View Full Version : Food for Thought: 2003 Montreal Park Factor, Vazquez and ERA+

12-25-04, 04:35 AM
While typing my extremely long post on the thread (which if interested one should read to see the streakiness of Vazquez' career) <a href="http://forums.nyyfans.com/showthread.php?postid=1837109#post1837109"> Vazquez wants to stay a Yankee </a> I came to note at the end of the very long thread this part:

-His 154 ERA+ in 2003 was very excellent (for any pitcher indeed). But someting jumped out to me - the league ERA adjusted to park factor for Javy and other MTL pitchers was 4.97! In Monreal no less!

From 2000-2004, league ERA for Monreal pitchers was 4.64 4.62 4.16 4.97 4.14.

Now while looking at those numbers, I have to wonder what factors came into play.

Using baseball-reference's year by year park factors, from 1995-2004, Montreal had these numbers (hitters/pitchers factor): 102/101, 103/102, 100/100, 96/97, 104/103, 100/100, 107/106, 101/101, 118/116, 95/96

Now looking at those numbers, one sees that Montreal has pretty consistently been put near neutral in park factor with years tending to favor batters only slightly.

But it immediately jumped out that in 2003, Monreal had a batting factor of 118 and pitch factor of 116! By contrast, in 2003, Colorado was 112/111 (2004 they were 120/117, 2002 they were 121/119).

Now obviously, through Montreal's history, they haven't been close to being considered anywhere near Colorado and the pitcher's deathtrap of Coors Field. They've had a neutral ballpark. So why did 2003 jump out?

As I thought about it, I remembered they played many games in San Juan, Puerto Rico, at Hiram Bithorn Stadium. 22 of 81 "home" games were played there in 2003 (as well as 2004).

But would a park suddenly have such a factor?

Digging up some links, I found it troubling to see that in 2003, a record in homeruns and runs were scored in games in San Juan. Take, for instance <a href="http://www.cfl.ca/Slam030607/mlb_mtl-ap.html"> Hiram Bithorn Stadium playing like Home Run Bithorn</a>:

And Garret Anderson hit five homers in a three-game series, including three Wednesday night. But even Anderson admits this park isn't fair.

"It's small. I've played in a lot of hitters' parks but you still have to hit the ball halfway decent," he said. "Here, if you hit the ball down the line you don't even have to hit it well."

The small stadium with the fast artificial turf, eight-foot high fences and the wind that usually blows out to right field has easily been the best hitters' park in the majors.

There have been 57 homers and an average of 12.2 runs in the first 14 games at the ballpark, where it's 399 feet to centre, 315 to left and 313 to right.

That's 54 per cent more runs per game than in Montreal's other 47 contests and 29 per cent more than the average for the rest of baseball.

But then in 2004, Montreal began their year off in San Juan with 22 games as well. But why did Montreal get rated a pitcher's park this year?

In 2004, in 21 games in Puerto Rico, Montreal and opponents scored 134 runs at 6.4 runs per game. In 24 games at Olympic Stadium, they combined for 203 runs at 8.5 runs per game. 24 home runs in San Juan comapred to 60 in Montreal.

At first, this was considered random variation, but given the huge difference in offensive production, I searched for other possibilites.

<a href="http://www.macon.com/mld/macon/sports/baseball/mlb/atlanta_braves/9086036.htm"> This article </a> gave me an answer.

FIELD CHANGES: Hiram Bithorn Stadium has a new look and Bobby Cox approves. Last year's hard artificial turf has been replaced by Field Turf, which better duplicates real grass, and the outfield fences have been moved back 10 feet to Montreal's Olympic Stadium dimensions. The result has been a drastic reduction in home runs.

"I think it was much easier to hit one out last year and the new carpet has made it slower," Cox said. "I think it's good. Last year it played a little small."

The increase to 404 feet to center and 375 feet to the power alleys isn't the primary reason balls are no longer flying out of the stadium. The trade winds are blowing in, knocking balls down and preventing the Expos from even reaching the warning track with most of their drives

Suddenly, it became obvious - 2003's park factor seemed inflated because of the park factor of playing 22 games in San Juan.

Checking out game logs, I found out these factors of their games in San Juan in 2003:
In 22 games in San Juan, 114 runs were scored by Monreal, 124 allowed (including 33 allowed in 3 games against ANA). That is a combined 10.82 runs / game.

Now why is this important? ERA+ is often what we use to calculate a player's value compared to the average player. ERA+ is taken by the league average ERA adjusted to park factors divided by player ERA.

League average ERA is calculated by what the average player in the league would have as an ERA if he played in that park. But considering that Montreal had played 22 games in San Juan in a park highly conducive to the long ball and high scoring games, it may have inflated the league average ERA for Montreal.

Higher league average ERA for Montreal would mean a much higher overall ERA+.

The team ERA of 4.01 calculated against their league average ERA of 4.97 would bring about an ERA+ of about 124. The team's pitching as a whole was 24% better than average.

While the team's ERA was #6 out of 16 in the NL, runs allwoed per game put them at #7. They allowed 4.42 runs per game compare to league average of 4.61. League average ERA was 4.28 compared to 4.01 of the team. The park factor for Montreal clearly played a huge factor as Montreal had the lg_ERA of 4.97 while the league's actual average ERA was 4.28.

