Evaluating SF's Pitchers by RE24 Average

With relief pitchers, it’s easy enough to cite – it’s just not a readily available statistic. You more or less have to go….

[(IP x .426) – RE24]

This gets you an RE24 by R9, which you can compare to earned runs

To compare to ERA you then multiply that figure by .92, divide by innings pitched, and multiply by 9.

…and just like that you have a much better picture of a relief pitcher’s true performance. It’s a bit like comparing RE24 to wRAA to understand situation hitting. Kind of.

In terms of what RE24 does, I wasn’t crystal clear.

It assigns run expectancy at the time of the play – so the starter gets his credit and the reliever gets theirs. Ultimately, starters get to subtract a poor relief performance using this measure.

In addition, for reasons that you can Google elsewhere, ERA is a stat that tends to favor ground-ball pitchers, so they get an unfair boost.

Finally, I like dividing the RE24 by leverage because it creates a neutral baseline for comparison.

It’s still a useful stat for starters but it doesn’t replace ERA. For relievers, it’s vastly superior.

In order to really grasp RE24, one first has to understand what run expectancy is and what the 24 base/out states are. I’d recommend reading the opening pages of The Book at Amazon using the Look Inside feature.

Essentially, a base/out state describes what is happening on the diamond with the batter in the box. No outs and no one on is the first base-out state. Three outs is the 25th but we don’t use it. In the middle you have things like two outs with runners on first and third. In total, there are 24 possible base/out states.

Run expectancy is derived by looking at the number of runs that a team scores in an average game over a five year time span and dividing it by 9. This number is accepted to be at .555 runs per inning. Therefore, when the first batter comes up in an inning, the run expectancy is .555 for that base-out state.

Let’s say the first batter gets to first. We have now transitioned to the first base/out state: a runner at first with no outs. Using a similar calculation, we look at the 5 year average for run expectancy in this base-out state. As it turns out, that number is .953. In making this transition, the rune expectancy changed by .398. It doesn’t matter how the batter reached first, just that he did.

RE24, using similar math, assigns a value to each one of the events that can occur in a game during a base/out state based on how likely it is to produce a run. The difference in RE during each transition is charged to the hitter and the pitcher. When taken as a whole, this gives us a player’s RE24. For batters, RE24 indicates the likelihood that they will produce runs with their at bats given their run expectancy across all 24 states. For a pitcher, it shows the likelihood that they will prevent or allow those runs to score.

A caveat here – be careful with Fangraphs because it shows a pitcher’s offensive RE24 in the player page and does not show the pitching calculation. You have to use baseball-reference.com for that.

So what we have done with the calculation above is apply a little math to get an RE24 average the same way we would turn earned runs into an ERA. This is why the first easily understood conversion takes us to RE24 expressed as an R/9, which compares to earned runs as R/9.

A quick note on leverage: Leverage looks at what high pressure situations do to players so dividing a player’s RE24 by his base/out leverage (boLI) levels the playing field and lets us look at players as though they all faced the same 50% pressure in every at bat. The resulting stat here is bo/LI – which I suppose could be applied to ERA as well, though I haven’t seen it done.

For pitchers, the resulting RE24/boLI tells us how often they allow batters on base and how often they pitch out of that situation. For batters, it tells us what the likelihood is that they will drive in a run. As I alluded to earlier, comparing RE24 to wRAA gives us a player’s situational hitting numbers. Aubrey Huff in 2011 was pretty decent when he was at the plate solo but fell apart with runners on.

As a statistic, RE24 is a holistic look at how likely a player is to contribute to runs during a game across all the 24 base/out states instead of just the ones that we like to measure. Even if a player reaches on an error, a pitcher like Cain or Romo has the ability to get that runner out, as seen by the RE24 average being lower than the earned run average. Without RE24, you wouldn’t know that.

So while the RE24 average is an especially well pronounced metric for relief pitchers due to the fact they inherit base runners, it also works well to look at a starting pitcher’s performance across all situations over a full 5-7 innings of work. In this sense, it is a nice compliment to FIP, which seeks to take fielding out of the equation completely.

One final note – RE24 can also be expressed as REW, or a player’s contribution to a team’s wins over the 24 base/out states. This has some interesting comparisons to WPA but I’ll let you Google that one for yourself.

I’m glad people enjoyed this.

I’ll re-build my situation hitting analysis when I figure out why Excel barfed on all my averages last week. Gosh, that was embarrassing.

Either way, RE24 is a very useful statistic and in a lot of ways the least artificial because it is based on sample sizes of around 5000.

It goes well with “ERA” as well

Will Brian Wilson’s ego allow him to start behind Romo?

I think the bigger questions are: Will Botchey adjest he views of Wilson has his skill mature/decline?
And do we realy want him to move Romo out of the higher leverage situations that come beofre the 9th?

To the last I can say I rather the best reliever be ready to deploy any time after the 5th ( say 11 outs or less left in a game) when getting out of control. This is even more critical in the squad that can’t score many runs with an annoying frequency. I am not saying turn a team willing to over pay ( which I doubt exists in this case) but look what Sabean usually buys in position player help $6mm -$8mm (belch obp, not much iso) then look at the ability to reliably count on getting the last 6 outs with giving up a run in 2 games of the 3 game series and that is pretty reasonible return on cost.

Better seen than not at all. I’ve actually written a couple of the big sites and suggested RE24 AVG (I’d call it R2A) as a stand alone statistic. It’s a much more sensible way to evaluate the bullpen and it’s got a lot of value for starters too.