 |
MVP Predictor Model
Predicting The MVP
Bryan Ellis and Tyson Lowery
SimDynasty.com
Over the years, the MVP formula in Sim Dynasty's baseball simulation has been a sticking point for many users. As with real life baseball awards, many team owners have debated or questioned the simulation's selection of the MVP winner.
Recently, we began to take a fresh look at our MVP formula. The goal of Sim Dynasty's MVP formula is to award a league MVP using the same set of criteria that the writers use when voting for the MLB MVPs. We took a look at how our formula did from 1995 to 2006. Our MVP formula picked the correct MVP winner only 7 out of 24 times. We went back to the drawing board and took a close look at how MVP races have played out 1995. This resulted in a new formula that correctly picked 22 of the past 26 MVP winners.
What factors were found to contribute to a candidate's chances of winning the award? There are a few things that one may not consider at first glance.
Statistics
Obviously, statistics such as home runs, RBIs, and batting average are among the most important factors. You can't win the award without putting up big numbers. But what else goes into it?
Team Performance
Did a team make the playoffs? Did they win 90 or 100 games? Did they win the division by a landslide? Did the team do better than they did the previous season? If a team didn't make the playoffs, did they at least come close? Were they involved in a tight playoff race? All of these are factors. Making the playoffs is important; only 4 MVPs have come from non-playoff teams since 1995, and half of those were involved in tight playoff races (Barry Bonds in 2004 and Ryan Howard in 2006). The other two put up monster numbers when compared to the competition (Larry Walker in 1997 and Alex Rodriguez in 2003).
Defensive Position and Gold Gloves
The voters tend to build in a bonus for players that play traditionally defensive positions, such as middle infielders and catchers. And they tend to build in a penalty for designated hitters. Surprisingly, there seems to be a bit of extra love for right fielders that can't be explained by other factors. In addition, voters logically give a player an additional bonus if he has recently won a Gold Glove.
Repeat Winners
It is difficult to repeat as MVP. One might think that voters would tend to favor a guy who has won the award before, but it seems to be just the opposite. There are 3 cases in our model where we feel a former MVP should have won but did not: Barry Bonds in 1997, Jeff Bagwell in 1999, and Albert Pujols in 2006. The data shows that writers typically tend to spread the wealth and give another guy a shot at the award.
The Colorado Factor
People often talk about hitter's parks and pitcher's parks. The basic premise is that some baseball parks are more beneficial to hitters (or pitchers), and players who play 81 games (half the season) in these parks tend to have skewed statistics. We have not yet seen evidence that this shows up in the voting, with the exception being Colorado and Coors Field. According to ESPN's calculations at http://sports.espn.go.com/mlb/stats/parkfactor, Coors has been a top 3 hitter's park during each of the past 8 seasons.
Looking beyond Coors for a minute, during those 8 seasons, Chase Field and U.S. Cellular Field finished in the top 10 in all but one year. Rangers Ballpark was in the top 10 in all but 2 years. During that time, there have been few legitimate MVP candidates from these teams besides Alex Rodriguez and Luis Gonzalez. In 2002, Alex Rodriguez put up MVP caliber numbers on a bad Texas team but did not win the MVP. We think this had more to do with the Rangers than their park. So do the voters apply a park penalty outside of Coors? It hasn't appeared that way, but if Jermaine Dye doesn't do well this year in the voting, we may have to revise that thinking.
Okay, so what about Coors? Larry Walker is the only Rockie to win the MVP; he did so in 1997. He had to put up monster numbers to overcome the writers' discrimination against Colorado players. There are 5 other cases where we feel a Colorado player would have won the MVP, if not for the Coors penalty: Dante Bichette (1995), Larry Walker (1995 and 1999), Todd Helton (2000), and Matt Holliday (2007). Justified or not, this is how the writers tend to vote.
A Fresh Face
Sometimes a team will add a new piece of the puzzle that will help it make the leap to the next level. The real effect of this can be debated, but the voters tend to like players that were not on that team the previous season. In other words, if a team was good this year but bad last year, the voters tend to attribute that to a new player in town.
What about Pitchers?
Every once in awhile, voters will give the award to a pitcher, usually a relief pitcher. We have not built this into the formula, but have examined two of these cases a bit in the discussion of the AL race below.
The Years We Got It Wrong
So we took all these factors, stirred them into a big pot, and found our model selected the MVP winner 22 of the past 26 times. This means our model failed to get four of them right. Let's take a closer look at each of these.
