All about the long overdue death of Range Factor
The third big change slated for early 2018 is all about defense. I wanted to talk about it now, because it?s likely going to roll out with the player rank change discussed in my last blog post.
As everyone knows, Range Factor is currently the stat used on Pennant Chase to add a defensive element to the game (beside Fielding Percentage, which simply determines the likelihood of a player making an error).
Range Factor was not an element that PC launched with back in 2008. The only defensive element was Fielding Percentage. Range Factor came in later, once users started requesting a feature to add defensive strategy to the game. It was an easy stat to use, because we already had what we needed in the database: Putouts, Assists and Games Played.
Range Factor is a simple stat. It?s just (PO+A) / GP. Another version of Range Factor breaks it down into PO+A per nine innings, which is a better version, but one that we didn?t have access to. The idea behind Range Factor is that a player who is involved in a lot of plays must be able to cover a wider portion of the field. Not surprisingly, this stat no longer has any value (if it ever did) in a world where many other attempts to measure defense have cropped up.
Anyone who has watched a lot of baseball knows that a fielder involved in a lot of plays doesn?t necessarily have good range. Maybe he?s an infielder playing behind a lot of ground-ball pitchers. Maybe he?s an outfielder playing in a large ballpark. There are so many reasons why one player might accumulate more opportunities than his peers.
Nonetheless, the Pennant Chase community embraced Range Factor with very few complaints. Sure, people would comment once in a while on its inaccuracy, but everyone seemed to accept it was the only way to measure defense in the sim. Most people seemed grateful there was some measure of defense involved, even if it was an imperfect one.
Defense continues to be an incredibly challenging thing to measure. Newer technology has led to stats like ?Ultimate Zone Rating? where statisticians analyze tracking data to determine if players made exceptional plays, but defensive shifts and other factors can result in inaccuracies. Data scientists try to get as accurate as possible by comparing historical data, but there?s a level of subjectivity that?s inherently baked into the entire formula.
That said, there?s not much choice but to use a slightly subjective statistic if we want to make defense a part of the game. Defensive Runs Saved is now a popular measure, which attempts to figure out how many runs a fielder saved by quantifying his entire defensive performance. Defensive WAR uses both UZR and DRS to figure out how many wins a player contributed to his team with his defensive.
By now, any baseball fan has heard about WAR. The concept of Wins Above Replacement (WAR) is to try to figure out how much value a player has over the average ?replacement? player. WAR is far too complex to get into in this blog post, but it?s important to note that it?s relative to the player?s league and the position he plays. A second baseman who clubs 50 homers is likely to have a huge WAR because most second basemen don?t produce that kind of offense. [Edit: A user informed me this is inaccurate because homers are not heavily weighted in WAR. The point I believe is still valid - WAR does have some position adjustments.] WAR is also very subjective - so much so that there is no one universal WAR formula. Two major websites, Baseball Reference and Fangraphs, have completely different formulas (although they have aligned closer recently).
So what are we going to use and what?s happening to Range Factor?
Range Factor is going to be removed from the sim, but not right away. Since so many people have built teams around RF, I?m not going to yank it immediately. At the same time, I?m going to rollout the stat we?re going to use, which is Defensive WAR. We?ll be using the Baseball Reference calculation as of January 2018. It?s important to note, websites like Baseball Reference are constantly adjusting their WAR formulas, but I?m not going to be re-importing WAR for every player on a regular basis. So what you see on PC may not match any other website.
The historical range of dWAR is basically somewhere around -5.0 to 5.0. Elite players tend to be over 2.0. If your dWAR is zero, that means you are a perfectly average player - your defense adds no extra value, but it doesn?t hurt your team either. If your dWAR is negative, it means you?re usually a liability in the field and not an asset. If your dWAR is positive, your defense can contribute to your team winning games that you may not have otherwise won.
Okay, so how is dWAR going to work?
It?s important to note that unlike Range Factor, dWAR is going to work on an individual basis and not a team basis. (Because of the flaws in Range Factor, it was added up and used at a team level in the sim. If your team RF was high, you could benefit.) But dWAR will come into play when the fielder is in action.
Here?s the best way to figure out what the number means in PC: multiply it by 10 and then divide it by 3. That number is (roughly) the odds your player will make a positive or negative play any time a ball is hit to him. For example, if someone?s dWAR is 1.2, you?d get the number 4 (if you multiply it by 10 and then divided by 3). That means there?s a 4% chance your player will rob a hit any time the ball is hit to him. The opposite is true for a negative number. A negative dWAR means there?s a 4% chance your player will allow a hit that should have been fielded. (This is completely different than an error.)
