Replacement Level Players

by Jeff Sullivan on Feb 3, 2009 4:25 PM PST in Seattle Mariners Stats

For those of you who are familiar with the idea but somewhat skeptical of its legitimacy, you're going to want to check out the work Dave's been doing over at Fangraphs:

I don't want to spoil anything, since there are still some more entries left in the series, but these players are real, and - surprise! - they're worth right around two wins below average. Which is where the WAR equation puts them.

Solid stuff. Whenever you raise the replacement level idea around someone new to the subject, that person always asks for examples. This has the makings of an excellent list.

Tweet Comment 15 comments | 0 recs |

Do you like this story?

Comments

What I learned from this is

1 point of wOBA in relation to the .330 average is ~.5 runs difference.

Is this a typically true statement?

by seattlebruin on Feb 3, 2009 4:33 PM PST reply actions

The runs difference is (X - .330) / 1.15 * 600

Where the 600 represents PAs in a full season. (A full full season is 700 PAs, but people are generally projected to miss some time.) Given this equation, a .5 run difference is equal to just under one point of wOBA.

by Jeff Sullivan on Feb 3, 2009 4:43 PM PST up reply actions

Gotcha

Why did Dave use 1.20 in the FanGraphs article?

To me, it’s just a bit easier to be able to grasp some kind of run value that a point of wOBA represents in relation to average, so this was actually really helpful.

by seattlebruin on Feb 3, 2009 4:46 PM PST up reply actions

I'm not sure and am hoping he drops by to answer that

by Jeff Sullivan on Feb 3, 2009 4:48 PM PST up reply actions

And as an aside, where does the 1.15 come from?

Is it just some scalar that’s already been calculated to translate wOBA to runs?

by seattlebruin on Feb 3, 2009 4:53 PM PST up reply actions

Good explanation in here

http://www.insidethebook.com/woba.shtml

by Jeff Sullivan on Feb 3, 2009 4:53 PM PST up reply actions

I think it would have made more sense to just do away with the 15% scalar to make it look like OBP and let it look like batting average

but whatever, that’s easy enough to understand, at least.

by seattlebruin on Feb 3, 2009 4:56 PM PST up reply actions

Using 1.2 means that one point is exactly 0.5 runs, by the way

by Jeff Sullivan on Feb 3, 2009 5:01 PM PST up reply actions

The wOBA scale isn't fixed

It varies from year to year, depending on the league wOBA. When I talked David Appelman into adding wOBA on FanGraphs, Tom walked us through the process of calculating wOBA year by year, and then posted the results on his site. The year by year values can be found here. The scale hasn’t been exactly 1.15 since 2000, and it’s been higher the the last few years.

by davidcameron on Feb 3, 2009 5:49 PM PST up reply actions

Really cool series

What I’d love to see is some work on the not-quite-replacement-level and yet not-quite-MLB-regular types that are on each team’s 40-man roster.
Some of them may actually BE replacement level, but many of them are something more akin to a bench player: Tug Hulett for the M’s has a wOBA projection that’s substantially above any of the SS’s and it’s above the average of the 2b group by a noticeable amount. You’d sort of expect that, given that Hulett isn’t ‘freely available.’
We typically think of MiLB 40 man players as prospects, but some are sort of in between prospects and bench players – Wlad’s another one for us, but you could use Brad Nelson of the Brewers too.

So: is there ANY consistency in value for a group like this? I remember Tango doing a study of MLB regulars, bench players and then the freely-available/replacement level types, and maybe averaging the last two groups might get it close.

I’m rambling, so again: nice work Dave.

by marc w on Feb 3, 2009 5:04 PM PST reply actions

I still have to wonder if the baseline of "replacement player"

is the best we can come up with. I feel like replacement players might be a little too poor. I feel like any team could assemble a league minimum team and win about 50 games fairly easily. Heck, for a team to lose more than 100 they must be pretty horrible. Usually teams have plenty of AAAA players who are worth more than 0 WAR. Any team might not have enough of these AAAA level players to construct a 25 man roster but usually I’d say there are 5-10 AAAA and young MLB players that would be above replacement.

For example, most teams seem to have a #6 SP stashed in AAA that would be above replacement.

I know a lot of thought has gone into setting these levels. Setting the level where it has been set at makes sense to me but it often seems that there are a fair amount of freely available players that would be slightly above replacement.

by Edgar for Pres on Feb 3, 2009 11:12 PM PST reply actions

But what do those players cost?

Those players have an undefined cost value to acquire, because they aren’t freely available in the market. So, now, you’re setting the baseline at something above +0 WAR and some undefined cost above the league minimum. That’s a problem.

by davidcameron on Feb 4, 2009 5:55 AM PST up reply actions

Yeah I totally see your point and WAR has probably the best baseline you can think of

Its just that the word replacement makes me think of the quality of player that will replace another player. Most teams usually have a guy better than that around. The difference though probably is usually 0.5 WAR or less so its not a big deal.

I guess I just need to associated WAR with its definition and not try to stray my thinking from that.

by Edgar for Pres on Feb 4, 2009 8:35 AM PST up reply actions

Would it be easier if, instead of "replacement level", we used "freely available level"?

by Jeff Sullivan on Feb 4, 2009 8:42 AM PST up reply actions

I think I can handle replacement because freely availible is just is too much to say

Plus WAFA sounds like a trade treaty. Something going against WAR though is that the detractors of WAR already have the great chant of “WAR what is it good for, absolutely nothing!”

by Edgar for Pres on Feb 4, 2009 9:16 AM PST up reply actions