For the 1995 AL race, our model predicted Albert Belle would win the award. Mo Vaughn was the 1995 MVP, narrowly edging him out 308-300. The oft-cited reason for giving the award to Vaughn that year was that the Indians didn't play under any pressure because they finished 30 games ahead of second place Kansas City. Others say that Belle's surly attitude with the press didn't help, but then again that never seemed to affect Bonds.
For the 2001 AL MVP award, our model predicted Roberto Alomar would win the award. That of course was the year Ichiro won both the MVP and Rookie of the Year. Incidentally, our model had Alomar first, Bret Boone second, and Ichiro third. The voters had Ichiro first and Jason Giambi a close second, followed by Boone and Alomar. We were not surprised that our model got this one wrong; the buzz following Ichiro that summer was unprecedented and difficult to measure.
The 2002 AL MVP was Miguel Tejada; our model predicted Alex Rodriguez by a good margin. A-Rod finished second in the voting, well behind Tejada. So why did Tejada win the MVP that year? You may recall some clutch hits that season by Tejada to propel the A's into the playoffs. The A's had a great stretch that year, winning 20 in a row at one point. According to Tejada's Wikipedia page, "Tejada contributed one-out, game-winning hits in the 18th and 19th games of that run: a three-run homer off Minnesota Twins closer Eddie Guardado for a 7-5 victory and a bases-loaded single against Kansas City Royals reliever Jason Grimsley to break a 6-6 tie." We think these two games made a huge impression on the voters, which is difficult to measure.
Finally, the 2006 AL MVP was Justin Morneau, while our model predicted Derek Jeter. Jeter finished a close second to Morneau, losing 320-306. Our model had Jeter first, Joe Mauer second, Carlos Guillen third, and Morneau fourth. Mauer, Guillen, and Morneau were relatively close in our model. It is interested to note that our model had teammate Mauer higher than Morneau.
Who Will Win in 2008?
Using our formulas, here are the top point totals for players for this year's awards.
National League
Had the Mets made the playoffs and won 90 games, David Wright would be only a few points behind Pujols. We are fairly confident that Pujols will be the 2008 NL MVP, although given this relatively low point total we would not be surprised to see a pitcher get some consideration.
American League
The AL field is historically weak when compared to other years. To give you some perspective, in only 1 case since 1995 has the MVP winner scored fewer than 1600 points in our formula: Justin Morneau, who scored 1532 points in 2006. The teams that made the playoffs this season seem to have a bunch of good players rather than one guy who put up monster numbers. Had the Twins beat the White Sox in the 1 game playoff, Mauer would have leapfrogged the field and been our choice to win the MVP.
Because of the weak field, we could see Francisco Rodriguez getting some serious consideration. Since 1980, there have been 3 relief pitchers to win the MVP: Rollie Fingers in strike shortened 1981, Willie Hernandez in 1984, and Dennis Eckersley in 1992. Eckersley saved 51 games on a team that won 96 games. We took a look at the 1984 and 1992 runner-ups. In 1984, Kent Hrbek and Eddie Murray scored 1138 and 932 points, respectively. In 1992, Joe Carter and Kirby Puckett scored 1218 and 1165 points, respectively. Each of those seasons lacked hitters who were strong MVP candidates. This precedent points to a pitcher MVP in 2008.
But What About Manny Ramirez?
For what it is worth, if you take Manny Ramirez's full season totals and treat him as a Dodger, he ends up with 1604 points. If he had stayed in Boston, his total drops a bit to 1532. We don't believe he has a chance of winning the MVP because he was only with LA for 2 months, but it is worth mentioning that he likely would have been the AL MVP if he had stayed in Boston and put up even his pre-trade numbers.
Sim Dynasty Baseball is an online, simulated baseball game. You act as the GM and Manager of a baseball franchise and direct all aspects of that franchise, such as drafting, trades, lineups, in-game manager preferences, grooming minor leaguers, and much more. The game is free to try, sign up today at www.simdynasty.com.
Bryan Ellis and Tyson Lowery
SimDynasty.com
Over the years, the MVP formula in Sim Dynasty's baseball simulation has been a sticking point for many users. As with real life baseball awards, many team owners have debated or questioned the simulation's selection of the MVP winner.
