Runs Created -- coincidence?

[Ytown Tribe fan]

What if the Runs Created formula is based on a mere coincidence?

It almost seems too easy and convenient: take total bases and multiply by OBA (with a few tweaks here and there) and you get a team's runs scored in a set period. Anyway you try it, it seems to work, that is beyond doubt. And that formula has been used to determine an individual player's offensive contribution to the team many times - in arbitration, in player analysis, in rating the best players, even in MVP voting.

But when you watch a baseball game, every run scored MUST be scored within a single inning -- between three outs. There is no average or set number of bases each inning that puts a run on the scoreboard. A leadoff double can lay stranded; a runner on third with no outs may score on a batter's GiDP; three hits strung together may result in no runs; runs may result from no hits -- the combinations are limitless.

Bill James discovered the basic formula two decades ago, and has refined it many times to include "clutch" hitting and other factors, all to make the formula fit the actual number of runs scored. But isn't it possible that the formula is merely coincidental? Does it really "even out the odds" in determining a player's worth over the course of a season or a lifetime, as opposed to raw RBI and Runs totals?

[Duque]

Do MVP voters really use RC?

[JamesI]

Originally posted by Duque
Do MVP voters really use RC?

]

I'd be surprised if they did. When the "experts" talk about mvp voting, the most important stat to them always seems to be r.b.i.

On topic I don't know if the formula is a coincidence. I'd guess it probably isn't.

[VNV Nation]

I think you have a point on some level, because offensive events are not credited with fractions of runs, i.e., when you get a single in the first inning you don't go ahead .34-0. You have a point about three outs in an inning, too, the value of outs are situation-dependent just like everything else. Since there are many different formulas that predict the number of runs scored, and they use vastly different methods, we probably will never know which is closer to accurately describing the alchemy by which runs are scored.

[sweaver]

I'm not sure how getting on base and amassing total bases being related to run scoring could be a coincidence. That's kind of like rainfall and soil moisture content being related.

[SmedIndy]

THE GRAVY SOAKED TIE and his ilk wouldn't know runs created if it was featured in their pre-game buffet.

It's not coincidence, it's good analysis. You find the goal (scoring runs) and try to break down scoring runs into its components. The first stab at runs created was finding the protons, neutrons, and electrons. The refinements have all of the other neat particles that get Sweavers undies in a bunch!

[sweaver]

Quarks! I love quarks!

[Skip]

Originally posted by SmedIndy
THE GRAVY SOAKED TIE and his ilk wouldn't know runs created if it was featured in their pre-game buffet.

Ok Smed, once again I will play your comic foil ... who is THE GRAVY SOAKED TIE - anyone in particular, or just the generic mediot.

And P.S. ... based on recent public appearances, Petey is increasingly hard to refer to as PRINCE VALIANT; he looks more like a "before" ad for the hair cutting/replacement/styling/controlling place of your choosing.

[VNV Nation]

Conlin

[KCBOOMER]

Isn't the right phrase "correlated"??

Anyway I don't think it is a coincidence. OBP means how often do you get on base and TB means what are the quality of those on base efforts. Seems pretty closely tied to run production to me. Certainly there are no partial runs being scored but as the sample size gets large this smooths out.

[SmedIndy]

Originally posted by Skip
Ok Smed, once again I will play your comic foil ... who is THE GRAVY SOAKED TIE - anyone in particular, or just the generic mediot.

And P.S. ... based on recent public appearances, Petey is increasingly hard to refer to as PRINCE VALIANT; he looks more like a "before" ad for the hair cutting/replacement/styling/controlling place of your choosing.

Bill Conlin is the GRAVY SOAKED TIE.

Petey will always be PRINCE VALIANT. But he could create runs.

[Ytown Tribe fan]

sweaver, you bring up a good point I was going to use. I'm a weather forecaster in real life (no, not the kind you see on TV with a bowtie and a bad combover). Every day I deal with numbers that appear to say one thing but mean another.

Example: so far this month, at the weather station where I work, we've received 1.50" of rainfall. This is perfectly normal for July. June rainfall was 0.50" above normal, and May rainfall was 2.30" above normal. We were above normal in the three months before that as well. Yet, the reservoirs are low and the grass is brown. Why?

Because we got almost ALL of our July rainfall (1.20") in a three-hour period on the 9th. The rest (.30") came three days later in one thunderstorm. It has been perfectly hot and dry every other day of the month. Nearly all of the rainfall since the last week of June simply ran off into the rivers and streams and is long gone. The raw numbers say we should be green as can be with full reservoirs. The reality is that the crops are suffering.

The problem I have with Runs Created formula is that no one can say exactly why it works. Bill James cannot say why a double today, combined with one fourth of an intentional walk last week, divided by three plate appearances in April, adding a home run in August, 62/100ths of a stolen base in September, and everything else from every game in between, should combine in that simple manner to equal the total runs scored by a team in a season. It is not enough, to me, that it works. There has to be a compelling reason, other than that "it seems to make sense and, what the hay -- it works, doesn't it?"

