Saving Bonds

Dave Paisley

As many of our regular readers know, here at Strikethree.com we try to walk that fine line between the under-analysis that's typical of most mainstream sports journalism, and over-analysis, which is a common problem with many stat geeks. Last season provided a classic example of the debate in Barry Bonds' historic year.

On the one hand, the traditionalists pointed to Bonds' lower RBI total than Sammy Sosa, and could often be found comparing the two, with Bonds coming out unfavorably. On the other hand, we had the stat geeks falling all over themselves in describing Bonds' season as the best ever, or at least one of the best three seasons ever. On the other other hand (maybe it's that nuclear plant I live next to) I believe the truth is somewhere in between, and that there is a way to bridge the gap between the camps.

Last season, Bonds' on-base percentage was phenomenal (.515) and his slugging average stratospheric (.863). But, say his detractors, so what if Bonds led in OPS (adding OBP and SLG) by over 200 points? They pointed to the fact that he only drove in 137 runs compared to Sosa's 160, and scored only 129 to Sosa's 146. Bonds worse than Sosa? Why, that's unpossible!

The answer is simple — the contexts within which they played were very different. Rich Aurilia's power surge robbed Bonds of RBI opportunities by clearting the bases ahead of him, and a merely excellent year for Jeff Kent didn't score Bonds as often as he might have.

Is it possible, though, to bridge the gap from traditional stats like RBI and Runs to OPS and Runs Created?

First, it is necessary to somehow convert R and RBI to a meaningful combined number. I've done this in past years. There is an awful stat called Runs Produced that is calculated by adding R and RBI and subtracting home runs. Why subtract home runs? So you don't count them twice, says the logic. However, that logic misses the point that there are two facets to scoring — crossing the plate, and driving in the run. The home run does both parts, and consequently deserves to be counted twice.

Imagine, if you will, two guys who each score and drive in 100 runs. Guy #1 is a John Olerud type who somehow never hits a home run, but gets on base a lot and drives other guys home with singles and doubles. Guy #2 is a Mark McGwire type who has all kinds of power, but for some reason never comes up with guys on base. He does, however, hit 100 solo home runs. According to Runs Produced, the first guys gets 200 RP, while the second gets only half that. And yet they were each involved in their team scoring 100 runs.

The way to fix this is to simply add R and RBI and divide by two. In the past I've called this Runs Generated, but apparently it was introduced in 1989 as Runs Tallied or RT. It was invented by Barry Codell who introduced the stat in Baseball Research Journal #18 (1989). Hands up anyone who knew that. Hmm, thought not. Anyway, as it's not exactly a patented formula, I shall continue to call it Runs Generated and move on from there.

Next we have to find a way to normalize it to some kind of rate. You might think that dividing it by plate appearances or at bats would be a good idea. Well, that's not bad, but they aren't the limiting factors. The only commodity of which a team has a fixed quantity is outs. You get 27 per game, and making the most of those is what is critical. Barry Bonds generated 133 runs to Sosa's 153. He only used 320 outs to do it, though, while Sammy used 388.

While this may just look like games with numbers, let me relate this to OPS and show the relationship graphically. Here's a chart of Runs Generated per Out plotted against OPS for all batters with at least 250 AB in 2001.

First, let me point out that the correlation is phenomenally good, so I have apparently been able to tie traditional stats (R, RBI, Outs (i.e. AB-H)) to less situation-dependent stats (OPS) very strongly. If I could be bothered, I could calculate Runs Created per 27 Outs, but I don't believe the results would change much.

There are only a handful of players who don't fall within a tight band of the mean line (red), and oddly enough, Bonds and Sosa are two of them. I believe that major deviations away form the mean line are either luck (good = above the line, bad = below), or are due to systemic factors beyond the individual player's control. Those factors include batting order position and the types of player who bat before and after. It's evident that Bonds had an unlucky year, or that circumstances just worked against him so far as turning raw offense into runs was concerned. Sosa, on the other hand, got a tidy little bonus from circumstances.

So Bonds MVP was truly deserved, and I can use whatever kind of stat you want to prove it.

Other players of note from the chart:

Todd Helton and Larry Walker both benefit from Coors, but they still lie close to the line. Correcting for park factor would move them down the line, but they're still both pretty good.

Chipper Jones, Jason Giambi and Luis Gonzalez had great years.

Juan Gonzalez managed to get more than his fair share of RBIs. Again.

Bret Boone's production was, predictably, more than could have been expected from his OPS. Mix together a drop in OPS and a return to more normal luck, and his R and RBI will probably drop off quite a bit in 2002. Much as I hate to believe it, the same fate may befall Mike Cameron.

Rondell White should look a lot better in NY, with better luck and a better supporting cast.

Benny Agbayani was really robbed in the luck department. Coors should help him out quite a bit.

Pat Meares, Paul Bergeron and Jack Wilson all suck. Shame the Pirates owned two of them.

Jolbert Cabrera was darn lucky in 2001.

As the Dodgers discovered, Alex Cora (lowest man on the RG/Out totem) is not the answer at short.

As bad as he is, Tom Goodwin is overrated.

about the author

The way today's stock market is going, can you afford to be without Savings Bonds? Get Dave Paisley's financial advice at drdjp@strikethree.com. Until he recommends that hot Enron stock, that is.