Help Me Understand fWAR

[Soxsi75]

I'm an old school baseball fan too. Been following the White Sox since 1975. (Hence my name on this thread.) But I tried to keep an open mind and wanted to learn and look at more statistics like WAR. Just for the fun of it. Then I saw that our very own South Side Hitmen first baseman Jim Spencer had a WAR of -.03 in 1977. So one of the best defensive first basemen of this era and a solid power hitter was worse than a replacement player? And what made it even worse was his defensive rating was -.08?? What?? He saved Alan Bannister countless errors that year!! But a replacement player would be better than him? Had me and you lost me there. 

And please. I don't want any "Sabergeeks" getting on here and responding why this should be. It shouldn't. Pure and simple. If someone wants to talk baseball, I'd love to talk about that. But I don't want to talk any "Sabergeekness."

 

[michelangelosmonkey]

19 minutes ago, Soxsi75 said:

I'm an old school baseball fan too. Been following the White Sox since 1975. (Hence my name on this thread.) But I tried to keep an open mind and wanted to learn and look at more statistics like WAR. Just for the fun of it. Then I saw that our very own South Side Hitmen first baseman Jim Spencer had a WAR of -.03 in 1977. So one of the best defensive first basemen of this era and a solid power hitter was worse than a replacement player? And what made it even worse was his defensive rating was -.08?? What?? He saved Alan Bannister countless errors that year!! But a replacement player would be better than him? Had me and you lost me there. 

And please. I don't want any "Sabergeeks" getting on here and responding why this should be. It shouldn't. Pure and simple. If someone wants to talk baseball, I'd love to talk about that. But I don't want to talk any "Sabergeekness."

 

That's interesting...I remember Jim Spencer too and that was what I remembered too...that he was this amazing fielder.   But he was NOT a good hitting 1B.   Even using old timey stats...in 76 and 77 he averaged about 550 at bats, hit about .250 and averaged 16 home runs.   We were just so broken as baseball fans in 1976 that we pretended that was good.  STill the fielding thing is a bit strange...maybe we just were kids and heard the announcers say it over and over again that he was an amazing fielder.  When you look at fielding stats he never made errors....which is good...and had a lot of put outs...but that could be because their rotation was an extreme ground ball rotation.   Remember...we were all told that Derek Jeter was a GREAT fielder until the saber heads proved otherwise.  Don't give up...the new stats are worth learning.    

[Soxsi75]

3 minutes ago, michelangelosmonkey said:

That's interesting...I remember Jim Spencer too and that was what I remembered too...that he was this amazing fielder.   But he was NOT a good hitting 1B.   Even using old timey stats...in 76 and 77 he averaged about 550 at bats, hit about .250 and averaged 16 home runs.   We were just so broken as baseball fans in 1976 that we pretended that was good.  STill the fielding thing is a bit strange...maybe we just were kids and heard the announcers say it over and over again that he was an amazing fielder.  When you look at fielding stats he never made errors....which is good...and had a lot of put outs...but that could be because their rotation was an extreme ground ball rotation.   Remember...we were all told that Derek Jeter was a GREAT fielder until the saber heads proved otherwise.  Don't give up...the new stats are worth learning.    

Good point and well put. Thanks.

 

[South Side Hit Men]

19 minutes ago, Soxsi75 said:

I'm an old school baseball fan too. Been following the White Sox since 1975. (Hence my name on this thread.) But I tried to keep an open mind and wanted to learn and look at more statistics like WAR. Just for the fun of it. Then I saw that our very own South Side Hitmen first baseman Jim Spencer had a WAR of -.03 in 1977. So one of the best defensive first basemen of this era and a solid power hitter was worse than a replacement player? And what made it even worse was his defensive rating was -.08?? What?? He saved Alan Bannister countless errors that year!! But a replacement player would be better than him? Had me and you lost me there. 

And please. I don't want any "Sabergeeks" getting on here and responding why this should be. It shouldn't. Pure and simple. If someone wants to talk baseball, I'd love to talk about that. But I don't want to talk any "Sabergeekness."