Let us now look at Montreal's league average ERA since 1998:
4.11, 4.70, 4.64, 4.62, 4.16, 4.97, 4.14

League pitching average ERA since 1999:
4.24, 4.56, 4.64, 4.35, 4.10, 4.28, 4.30

As we can see, Montreal's park adjusted average ERA is almost always near league average with 2003 again as the big anomaly

Now again why is 2003 so important? Because 2003 is the year that Javy had his best year - with a 3.24 ERA he put up a 154 ERA+ which is very good.

His teammate Livan Hernandez with a 3.20 ERA put up a 155 ERA+.
Tomo Ohka had a 4.16 ERA and a 119 ERA+.
Zach Day had a 4.18 ERA with a 119 ERA+.

In 2004 though, Livan Hernandez put up a 3.60 ERA with a 115 ERA+. 40 points of ERA dropping a ratio stat (ERA+) by 40 points is huge.

Ohka 3.40 ERA with ERA+ of 122. This time, a large change of ERA (0.76) resulted in just a 3 ERA+ change.
Day 3.93 ERA with ERA+ of 105.

So the question now is, was the league average ERA for Montreal inflated due to their 22 game stint in San Juan which in turn inflated ERA+ numbers? While they certainly had a great pitching staff that year, the fact that their team era+ would be so high when they were near the center in terms of league pitching is baffling.

And when we look at Vazquez' numbers in respects to ERA compared to league average ERA and ERA+, we see how a few numbers or park factor may suddenly change a year's performance.

His 3.24 ERA in 2003 compared to the 2004 league average ERA would bring his ERA+ to 128 from 153. Certainly still a very good number, but he is considered then half the pitcher he was. Take 2002, his ERA+ would be 128 to 129 again. Certainly great but nowhere near the 154 performance.

Then take his 2001 league average ERA (when the team park factor was 107/106) and we see that his ERA+ is 142. Certainly much better than 128 but compared to 154, that is 12points or nearly 20% less effective.

If we toy around with 3 year splits for Vazquez (2001-2003), and use that league average ERA, then his ERA+ from 2003 comapred to split years is 140. But if you use 2002-2004 (montreal), his ERA+ would translate to 138.

Still very good year. But compared to 154? Certainly a sudden drop off.

So my current conclusion?

First, watch out when just using ERA+ because it calculates league average ERA for that team against player performance. Unlike OPS+ which uses a much mroe complicated formula, this is simply a ratio of the average player with park factors compared to player ERA. Hence, relievers should never be compared with ERA+ much less with starters ERA+ given the difference in actual ERA.

FWIW, Javy had a DER of 4.61 which means in ERA+ terms he had a 98 instead of 92 for 2004. Making him just barely under average.

Next, when looking at his 2003, one should watch out for park factors. 2003's park factor seems to be the big abnormality compared to the rest of the years surrounding Montreal. Playing 22 games in San Juan which was an extreme hitters park may have inflated the park factor rating for the team's calculation of ERA and hence ERA+ is higher than it should probably be if pitching in a neutral park (as Montreal has traditionally been).

So what do I think of Vazquez after all this? He's certainly a very good pitcher in the league but has proved very streaky (bad year then good year then bad year then good year etc.) but while 2003 was very good, the use of ERA+ might've been inflated by the experiment in San Juan. Furthermore, if you consider that to be the year where the law of averages balanced him out with his bad 2002 year (when you put stats together, they get eerily close to his career stats), it has changed my perspective on him even more.

He's a very good pitcher but we have overrated him greatly through the reliance on his good years (two of them). Good news is, next year he's likely to rebound and through law of averages, he'll have a good year. Problem I think is going from the NL to AL nontheless the weak hitting NL East to the very hard hitting AL East is going to mean it will never be near a best year again.

I think if we average his numbers out over the rest of his contract with us (provided he stays with us) we can expect an average ERA of around 4 to 4.2 with good years going down to 3.7 ERA and bad years up to 4.6 to 4.7 (no 4.91 again thankfully).

This all of course depends on whether he is 100% healthy, doesn't have mechanical issues, and can pitch at least at the average level he put up in Montreal. No amount of stats can change a person's actual health and numbers.

One concern I do have is that 2003 was his best year - it also happened to be when he was 27 when players usually reach their peak year. While 2004 seems too sharp a decline for his next year at 28, and one expects him to rebound, I'd have to highly caution against predicting him to put up numbers even near his 2003 year (translated to AL stats of course). Just for reference, he would've had to put an ERA of 2.92 last year to match his 154 ERA+ in Montreal in 2003. Possible? Anything is possible. Probably? Highly unlikely.

Also I do have concerns on his decline - he was brought up early and pitched an average of moe than 220 innings a year from 2000-2003. That's not good since many doctors have talked about arm injuries occuring to pitchers who pitch too much before the age of 25 (when one's arm matures). Again, I expect a rebound, but this fact (also abuse by Alou then Robinson and now maybe even Torre) may make it less likely he rebounds to great numbers.

In this light of course, the RJ for Javy swap is up to the front office's idea of risk vs. reward. Do you risk Javy performing good for 3 years for RJ who will almost without a doubt have 3 incredible years at the risk of not having RJ healthy for all 3 years? Those are the odds they have to weigh but it seems almost certain they have chosen the risk of 3 great years of RJ with health risks rather than 3 good (possibly streaky?) years of Javy with few health risks.

01-08-05, 05:07 AM
i don't think they went through this whole thought process before making the trade. it was probably a simple conclusion for them: RJ>vazquez, and that's it.

nice analysis, though.