Yes, a player with a negative dWAR can sometimes make a good play, and vice versa, but that?s rare. Much like Range Factor, these impacts don?t show up in the box score, but they can have a very real effect. You can see how 4% odds add up when you consider how many times a player has a ball hit in his zone over the course of a game. In my trial runs, dWAR was impacting up to 5 or 6 hits a game, although the norm was between 2 - 4. And some games it never comes into play.
What about WAR or oWAR?
I chose to use dWAR because it?s one of the better options available to measure defense. But there is no great way to use overall WAR or Offensive WAR in the sim. When you start looking at overall WAR, it?s very dependent on the league and position played. It?s a measure of how valuable a player was at his position, not a measure of how great he was in the history of baseball or even if he was the best overall player that year, although people do tend to use WAR in that way. The other problematic thing is that there?s very little difference in WAR of great players. Adrian Beltre in 2004 and Mookie Betts in 2016 are just 1.5 WAR points away from Mickey Mantle?s 1956 season. I don?t think anyone feels those are remotely equal comparisons. [Edit: A user informed me that using the term "WAR points" is not accurate. WAR measures "wins", which I was aware of. But the user also said WAR is an approximation (that's an understatement!) and that a higher WAR shouldn't necessarily be viewed as "better". In my opinion, that only further enhances the case that Overall WAR is not ideal for a sim that is trying to rely on tangible stats.]
But since dWAR is only impacting defense, and it does a good job of telling us if someone was a valuable defender, it?s much easier to use and get realistic results. We can all get behind Ozzie Smith or Andrelton Simmons robbing more hits than Manny Ramirez, and we can also agree that placing Manny in your outfield should come with a defensive risk.
I?m really excited for this change. I think it?s going to keep a very important defensive strategy in PC, but it?s going to create value for the truly exceptional defenders. It?s also going to do something that Range Factor never did - create liabilities in the field, which creates tough lineup decisions for managers. Much like we saw people create entire strategies around Range Factor, I think we?re going to see something similar with dWAR.
As usual, drop a note on the boards with any questions or comments.
As everyone knows, Range Factor is currently the stat used on Pennant Chase to add a defensive element to the game (beside Fielding Percentage, which simply determines the likelihood of a player making an error).
Range Factor was not an element that PC launched with back in 2008. The only defensive element was Fielding Percentage. Range Factor came in later, once users started requesting a feature to add defensive strategy to the game. It was an easy stat to use, because we already had what we needed in the database: Putouts, Assists and Games Played.
Range Factor is a simple stat. It?s just (PO+A) / GP. Another version of Range Factor breaks it down into PO+A per nine innings, which is a better version, but one that we didn?t have access to. The idea behind Range Factor is that a player who is involved in a lot of plays must be able to cover a wider portion of the field. Not surprisingly, this stat no longer has any value (if it ever did) in a world where many other attempts to measure defense have cropped up.
Anyone who has watched a lot of baseball knows that a fielder involved in a lot of plays doesn?t necessarily have good range. Maybe he?s an infielder playing behind a lot of ground-ball pitchers. Maybe he?s an outfielder playing in a large ballpark. There are so many reasons why one player might accumulate more opportunities than his peers.
Nonetheless, the Pennant Chase community embraced Range Factor with very few complaints. Sure, people would comment once in a while on its inaccuracy, but everyone seemed to accept it was the only way to measure defense in the sim. Most people seemed grateful there was some measure of defense involved, even if it was an imperfect one.
Defense continues to be an incredibly challenging thing to measure. Newer technology has led to stats like ?Ultimate Zone Rating? where statisticians analyze tracking data to determine if players made exceptional plays, but defensive shifts and other factors can result in inaccuracies. Data scientists try to get as accurate as possible by comparing historical data, but there?s a level of subjectivity that?s inherently baked into the entire formula.
That said, there?s not much choice but to use a slightly subjective statistic if we want to make defense a part of the game. Defensive Runs Saved is now a popular measure, which attempts to figure out how many runs a fielder saved by quantifying his entire defensive performance. Defensive WAR uses both UZR and DRS to figure out how many wins a player contributed to his team with his defensive.