Well, that is the year I started following team as well. Most friends and neighbors were Cub fans, growing up on the NW Side (family started out on West Side when immigrating here a hundred years ago), but family were White Sox (and Football Cardinals) fans. Really loved the White Sox uniforms in 1975, also loved the 1977 team and games WSNS. 

In terms of the bWAR numbers, he was always rated as a below average 1B by the numbers, and I'm not really sure why. He won the Gold Glove that year, and while sometimes the "eye ball test" does not give a full picture, looking at the range and other metrics in the fielding section, I'm don't understand the -0.8 dWAR number either. Would be interesting if the same information available now was available for 1977, and whether his dWAR would improve as a result.

A -0.3 bWAR is basically replacement level average. While Batting Average, Home Runs and Runs Batted In were the primary stats used to evaluate a first baseman back then, I do understand the low bWAR rating in this instance, as a .700 OPS (On Base + Slugging) is a poor number, especially for a first baseman. It was the second lowest on the team (Orta's .751 the next lowest, Bannister was the lowest at .672). The league average OPS across all AL players was .735 that year, so being below league average as a 1B really indicates a poor hitting season. He had a huge season two years later with the short RF porch in Yankee Stadium, bumping his OBP to .960. the bWAR was likely adjusted downward a bit, accounting for the ballpark difference (Comiskey much harder than Yankee Stadium for LHBs).

[Soxsi75]

3 minutes ago, South Side Hit Men said:

Well, that is the year I started following team as well. Most friends and neighbors were Cub fans, growing up on the NW Side (family started out on West Side when immigrating here a hundred years ago), but family were White Sox (and Football Cardinals) fans. Really loved the White Sox uniforms in 1975, also loved the 1977 team and games WSNS. 

In terms of the bWAR numbers, he was always rated as a below average 1B by the numbers, and I'm not really sure why. He won the Gold Glove that year, and while sometimes the "eye ball test" does not give a full picture, looking at the range and other metrics in the fielding section, I'm don't understand the -0.8 dWAR number either. Would be interesting if the same information available now was available for 1977, and whether his dWAR would improve as a result.

A -0.3 bWAR is basically replacement level average. While Batting Average, Home Runs and Runs Batted In were the primary stats used to evaluate a first baseman back then, I do understand the low bWAR rating in this instance, as a .700 OPS (On Base + Slugging) is a poor number, especially for a first baseman. It was the second lowest on the team (Orta's .751 the next lowest, Bannister was the lowest at .672). The league average OPS across all AL players was .735 that year, so being below league average as a 1B really indicates a poor hitting season. He had a huge season two years later with the short RF porch in Yankee Stadium, bumping his OBP to .960. the bWAR was likely adjusted downward a bit, accounting for the ballpark difference (Comiskey much harder than Yankee Stadium for LHBs).

I loved those 1975 uniforms too!! One of the reasons I became a White Sox fan!! And good old WSNS!! And maybe you're right, with the information available today, maybe it would be different?

[Dam8610]

On 1/24/2021 at 8:40 AM, poppysox said:

As an old school baseball fan, I relate to home runs, rbi's, sb's, runs scored, etc...  WAR seems to track everything that happens everywhere on the field and grade it.  How is this possible?  Wouldn't that require dozens of watchers trained to score performance?  It seems impossible to have all those scorekeepers using a fair and unbias system since we have trouble agreeing on an error call for example.  Jose Abreu won the MVP but 3 position players had better WAR scores...why did Jose win?  This question is not meant to be a clever way to say WAR is somehow BS.  Some of you put great stock in it but I have trouble understanding why you value the opinions of people you don't know.  I do see a usefulness for baseball executives having their own in-house grading systems but the average fan is lost when he tries to grasp what is WAR.  Please try and explain in a clear non-condescending manner as I am sure many would like a better grasp of this subject. 

I'm bored so I'll give this a go. 

The first thing to understand is that fWAR is not tracking everything. Rather, fWAR is the coming together of several other tracking statistics to quantify the player's overall performance and its relative value to the team in an easily articulable number that most baseball fans understand, namely "How many wins was [insert player here] worth to his team?" That may seem like an impossible task on its face, however, statistical modeling has shown us that the results that happen on the field correlate to an expected value of runs produced, and that a certain number of runs produced is equivalent to a win produced. Some of the math has already been referenced here, so I'm not going to get into the details, but the importance of this point is that the numbers from fWAR do not represent "someone else's opinion" but an objective calculation of a player's relative value based on concepts that have met the rigorous standards of mathematical proof. Having studied mathematics myself, I know how rigorous those standards are, they make a prosecuting attorney's job in a murder case look easy.

