Tervor Hoffman v Mariano Rivera using IR

[AztecBill]

I found an interesting site that tries to quantify how relief pitchers fare with inherited runners. I thought Rivera would own Hoffman because I heard how he was pulled into games earlier and how good he was coming into mid-inning games. But the stats on this site showed something different.

Career inherited runners.

Hoffman 346
Rivera 331

I am surprised that Hoffman had more inheried runners than Rivera. I guess my impression that the Yankees brought in Rivera more often to clean up 8th inning messes was wrong.

Here is an explanation of the ratings. It is based on what base the runners on and how many outs there were. And then they adjust the numbers to reflect and adjusted ERA.

Hoffman 3.04
Rivera 4.20

That means Hoffman was significantly better than Rivera at not allowing inherited runners to score.

[AztecBill]

The detail for Heath Bell at the site shows that Heath has had 21 inherited runners over the past two years and only allowed one of them to score. Note: "A" is appearances.

That one runner was a game he came in with runners at 2nd and 3rd with one out.

---

There is only one pitcher in the stats that has a better record with more inherited runners than Hoffman.

Ricardo Rincon - 2.55

[01aztecgrad]

It's an interesting concept, but a horribly designed statistic.

Problems with the stat:

1) It weights failure differently based on the situation (outs/runners/base) but doesn't weight success differently based on the situation.

2) It takes a straight average of the per game calculation to determine the season statistic.

An example of the absurdity of this methodology would be player B being ranked higher than player A in this situation:

Player A
Game 1: Bases Loaded 2 outs - grand slam - 46.6 RRA
Game 2: Bases Loaded 0 outs - no runs - 0.00 RRA
Game 3: Bases Loaded 0 outs - no runs - 0.00 RRA
Game 4: Bases Loaded 0 outs - no runs - 0.00 RRA
Game 5: Bases Loaded 0 outs - no runs - 0.00 RRA

5 Games - RRA = 9.32

Player B:
Game 1: Runner on 1st 0 outs - runner scores - 7.87 RRA
Game 2: Runner on 1st 0 outs - runner scores - 7.87 RRA
Game 3: Runner on 1st 0 outs - runner scores - 7.87 RRA
Game 4: Runner on 1st 0 outs - runner scores - 7.87 RRA
Game 5: Runner on 1st 0 outs - runner scores - 7.87 RRA

5 games - RRA = 7.87

If they want the stat to have any meaning at all, they have account for the varying degree of difficulty in successful situations, and account for the relatively low number of chances when computing the season/lifetime average.

[AztecBill]

May 9, 2011 18:55:13 GMT -8 01aztecgrad said:

It's an interesting concept, but a horribly designed statistic.

Problems with the stat:

1) It weights failure differently based on the situation (outs/runners/base) but doesn't weight success differently based on the situation.

2) It takes a straight average of the per game calculation to determine the season statistic.

An example of the absurdity of this methodology would be player B being ranked higher than player A in this situation:

Player A
Game 1: Bases Loaded 2 outs - grand slam - 46.6 RRA
Game 2: Bases Loaded 0 outs - no runs - 0.00 RRA
Game 3: Bases Loaded 0 outs - no runs - 0.00 RRA
Game 4: Bases Loaded 0 outs - no runs - 0.00 RRA
Game 5: Bases Loaded 0 outs - no runs - 0.00 RRA

5 Games - RRA = 9.32

Player B:
Game 1: Runner on 1st 0 outs - runner scores - 7.87 RRA
Game 2: Runner on 1st 0 outs - runner scores - 7.87 RRA
Game 3: Runner on 1st 0 outs - runner scores - 7.87 RRA
Game 4: Runner on 1st 0 outs - runner scores - 7.87 RRA
Game 5: Runner on 1st 0 outs - runner scores - 7.87 RRA

5 games - RRA = 7.87

If they want the stat to have any meaning at all, they have account for the varying degree of difficulty in successful situations, and account for the relatively low number of chances when computing the season/lifetime average.

I didn't read that deeply. You are right. Their methodology is very bad. They had the numbers of how hard it is to score from the bases, it wouldn't have taken long to find a 3x3 grid showing how hard it is to score for every out/base situation. Then take all out base situations for a year and calculate the expected number of runners that would score and how many actually did. That ratio is a much better rating. Taking an average of game scores is crazy. And you are right their method gives no credit for not allowing a run, other than getting a zero. But you would get a zero for not allowing a run with runners at second and third with no outs (most difficult situation), just the same as a runner at first with two outs. It also doesn't differentiate a runner at third with less than two outs enough. Because a runner can score on an out, those chances are way more likely than a runner at third and two outs.

It is hard to believe that someone would go to that much trouble and get those basic things so wrong. It screams of someone who doesn't know baseball or statistics but is good with computers.