By now, any baseball fan has heard about WAR. The concept of Wins Above Replacement (WAR) is to try to figure out how much value a player has over the average ?replacement? player. WAR is far too complex to get into in this blog post, but it?s important to note that it?s relative to the player?s league and the position he plays. A second baseman who clubs 50 homers is likely to have a huge WAR because most second basemen don?t produce that kind of offense. [Edit: A user informed me this is inaccurate because homers are not heavily weighted in WAR. The point I believe is still valid - WAR does have some position adjustments.] WAR is also very subjective - so much so that there is no one universal WAR formula. Two major websites, Baseball Reference and Fangraphs, have completely different formulas (although they have aligned closer recently).
So what are we going to use and what?s happening to Range Factor?
Range Factor is going to be removed from the sim, but not right away. Since so many people have built teams around RF, I?m not going to yank it immediately. At the same time, I?m going to rollout the stat we?re going to use, which is Defensive WAR. We?ll be using the Baseball Reference calculation as of January 2018. It?s important to note, websites like Baseball Reference are constantly adjusting their WAR formulas, but I?m not going to be re-importing WAR for every player on a regular basis. So what you see on PC may not match any other website.
The historical range of dWAR is basically somewhere around -5.0 to 5.0. Elite players tend to be over 2.0. If your dWAR is zero, that means you are a perfectly average player - your defense adds no extra value, but it doesn?t hurt your team either. If your dWAR is negative, it means you?re usually a liability in the field and not an asset. If your dWAR is positive, your defense can contribute to your team winning games that you may not have otherwise won.
Okay, so how is dWAR going to work?
It?s important to note that unlike Range Factor, dWAR is going to work on an individual basis and not a team basis. (Because of the flaws in Range Factor, it was added up and used at a team level in the sim. If your team RF was high, you could benefit.) But dWAR will come into play when the fielder is in action.
Here?s the best way to figure out what the number means in PC: multiply it by 10 and then divide it by 3. That number is (roughly) the odds your player will make a positive or negative play any time a ball is hit to him. For example, if someone?s dWAR is 1.2, you?d get the number 4 (if you multiply it by 10 and then divided by 3). That means there?s a 4% chance your player will rob a hit any time the ball is hit to him. The opposite is true for a negative number. A negative dWAR means there?s a 4% chance your player will allow a hit that should have been fielded. (This is completely different than an error.)
Yes, a player with a negative dWAR can sometimes make a good play, and vice versa, but that?s rare. Much like Range Factor, these impacts don?t show up in the box score, but they can have a very real effect. You can see how 4% odds add up when you consider how many times a player has a ball hit in his zone over the course of a game. In my trial runs, dWAR was impacting up to 5 or 6 hits a game, although the norm was between 2 - 4. And some games it never comes into play.
What about WAR or oWAR?
I chose to use dWAR because it?s one of the better options available to measure defense. But there is no great way to use overall WAR or Offensive WAR in the sim. When you start looking at overall WAR, it?s very dependent on the league and position played. It?s a measure of how valuable a player was at his position, not a measure of how great he was in the history of baseball or even if he was the best overall player that year, although people do tend to use WAR in that way. The other problematic thing is that there?s very little difference in WAR of great players. Adrian Beltre in 2004 and Mookie Betts in 2016 are just 1.5 WAR points away from Mickey Mantle?s 1956 season. I don?t think anyone feels those are remotely equal comparisons. [Edit: A user informed me that using the term "WAR points" is not accurate. WAR measures "wins", which I was aware of. But the user also said WAR is an approximation (that's an understatement!) and that a higher WAR shouldn't necessarily be viewed as "better". In my opinion, that only further enhances the case that Overall WAR is not ideal for a sim that is trying to rely on tangible stats.]
But since dWAR is only impacting defense, and it does a good job of telling us if someone was a valuable defender, it?s much easier to use and get realistic results. We can all get behind Ozzie Smith or Andrelton Simmons robbing more hits than Manny Ramirez, and we can also agree that placing Manny in your outfield should come with a defensive risk.
I?m really excited for this change. I think it?s going to keep a very important defensive strategy in PC, but it?s going to create value for the truly exceptional defenders. It?s also going to do something that Range Factor never did - create liabilities in the field, which creates tough lineup decisions for managers. Much like we saw people create entire strategies around Range Factor, I think we?re going to see something similar with dWAR.
As usual, drop a note on the boards with any questions or comments.
Logged in users can send feedback via the feedback form or make a comment on the public board.
