pre 1920

Grove - 2 times

Seaver - 1970s

Maddux,Pedro,Clemens, Johnson

9 pitchers who make a great argument for top 9 of all time.

Now, Grove was the best pitcher in a 50 year span - 1920-1969

Johnson was 1 of the 4 best in his era. He may be the best of the 4, but I think you will find backers and stats for all 4

So, it is real hard for me to understand how the Big Unit dominated his time as much as Grove did his.

There were 3 other players in Johnson’s time who were about as dominant as he was. Were each as dominant as a once in a 50 year player like Grove?

Two groups of kids take a test - both groups have an average of 80 - in group A, 1 kid scores a 120, the next highest is a 90. In group B 4 kids score between 115 and 120. Are the 4 kids in group B each as dominant in their class as the one kid in group A?

Grove stood out more in his era than any of Randy/Roger/Greg/Pedro did.  But I don’t know how meaningful that is.  Hit 3 of those 4 with a bus early in their careers and the one left standing dominates his era just as much as Grove did.

Also, we must consider that Grove may not have been the best pitcher of his era.  Probably was Satchel.

I don’t think that we know enough about his performance to say that Satchel was “probably” better than Grove.  That’s a pretty high bar.  I have no problem with “maybe” or “possibly”.

I know relatively little about the Negro Leagues, but Paige has always struck me as more of a Nolan Ryan/Cy Young hybrid...a great pitcher, but one who draws a lot of extra attention for his style (like Ryan) and who lasts forever and ever (both Young and Ryan) and thus has much more of a legend grow around him. 

When the HOM rated the pitchers from the 1930s-50s, Grove was a unanimous #1 and Paige finished third, and they take the NeLs very seriously.

I just finished the new Satchel Paige book, so he’s fresh on my mind.  I’m kind of extrapolating a bit from how good he was into his 40’s and beyond.  From his minor league record in his 50’s with the Miami Marlins it appears he still would have been a serviceable pitcher then.

It looks like the reason he didn’t pitch in the majors in 1950, or after his St Louis Browns stint until his 3 innings at age 59, was because there was only one owner who would sign him.  Bill Veeck.

It is really hard to say how good he was at his peak, but he quite possibly was the greatest pitcher to ever live.

Dave,

Trying to answer your question from #27, I did something a few years ago where I found the best seasons and careers in ERA by standard deviations below the mean. Here is what I posted on the SABR list in 2005:

Sometimes pitcher’s ERAs are compared to the league average (relative ERA). Then they are ranked by this. One criticism of this is that in very low or high scoring years it might be easier to get a good relative ERA. An alternative to relative ERA would be to calculate how many standard deviations below the mean ERA a pitcher is.

I thought compiling this for all of baseball history would be very time consuming. But I just happened to check Microsoft Excel’s “subtotals” options and standard deviation is there. So, using the Lee Sinins Sabermetric Encyclopedia, I quickly compiled all of the ERA title qualifying seasons in AL history. Then I got the mean and standard deviation for the qualifiers in each season, then calculate how many standard deviations below or above the mean ERA each pitcher was. Then I ranked them from lowest (most below) to highest. I have put the top 100 AL seasons at the following site

calculate how many standard deviations below the mean ERA a pitcher

http://cyrilmorong.com/ALERA.htm

At this site, the last column “REL” tells you how many standard deviations below or above the mean a pitcher was. Pedro Martinez had the best at -3.79 in 2000 (and 2nd best -3.18 in 1999). Dutch Leonard is 3rd with -2.94 in 1914. It still looks a little like the best seasons come in very high or low scoring years, but maybe not as much as with relative ERA. Santana of 2004 was 9th.

Also, I compiled the career average in terms of ERA above or below average for all AL pitchers with at least 2000 IP. I did weighted average of their “REL,” weighted by IP. Lefty Grove came out on top (Martinez does not have 2000 IP in the AL). But this only included data for each pitcher from seasons in which they qualified for the ERA title.

Just in case anyone does not know this, a weighted average in this case, in an Excel spreadsheet, can be calculated very easlily. For each season, you mutliply REL times IP. Call that MULT. Then this gets summed for each pitcher (using the Excel subtotal function). This gets divided by a pitcher’s total IP ( which also can get summed using Excel subtotal function).

Suppose a pitcher had a REL of -1 in a year with 300 IP and a -2 in a year with 200 IP. You tell Excel to add -300 and -400 to get -700. Divided by 500(you also tell Excel to sum IP, too), you get -1.4. When you have many pitchers who all pitched a different number of seasons, this gets you the weighted average very quickly.

As for using the Lee Sinins Sabermetric Encyclopedia, I had it call up the top 1000 seasons in ERA for each 20 year period in AL history, along with IP for each pitcher. The 1981-2004 period came in with 997 pitchers. The rest were well under 1000. It should not be hard to do all of this for strikeout to walk ratio, HRs allowed per IP (or per batter), or RSAA, Runs Saved Against League Average, which is park adjusted (ERA here is not).

Cyril Morong

http://www.baseballthinkfactory.org/files/dialed_in/discussion/steve_austin_is_not_a_baseball_player/

Tango, I assume you got your weights from the Lahman db, and so did I.  MLB got a little smaller recently, and so I don’t agree with the overall position that players are bigger (they aren’t).  If MLB were still all white it *would* be bigger than 1950, but since we have many Latinos, it isn’t.  the Latinos, who represent many of the best players in teh game are simply smaller than the players from the 1950s.  Size simply is not the tell-tale factor in baseball.

Let me clarify: MLB is overall bigger than 1950, but the latino players are not.  The size gain is US players.