# Using Park Factors

BTF Park Effects - Explanation of Methods

1997 Ball Park Factors

### Why do we need Ball Park Factors anyway?

Park Factors can be used to help answer many "what if" questions. What if all players played in the same park, who would be the best? How much does each park effect play? What if Fred McGriff played in Coors Field and Andres Galaragga played in Turner Field, who would help the Rockies more?

Park Factors help filter out the bias present in player's statistics due to the influence of ball parks.

Do park factors strain out all park related error? No, they don't. Unfortunately, there are too many factors muddled together to confidently filter out only park effects.

Park Factors are a best guess. They are a tool, just like other sabermetric methods are tools. We use these tools to enlighten us about something we don't already know or understand.

What I try to do is with park factors is take what a player did playing in one park, and translate his results to a different (or neutral) park. The translated numbers give me a better value for comparing players.

When using park factor to move a player from one park to another, remember the numbers are not predictions. Players possess unique skills; parks have unique characteristics. Some players' skills are better suited to a particular park. A player who plays in a park which complements his skills has a higher value in that park. It's difficult to discern, sometimes, whether a player's true value is enhanced by the park more than the standard park factors would suggest.

I try to filter out a park's effect on a player's statistics, but knowing that some players take better advantage of their playing conditions than others, I understand that the value of the resulting numbers may be misleading in some cases. Overall, I believe the work has tremendous value to baseball decision makers if the limitations are kept in mind.

### Galarraga and McGriff - What if they played on each other's team in 1997?

To illustrate how Ball Park Factors can be used, I played around with Fred McGriff's and Andres Galarraga's statistics. Here are their real life numbers last year:

```Name	           Avg  OBP  Slg  OPS   G  PA  AB  R   H  2B 3B HR  TB RBI SH SF HP BB   K SB CS GIDP
Andres Galarraga  .318 .389 .585 .974 154 674 600 120 191 31  3 41 351 140  0  3 17 54 141 15  8   16
Fred McGriff      .277 .356 .441 .797 152 641 564  77 156 25  1 22 249  97  0  5  4 68 112  5  0   22
```

Here are Galarraga and McGriff's park neutral statistics:

```Name              AVG  OBP  SLG  OPS AB   H  2B 3B HR BB  SO
Andres Galarraga .293 .369 .535 .904 597 175 29  3 37 57 152
Fred McGriff     .279 .354 .454 .808 567 158 26  1 24 65 113
```

OPS stands for On Base plus Slugging. This is a pretty good measure of overall offensive performance. Using OPS, Galarraga loses 7.2% (.070 total) of his value, while McGriff gains 1.4% (.011 total)

Just out of curiosity, I checked Galarraga's and McGriff's road stats for 1997. If I'm on the right track, they should compare to the park neutral stats. Here they are projected over a full season:

```Name              AVG  OBP  SLG  OPS AB   H  2B 3B HR BB  SO
Andres Galarraga .295 .372 .560 .932 604 178 28  6 40 50 156
Fred McGriff     .289 .375 .485 .860 540 156 22  0 28 70 104
```

Galarraga and McGriff's stats are pretty close. If both played more games in hitter's parks than average these numbers would be pretty good.

Currently, my park effect system is generic to all batters. In reality, most parks effect right-handed hitters differently than left-handed hitters. As a matter of fact, STATS ' hitting indexes indicate this. For Turner Field, the LH batters' home run index is 79, while the RHB index is 88. Hopefully at some point in the future, I'll be able to apply my methods using split data.

Let's look at how Galarraga and McGriff's statistics look like if they'd played in each other's home park in 1997?

```Name              AVG  OBP  SLG  OPS AB   H  2B 3B HR BB  SO
Andres Galarraga .289 .369 .516 .885 595 172 28  3 34 59 152```

Or McGriff playing in Coors Field for half his games?

```Name              AVG  OBP  SLG  OPS AB   H  2B 3B HR BB  SO
Fred McGriff     .302 .371 .493 .864 570 172 28  1 26 62 104
```

Checking the STATS home run index for lefty/righty batters, makes me believe that these figures might be understated. In Coors in 1997, the HR index for righties was 140, for lefties 94. Expect Galarraga to lose even more value, with his home run output declining much more than figured here. At least Andres won't have to face the Braves' starting rotation. That should help him out.

As for McGriff, his move to Tampa Bay clouds the crystal ball. I don't really know how that park will effect offense. If I had to make a prediction, I'd guess that his numbers will stay about the same. Atlanta was one of the tougher parks to hit in during the 1997 season. He's had a noticable drop in bat speed, but new venue sometimes do wonders. Having something to prove doesn't hurt either.

Keep in mind, park factors generated using split stats would definitely be more accurate.

Using the Park Factor data isn't extremely difficult. It is, however, time consuming. Thankfully for you, I'll be supplying the park neutral stats here on my web site. I'll also supply an Excel spreadsheet to make it easier to do "what if scenarios."

Here, I'll outline the steps to create park neutral statistics using Andres Galarraga's 1997 season in Colorado as an example.

1. You have to begin somewhere and a getting the player's PA are the starting point. PA are not modified and will be used in future steps. Galarraga had 674 PA in 1997.
2. Calculate Andres Balls in Play (BIP), subtracting his actual BB, SO, HP, SH and SF from his PA:
674 PA - 54 BB - 141 SO- 3 SF - 0 SH - 15 HP = 461 BIP
3. Calculate The Big Cat's adjusted walks and adjusted strikeouts by multiplying his Rockies totals with the corresponding park factor:
BB = 54 Factor = .950   54/.950 = 57 aBB
SO = 141 Factor = .925   141/.925 = 152 aSO
4. Calculate Andres' adjusted BIP, subtracting aBB, aSO, SF, SH, HP (SF, SH, and HP are not park corrected) from his PA:
674 PA - 51 aBB - 130 aSO - 3 SF - 0 SH - 15 HP = 475 aBIP
5. Calculate Andres' BIP adjustment factor, dividing his adjusted BIP by his original BIP:
475 aBIP/ 461 BIP = 1.030 BIPaf
6. Calculate adjusted 1B, 2B, 3B and HR, multiplying by the corresponding park factor and the BIP adjustment factor. (I'll just calculate 2B here. Just substitute the corresponding values for 1B, 3B and HR.)
31 2B * 1.030 BIPaf / 1.049 2B PF = 30 a2B
NOTE: You'll notice that Galarraga is listed with 29 2B above. The difference is due to rounding and decimal places.
7. Re-build the Cat's complete batting lines.

Note: To generate Galarraga's Turner Field statistics, substitute his park neutral stats for the the original stats. Follow the seven steps making one modification: In step 6, instead of dividing by the park factor, multiply by the new park's factor.

I realize that some of this is pretty numbing. The entire method is explained here for the readers who are interested in such things. (Believe me the number is not large.)

If you'd like park neutral player statistics or statistics modified for play in a particular ballpark, e-mail me.

Back to the top of page | BTF Homepage