Recently, we began to take a fresh look at our MVP formula. The goal of Sim Dynasty's MVP formula is to award a league MVP using the same set of criteria that the writers use when voting for the MLB MVPs. We took a look at how our formula did from 1995 to 2006. Our MVP formula picked the correct MVP winner only 7 out of 24 times. We went back to the drawing board and took a close look at how MVP races have played out 1995. This resulted in a new formula that correctly picked 22 of the past 26 MVP winners.
What factors were found to contribute to a candidate's chances of winning the award? There are a few things that one may not consider at first glance.
Statistics
Obviously, statistics such as home runs, RBIs, and batting average are among the most important factors. You can't win the award without putting up big numbers. But what else goes into it?
Team Performance
Did a team make the playoffs? Did they win 90 or 100 games? Did they win the division by a landslide? Did the team do better than they did the previous season? If a team didn't make the playoffs, did they at least come close? Were they involved in a tight playoff race? All of these are factors. Making the playoffs is important; only 4 MVPs have come from non-playoff teams since 1995, and half of those were involved in tight playoff races (Barry Bonds in 2004 and Ryan Howard in 2006). The other two put up monster numbers when compared to the competition (Larry Walker in 1997 and Alex Rodriguez in 2003).
Defensive Position and Gold Gloves
The voters tend to build in a bonus for players that play traditionally defensive positions, such as middle infielders and catchers. And they tend to build in a penalty for designated hitters. Surprisingly, there seems to be a bit of extra love for right fielders that can't be explained by other factors. In addition, voters logically give a player an additional bonus if he has recently won a Gold Glove.
Repeat Winners
It is difficult to repeat as MVP. One might think that voters would tend to favor a guy who has won the award before, but it seems to be just the opposite. There are 3 cases in our model where we feel a former MVP should have won but did not: Barry Bonds in 1997, Jeff Bagwell in 1999, and Albert Pujols in 2006. The data shows that writers typically tend to spread the wealth and give another guy a shot at the award.
The Colorado Factor
People often talk about hitter's parks and pitcher's parks. The basic premise is that some baseball parks are more beneficial to hitters (or pitchers), and players who play 81 games (half the season) in these parks tend to have skewed statistics. We have not yet seen evidence that this shows up in the voting, with the exception being Colorado and Coors Field. According to ESPN's calculations at http://sports.espn.go.com/mlb/stats/parkfactor, Coors has been a top 3 hitter's park during each of the past 8 seasons.
Looking beyond Coors for a minute, during those 8 seasons, Chase Field and U.S. Cellular Field finished in the top 10 in all but one year. Rangers Ballpark was in the top 10 in all but 2 years. During that time, there have been few legitimate MVP candidates from these teams besides Alex Rodriguez and Luis Gonzalez. In 2002, Alex Rodriguez put up MVP caliber numbers on a bad Texas team but did not win the MVP. We think this had more to do with the Rangers than their park. So do the voters apply a park penalty outside of Coors? It hasn't appeared that way, but if Jermaine Dye doesn't do well this year in the voting, we may have to revise that thinking.
Okay, so what about Coors? Larry Walker is the only Rockie to win the MVP; he did so in 1997. He had to put up monster numbers to overcome the writers' discrimination against Colorado players. There are 5 other cases where we feel a Colorado player would have won the MVP, if not for the Coors penalty: Dante Bichette (1995), Larry Walker (1995 and 1999), Todd Helton (2000), and Matt Holliday (2007). Justified or not, this is how the writers tend to vote.
A Fresh Face
Sometimes a team will add a new piece of the puzzle that will help it make the leap to the next level. The real effect of this can be debated, but the voters tend to like players that were not on that team the previous season. In other words, if a team was good this year but bad last year, the voters tend to attribute that to a new player in town.
What about Pitchers?
Every once in awhile, voters will give the award to a pitcher, usually a relief pitcher. We have not built this into the formula, but have examined two of these cases a bit in the discussion of the AL race below.
The Years We Got It Wrong
So we took all these factors, stirred them into a big pot, and found our model selected the MVP winner 22 of the past 26 times. This means our model failed to get four of them right. Let's take a closer look at each of these.
For the 1995 AL race, our model predicted Albert Belle would win the award. Mo Vaughn was the 1995 MVP, narrowly edging him out 308-300. The oft-cited reason for giving the award to Vaughn that year was that the Indians didn't play under any pressure because they finished 30 games ahead of second place Kansas City. Others say that Belle's surly attitude with the press didn't help, but then again that never seemed to affect Bonds.