Once we've established that fWAR is an objective measure, the next question is: what is fWAR measuring, exactly? I gave a rather simplistic definition earlier, but more context is needed. The name gives this context: fWAR stands for fangraphs Wins Above Replacement, which means that the value being measured is relative to a replacement player. This is where fWAR (and bWAR, because both use the same baseline for WAR) get into some assumptions that have certainly held up mathematically over time, but they are still assumptions, not proven, and therefore up for debate, though the evidence in their favor is strong. The theory behind this is that if a player gets hurt, that player must be replaced, and the type of player available to replace that player would not be a league average player, but a "replacement level" player, someone who is a free agent or currently on your AAA roster who will likely perform well below league average. Thus, the numbers that Fangraphs and Baseball Reference have agreed on are that they will assume that a team full of replacement players will produce a .294 winning percentage, or an average of 47.7 wins, leaving 1,000 WAR over 2,430 MLB games to be earned by the 780 players on MLB rosters at any given moment. Mathematical regression has shown these assumptions to be valid season after season, but that could change in the future. These 1,000 fWAR are currently divided up as 570 fWAR allocated to position players and 430 fWAR allocated to pitchers. Again, these allocations are supported by mathematical regression, but are subject to change in the future. The existence of a finite amount of fWAR per season can bring understanding to certain concepts like a player having a negative fWAR and illuminates the excellence of certain performances, such as Mike Trout accounting for ~1% of all fWAR in baseball for several years of his career.

Finally some things to keep in mind. Regarding relative value, fWAR is most appropriately viewed through the lens of ranges as opposed to exacts. For example, if Player A produces 4.4 fWAR and Player B produces 4.1 fWAR, they effectively produced the same value. If Player C produces 7.6 fWAR, we can conclude that Player C produced a great deal more value than Player A or Player B. That being the case, the question then becomes what's the demarcation point? The answer to that typically lies in the eye of the beholder, but 0.5 fWAR and 1.0 fWAR are good and fairly common ones, meaning one person could look at 0.5 fWAR as separating one player from another in value, where another could look at 1.0 fWAR as separating one player from another. Also, remember fWAR is a counting stat, akin to HR and RBI. This means the more opportunities a player has in a given season, the more likely they are to have a higher fWAR. This is why it is difficult for relievers and utility players to put up high fWAR totals, they don't get nearly as many opportunities as starting position players and starting pitchers. Thus, it is not likely the best measure of relative value of a utility player or reliever, whereas it is possibly the best measure of relative value of a starting position player or starting pitcher, and certainly is the quickest and most easily understandable measure of relative value of those players, hence the popularity it has enjoyed in the analytics community for some time now.

I hope this met the objective of "Please try and explain in a clear non-condescending manner as I am sure many would like a better grasp of this subject." If I wasn't clear, please ask any question, though I don't know that I'll have the answer, and if I was somehow condescending, that was not my intent.

[poppysox]

11 hours ago, Dam8610 said:

I'm bored so I'll give this a go. 

The first thing to understand is that fWAR is not tracking everything. Rather, fWAR is the coming together of several other tracking statistics to quantify the player's overall performance and its relative value to the team in an easily articulable number that most baseball fans understand, namely "How many wins was [insert player here] worth to his team?" That may seem like an impossible task on its face, however, statistical modeling has shown us that the results that happen on the field correlate to an expected value of runs produced, and that a certain number of runs produced is equivalent to a win produced. Some of the math has already been referenced here, so I'm not going to get into the details, but the importance of this point is that the numbers from fWAR do not represent "someone else's opinion" but an objective calculation of a player's relative value based on concepts that have met the rigorous standards of mathematical proof. Having studied mathematics myself, I know how rigorous those standards are, they make a prosecuting attorney's job in a murder case look easy.

