I’m curious. Given that it’s Wins Above Replacement, is it possible that Replacement Level is being redefined to a lower amount of wins? Either way, interesting article.
I’m not sure. It’s been a rant of mine the last few years that WAR, whatever version, doesn’t actually represent what the name implies. If replacement is average (0.0) and the Dodgers are projected to have 50+ fWAR should we expect them to win 131+ games? 81 (average) + 50 above average. The name messes with my logical brain and now I find out the projections apparently ignore the upper limits of fWAR.
There’s likely some reasonable explanation I’m missing.
Right, historically a replacement level team wins about 60 games. Replacement level is supposed to represent the wins a team has if all players were league minimum guys (i.e. replacement level is who a team could field with the absolute minimum resources). So this might be part of the explanation.
I think you hit on a part I was missing - 60 wins vs 81 for replacement. That would put the Dodgers at 110 in my example, and while not likely, at least theoretically possible.
I don’t think I’ve ever seen that number - 60 - can you point to a link?
Still wondering about the extra 180 WAR and “everyone” improving, though. 😂
I suspect this is happening because the preseason projections don't really have many (or any) sub-replacement level players -- since very few players projected to make and stay on an MLB roster are sub-replacement talent -- but in reality, some amount of players will perform at a sub-replacement level due to randomness, talent mis-judgment, etc. Some players _will_ be sub-replacement regardless, but the projections don't really know who will miss their mark and who won't.
So here we're kind of mixing backwards-looking value assessments (last year's WAR) with forward-looking talent projections (preseason forecasts), which have two distinct purposes.
Thanks, Neil. I understand your point about the mixing and plan to compare this year's projections with actual results next offseason.
Not eloquently written, but I would think the goal would be to reflect what is most likely to happen in the regular season, or am I missing the point of the projections?
It seems illogical (to me, anyway) to say, we know history shows X% of players will end up negative, but we're not going to project that (or even attempt to).
Last season 3.4% of Qualified players had negative fWAR and 11.4% of players with 160 PA had negative fWAR. It just feels like this is a missing piece of the projections.
An extreme example is the Rockies, who as a team had a negative fWAR last season and 32 players with negative fWAR. The projections show only 6 (of 50) projected negative fWARs and none lower than -0.1.
I'm mixing again, but it just seems way off to me, as does the Rockies ending up with 22.4 fWAR this coming season, but I guess we'll see.
I definitely had to do a double-take when I saw your chart as well, because you are right, it is odd to see (especially that Rockies number, lol).
Maybe there are also built-in, possibly faulty, assumptions that teams will not systematically tolerate sustained sub-replacement play without looking for improved production at a position, and that players' individual odds of over/under-performing their projections are independent within a team.
In reality, some teams are perhaps, um, deliberately tolerating sub-replacement play for their own reasons, and we've all seen seasons snowball out of control where it seems like every player on a roster has a bad year at once.
I’m curious. Given that it’s Wins Above Replacement, is it possible that Replacement Level is being redefined to a lower amount of wins? Either way, interesting article.
I’m not sure. It’s been a rant of mine the last few years that WAR, whatever version, doesn’t actually represent what the name implies. If replacement is average (0.0) and the Dodgers are projected to have 50+ fWAR should we expect them to win 131+ games? 81 (average) + 50 above average. The name messes with my logical brain and now I find out the projections apparently ignore the upper limits of fWAR.
There’s likely some reasonable explanation I’m missing.
Thanks for chiming in.
Right, historically a replacement level team wins about 60 games. Replacement level is supposed to represent the wins a team has if all players were league minimum guys (i.e. replacement level is who a team could field with the absolute minimum resources). So this might be part of the explanation.
I think you hit on a part I was missing - 60 wins vs 81 for replacement. That would put the Dodgers at 110 in my example, and while not likely, at least theoretically possible.
I don’t think I’ve ever seen that number - 60 - can you point to a link?
Still wondering about the extra 180 WAR and “everyone” improving, though. 😂
Also I might be wrong about the 60, but that’s the general idea
Right yeah I’m not sure the answer to the last part hahaha
but here’s a link that should help with the rest: https://library.fangraphs.com/misc/war/replacement-level/
That’s good info as they discern between average and replacement, so that helps, thanks.
They also confirm the 1,000 total.
Appreciate the convo and info.
Very cool. Well we didn't quite answer the question (why it's over 1,000 WAR in projections) but we sure had fun trying!
I suspect this is happening because the preseason projections don't really have many (or any) sub-replacement level players -- since very few players projected to make and stay on an MLB roster are sub-replacement talent -- but in reality, some amount of players will perform at a sub-replacement level due to randomness, talent mis-judgment, etc. Some players _will_ be sub-replacement regardless, but the projections don't really know who will miss their mark and who won't.
So here we're kind of mixing backwards-looking value assessments (last year's WAR) with forward-looking talent projections (preseason forecasts), which have two distinct purposes.
Thanks, Neil. I understand your point about the mixing and plan to compare this year's projections with actual results next offseason.
Not eloquently written, but I would think the goal would be to reflect what is most likely to happen in the regular season, or am I missing the point of the projections?
It seems illogical (to me, anyway) to say, we know history shows X% of players will end up negative, but we're not going to project that (or even attempt to).
Last season 3.4% of Qualified players had negative fWAR and 11.4% of players with 160 PA had negative fWAR. It just feels like this is a missing piece of the projections.
An extreme example is the Rockies, who as a team had a negative fWAR last season and 32 players with negative fWAR. The projections show only 6 (of 50) projected negative fWARs and none lower than -0.1.
I'm mixing again, but it just seems way off to me, as does the Rockies ending up with 22.4 fWAR this coming season, but I guess we'll see.
Appreciate your feedback.
I definitely had to do a double-take when I saw your chart as well, because you are right, it is odd to see (especially that Rockies number, lol).
Maybe there are also built-in, possibly faulty, assumptions that teams will not systematically tolerate sustained sub-replacement play without looking for improved production at a position, and that players' individual odds of over/under-performing their projections are independent within a team.
In reality, some teams are perhaps, um, deliberately tolerating sub-replacement play for their own reasons, and we've all seen seasons snowball out of control where it seems like every player on a roster has a bad year at once.