For the 2001 AL MVP award, our model predicted Roberto Alomar would win the award. That of course was the year Ichiro won both the MVP and Rookie of the Year. Incidentally, our model had Alomar first, Bret Boone second, and Ichiro third. The voters had Ichiro first and Jason Giambi a close second, followed by Boone and Alomar. We were not surprised that our model got this one wrong; the buzz following Ichiro that summer was unprecedented and difficult to measure.
The 2002 AL MVP was Miguel Tejada; our model predicted Alex Rodriguez by a good margin. A-Rod finished second in the voting, well behind Tejada. So why did Tejada win the MVP that year? You may recall some clutch hits that season by Tejada to propel the A's into the playoffs. The A's had a great stretch that year, winning 20 in a row at one point. According to Tejada's Wikipedia page, "Tejada contributed one-out, game-winning hits in the 18th and 19th games of that run: a three-run homer off Minnesota Twins closer Eddie Guardado for a 7-5 victory and a bases-loaded single against Kansas City Royals reliever Jason Grimsley to break a 6-6 tie." We think these two games made a huge impression on the voters, which is difficult to measure.
Finally, the 2006 AL MVP was Justin Morneau, while our model predicted Derek Jeter. Jeter finished a close second to Morneau, losing 320-306. Our model had Jeter first, Joe Mauer second, Carlos Guillen third, and Morneau fourth. Mauer, Guillen, and Morneau were relatively close in our model. It is interested to note that our model had teammate Mauer higher than Morneau.
Who Will Win in 2008?
Using our formulas, here are the top point totals for players for this year's awards.
National League
| Place | Player | Points |
| 1 | Albert Pujols | 1505 |
| 2 | Chase Utley | 1451 |
| 3 | Ryan Howard | 1361 |
| 4 | David Wright | 1338 |
| 5 | Jose Reyes | 1256 |
| 6 | Ryan Braun | 1249 |
| 7 | Hanley Ramirez | 1243 |
Had the Mets made the playoffs and won 90 games, David Wright would be only a few points behind Pujols. We are fairly confident that Pujols will be the 2008 NL MVP, although given this relatively low point total we would not be surprised to see a pitcher get some consideration.
American League
| Place | Player | Points |
| 1 | Jermaine Dye | 1328 |
| 2 | Carlos Quentin | 1261 |
| 3 | Joe Mauer | 1217 |
| 4 | Kevin Youkilis | 1208 |
| 5 | Dustin Pedroia | 1159 |
| 6 | Vlad Guerrero | 1149 |
| 7 | Josh Hamilton | 1100 |
The AL field is historically weak when compared to other years. To give you some perspective, in only 1 case since 1995 has the MVP winner scored fewer than 1600 points in our formula: Justin Morneau, who scored 1532 points in 2006. The teams that made the playoffs this season seem to have a bunch of good players rather than one guy who put up monster numbers. Had the Twins beat the White Sox in the 1 game playoff, Mauer would have leapfrogged the field and been our choice to win the MVP.
Because of the weak field, we could see Francisco Rodriguez getting some serious consideration. Since 1980, there have been 3 relief pitchers to win the MVP: Rollie Fingers in strike shortened 1981, Willie Hernandez in 1984, and Dennis Eckersley in 1992. Eckersley saved 51 games on a team that won 96 games. We took a look at the 1984 and 1992 runner-ups. In 1984, Kent Hrbek and Eddie Murray scored 1138 and 932 points, respectively. In 1992, Joe Carter and Kirby Puckett scored 1218 and 1165 points, respectively. Each of those seasons lacked hitters who were strong MVP candidates. This precedent points to a pitcher MVP in 2008.
But What About Manny Ramirez?
For what it is worth, if you take Manny Ramirez's full season totals and treat him as a Dodger, he ends up with 1604 points. If he had stayed in Boston, his total drops a bit to 1532. We don't believe he has a chance of winning the MVP because he was only with LA for 2 months, but it is worth mentioning that he likely would have been the AL MVP if he had stayed in Boston and put up even his pre-trade numbers.
Sim Dynasty Baseball is an online, simulated baseball game. You act as the GM and Manager of a baseball franchise and direct all aspects of that franchise, such as drafting, trades, lineups, in-game manager preferences, grooming minor leaguers, and much more. The game is free to try, sign up today at www.simdynasty.com.