The other problem is that he "tweaks" the formula all the time to get it to equal the number of actual runs scored. This "fudging" is fuzzy math any way you look at it.

There has to be a better way, other than to simply add up runs and RBI (not that there's anything wrong with that, and runs should count equally if that's the case). I'm convinced that runs and RBI are about 80-90% of a players offensive contribution, and that the other 10-20% can be measured in another way.

That way may be to simply look at every single inning of every game in which a team has scored a run and give every player in that inning some credit if he A) didn't make an out or B) made an out that resulted in advancing a runner or driving in a run without ending the inning. Project Scoresheet could easily do that. You could take the number of runs scored in that inning and divide that by the number of batters who contributed in the above manner and each such batter would get a "Run Share" for that inning. That number would not replace runs or RBI, but would augment it.

Run Shares would be a separate category, just as Wins, Holds and Saves are for pitchers. Each player would have runs scored, RBI and run shares, and each category would be considered equally in determining a players offensive contribution to his team. The advantage over Runs Created is that each Run Share is exactly that - a share of a run (or runs) that a team has created in a given inning.

Different players could have different "profiles" (that is, different ratios of runs to RBI to run shares). A player who had a low run or RBI total but a high number of run shares would get credit for doing the things necessary in an inning to put runs on the board. A player with high RBI totals and low run and run shares totals would be somewhat exposed -- I'd guess it would be the all-or-nothing type of hitter, the Dave Kingman type whose RBI totals inflate the perception of his actual contribution. A Rickey Henderson-type player would have high runs and run shares totals with low RBI numbers. Other players would have different spikes or no spikes at all. Some players with high batting averages might have low run/RBI totals and high run shares, others might not - it could be a measure of true "clutch" hitting.

I'm not saying that Runs Created doesn't already do those things to some extent, just that run shares would not be based on seasonal averages but on actual innings where runs were scored. It would be work, but it could be worth it.

[VNV Nation]

[i]
The other problem is that he "tweaks" the formula all the time to get it to equal the number of actual runs scored. This "fudging" is fuzzy math any way you look at it.
[/b]

This is an inevitable change in the purpose of the runs created formula. Twenty years ago, it was very much an open question of whether or not something like runs created would work. It has become generally accepted that you can predict the number of runs a team will score based on its number of hits, walks, etc.
Instead of predicting the number of runs a team will score, the formula is now trying to determine more accurately how many runs each individual player has contributed.

[sweaver]

So Y'town, you ARE a weather forecaster, but you DON'T play one on TV?

The problem with the idea of run shares is the same problem I have with win shares: we are tying predictive measures to actual results, and as a result "fixing" our predictions to make them accurate after the fact.

I always liked runs created, in the original form, because they showed a player's performance removed from the context of his team. Good teammates mean more runs and RBI, bad teammates mean less, regardless of how the player himself does.

I am afraid with all these "shares" measures, we are examining an elephant under a microscope, and losing the big picture. To extend the precipitation metaphor, the type of precipitation matters. A slow, steady rain will saturate the soil much better than a quick thunderstorm, even if the rainfall amounts are the same. And that can make a telling difference in the water table.

But, a home run is a home run, no matter how long it is.

[Ytown Tribe fan]

Run Shares isn't intended to be a predictive measurement but an additional stat on a par with runs and RBI. It would be based on every inning in which a team actually scored runs, and would be awarded to every player that was a part of the scoring, whether they got an RBI or scored a run in that inning or not, so long as they did something positive to help the team put runs on the board. If a player's actions didn't result in a run eventually being scored, they don't get a Run Share, even if they get a hit or steal a base.

Example inning: Lonnie leads off with a walk and steals second base. Willie hits a sharp single, Lonnie advancing to third base. George grounds into a double-play, Lonnie scoring and Willie out at second. Steve hits a double. Frank hits a single, Steve holding at third. Darryl hits a sharp grounder and is thrown out at first. Inning over.

The result of all that effort is one run for the 1985 Royals. Lonnie gets one run scored, one walk, one stolen base; Willie gets a single; George gets one RBI and one GiDP; Steve gets one double; Frank gets one single; Darryl (Motley) gets a groundout.

Each of those stats counts toward the players season totals, and towards Runs Created , OBP, SLG, BA and everything else that we count. But only three players were involved in the scoring. In that inning, Lonnie, Willie and George would each get 1/3rd of a Run Share. Steve, Frank and Darryl would get no Run Shares.

Is it fair? Irrelevant. Baseball isn't fair. It doesn't matter one whit to the Baseball Gods HOW you win a game, or how you score (or prevent) a run. A player can hit a triple and not score and not get a Run Share; a player can ground into a double-play and drive in a run and get a Run Share. That's real baseball and that's real life.

I believe that Run Shares, when counted along with Runs and RBI, helps to fill out the picture of a player's offensive contributions based on what his team actually did, not on what he or his team did "on average". And I'm not even plugging an upcoming book.

BTW - I'll offer a reward to any person who can demonstrate WHY the Pythagorean W-L formula works. But that belongs in the "Win Shares" thread.