Comments on: Stats Discussion Part I: What’s wrong with Wins and RBI?

By: Todd Boss

Todd Boss — Wed, 06 Feb 2013 14:59:15 +0000

If this is the same WPA that I’ve looked into before, then I have some serious concerns about it on a game-per-game basis. I’ll talk about this in a later post when I look at the “player value” stats.

By: Zac

Zac — Wed, 06 Feb 2013 13:43:49 +0000

Hey guys, just wanted to let you know if you’re interested in that Win Probability Added that Kevin R mentioned, you can find that at fangraphs. They update that every day just the way Todd is interested in.

By: Todd Boss

Todd Boss — Tue, 05 Feb 2013 17:55:55 +0000

Thanks for the correction; I just fixed it in the post.

I like this RBI parable to Sales organizations. I may use that

By: RMR

RMR — Tue, 05 Feb 2013 17:24:07 +0000

Great summary. However, regarding OPS, you have it backward. In terms of the relative run value of a point of OBP to SLG, a point of OBP is worth 1.7 times as much as a point of SLG.

As you note, wOBA gets to this using a linear weights based approach, but you mention that.

Regarding the RBI conversation, I think the best business world comp is with sales. Imagine you have a bunch of sales guys who are fed leads by a sales manager. Looking at RBI is like looking at the total $ in deals closed by the sales rep. Sure, it’s a decent indicator of who did the bets job. But without accounting for who got what deals, of what size and in what circumstances, it’s a rough, biased measure of the a guy’s ability to close business.

Now imagine that those guys were also responsible for generating the leads, which got fed in to the pool of leads, only they couldn’t do the sales for most of the leads they generate.

Now imagine you’re starting a new company and you’re building out your sales staff. How do you pick which guys to hire? By just (or primarily) looking at the total $ of business each one closed? I think not. And which guy gets the award for the best salesman?

By: kevin r

kevin r — Sat, 02 Feb 2013 18:52:43 +0000

Todd, go here http://www.baseball-reference.com/about/wpa.shtml and scroll down to “run expectancy”. I haven’t done much homework on it, but it does seem very useful. It’s also useful for things like “productive outs”, which always sounded like crap, but you can use it to capture things like how much benefit you can derive from trying to go first-to-third, etc. (I’d like to see a stat for that – how much worse a K is than F9, do different hitters move runners further on average than others because they have more deep singles, etc..)

By: kevin r

kevin r — Sat, 02 Feb 2013 18:46:52 +0000

Wally,
The biggest problem with using RBI as a measure of performance is that it’s subject to the same flaw that humans are – we remember the clutch moments. We remember Riggins shrugging off the CB in the Super Bowl, but we don’t remember Joe Washington leaping to get a shoulder on a linebacker which provided the hole. Also, we remember the Werth homer in the 9th which won the game, but without the run in the 3rd, we’d have just tied rather than won. And that’s the problem with RBIs – they only credit the thing we already would have remembered rather than taking into account all the boring stuff that resulted in the win or loss. At least the newer stats attempt to take all the contributions into account.

By: Wally

Wally — Fri, 01 Feb 2013 22:57:06 +0000

Todd – good discussion.

My view on the old school stats, and why I have personally slid towards the new school ones, is that, even with the valid flaws you mention, they are ok as one way to see if someone was good, but they were too often used as the only way to show it so they should be less emphasized. Take two of the most criticized ones – Wins and RBIs. If someone wins 20 games, or drives in 125 RBIs, the odds are pretty highly correlated that they will look good under the other metrics too. Not necessarily, but still pretty good. But many, many players who didn’t hit those numbers will also have had good seasons, so you need more ways to think about it.

As for players loving RBIs, I don’t think you can state strongly enough the motivation that you got from Clip – pay. Players will strongly gravitate towards what earns the most money. It is human nature, and also the biggest grade that they see. If walks became the highest paid metric in the game, I feel pretty strongly that you’d see players going hard for that. Is it really any different for any of us in our jobs that have some amount of incentive bonuses? People generally figure out quickly what kind of work gets paid the good bonuses, and gravitate towards that. At least that is what I have seen.

The other reason i think that people (fans, writers) like RBIs is the belief in clutch and choking. RBIs become a shorthand for players that rise to the moment. If you believe that some do and some don’t, you’d be inclined to place a high value on RBIs. Even if other guys didn’t get as many chances, the best that you can say is that maybe they would have delivered. But you know for sure that these guys did. I do not subscribe to this theory, but I think some others do.

By: Todd Boss

Todd Boss — Fri, 01 Feb 2013 21:49:40 +0000

Very interesting. You could still keep running totals in the current year just by basing the runs expectancy on the prior year's numbers, so that we could have "leader boards." Who wants a stat that is only good once the season is over? But yeah I like the concept. Did they actually go through with this? I did some digging at BP and couldn't find it. But I did find the one and only Nate Silver introducing a "Context Independent RBI" in 2004. You dont' happen to have a link to this "run probability" RBI stat do you?

By: Todd Boss

Todd Boss — Fri, 01 Feb 2013 21:40:27 +0000

Hey, I warned you it was long in the beginning The others in the series aren’t nearly as long right now.

By: peric

peric — Fri, 01 Feb 2013 20:19:21 +0000

Not bad. Very good summary … albeit long but length seems necessary given the number of stats reviewed.