Once we've established that fWAR is an objective measure, the next question is: what is fWAR measuring, exactly? I gave a rather simplistic definition earlier, but more context is needed. The name gives this context: fWAR stands for fangraphs Wins Above Replacement, which means that the value being measured is relative to a replacement player. This is where fWAR (and bWAR, because both use the same baseline for WAR) get into some assumptions that have certainly held up mathematically over time, but they are still assumptions, not proven, and therefore up for debate, though the evidence in their favor is strong. The theory behind this is that if a player gets hurt, that player must be replaced, and the type of player available to replace that player would not be a league average player, but a "replacement level" player, someone who is a free agent or currently on your AAA roster who will likely perform well below league average. Thus, the numbers that Fangraphs and Baseball Reference have agreed on are that they will assume that a team full of replacement players will produce a .294 winning percentage, or an average of 47.7 wins, leaving 1,000 WAR over 2,430 MLB games to be earned by the 780 players on MLB rosters at any given moment. Mathematical regression has shown these assumptions to be valid season after season, but that could change in the future. These 1,000 fWAR are currently divided up as 570 fWAR allocated to position players and 430 fWAR allocated to pitchers. Again, these allocations are supported by mathematical regression, but are subject to change in the future. The existence of a finite amount of fWAR per season can bring understanding to certain concepts like a player having a negative fWAR and illuminates the excellence of certain performances, such as Mike Trout accounting for ~1% of all fWAR in baseball for several years of his career.

Finally some things to keep in mind. Regarding relative value, fWAR is most appropriately viewed through the lens of ranges as opposed to exacts. For example, if Player A produces 4.4 fWAR and Player B produces 4.1 fWAR, they effectively produced the same value. If Player C produces 7.6 fWAR, we can conclude that Player C produced a great deal more value than Player A or Player B. That being the case, the question then becomes what's the demarcation point? The answer to that typically lies in the eye of the beholder, but 0.5 fWAR and 1.0 fWAR are good and fairly common ones, meaning one person could look at 0.5 fWAR as separating one player from another in value, where another could look at 1.0 fWAR as separating one player from another. Also, remember fWAR is a counting stat, akin to HR and RBI. This means the more opportunities a player has in a given season, the more likely they are to have a higher fWAR. This is why it is difficult for relievers and utility players to put up high fWAR totals, they don't get nearly as many opportunities as starting position players and starting pitchers. Thus, it is not likely the best measure of relative value of a utility player or reliever, whereas it is possibly the best measure of relative value of a starting position player or starting pitcher, and certainly is the quickest and most easily understandable measure of relative value of those players, hence the popularity it has enjoyed in the analytics community for some time now.

I hope this met the objective of "Please try and explain in a clear non-condescending manner as I am sure many would like a better grasp of this subject." If I wasn't clear, please ask any question, though I don't know that I'll have the answer, and if I was somehow condescending, that was not my intent.

Very good attempt at explaining a very difficult (IMO) subject.  I especially got a better grasp of the utility player vs full-time player.  The Moneyball image I had was that fWAR was used by FO types to uncover the hidden gems among players being underutilized.  In any event...thanks for your efforts.

[Tony]

1 hour ago, poppysox said:

Very good attempt at explaining a very difficult (IMO) subject.  I especially got a better grasp of the utility player vs full-time player.  The Moneyball image I had was that fWAR was used by FO types to uncover the hidden gems among players being underutilized.  In any event...thanks for your efforts.

I think that’s part of your problems, poppy. You aren’t going to agree with it or not going to understand it if you think it’s used for one thing...when it’s really used for another. 
 

Here is what my advice would be, and this isn’t just about WAR. Use as much information as possible, including your eyes. I’ve beaten this point into the ground and I will continue to: No front office just uses one stat to evaluate a player. Neither should you. 
 

We have SO much information available to us in 2021. Why would you use one stat to try and tell you about a player when you have hundreds to look at? I don’t look at just K/9 when looking at a relief pitcher. It’s one of the stats I like to look at, but it doesn’t tell the whole story. I love OBP, but I can’t just use OBP to evaluate a hitter. And why would I? I can look at so many other stats to help me paint a better picture of a player. 
 

WAR tries to evaluate overall performance and there is some value there for sure. But only using WAR would be a mistake. You should use it! Just don’t use it alone.