I'm with you on Blyleven verus Morris, and FWIW, I'd vote for Blyleven if I could.

But it is frustrating how his advocates just define away the biggest argument against him. A .534 winning percentage is ... not a HOF winning percentage. Yes, yes, his advocates are convinced that that was just bad luck on Blyleven's case. But it seems to me that I'd like that to be proven, not just asserted. His teams were not, by and large, horrible offensively. They were average. With average offensive support, a pitcher with an ERA+ of 118 should have a much higher winning percentage than .534.

Not precise enough numbers for you? Well, using the pythagorean theorem, Minn's actual offensive performance, Blyleven's RA, and assuming that Minn supported him to the same extent that they supported the rest of the team's pitchers, he should have been 113-67, not 95-85, over his first six years. I haven't figured the rest of his career yet.

So did his teams on the whole support him less than the other pitchers on the team? Over the course of his career, I find that unlikely. At least I'd like to see some evidence to that effect.

I'd like to see more of the work done, too (I'm completely incompetent to do it myself), but more as a curiosity. I guess I would disagree with you as to how "big" of an argument this is. It's certainly the most often-cited one against Bert, but I don't see it as having any force at all, and I don't think it particularly matters exactly why his winning percentage is lower than you'd expect.

As I see it, a pitcher's job is not to "win games," since no pitcher can actually win a game by himself*, but to pitch as well as he can and hope that his offense scores enough runs to win (and that his bullpen doesn't blow it). He has no control over anything but how well he does at preventing the other team from scoring runs. Thus, if pitcher A prevents runs more effectively than pitcher B, pitcher A is a better pitcher, regardless of their won-loss records.

* Well, I saw Pete Harnisch win a game once. He pitched a shutout and hit a home run. But you know what I mean.

So to me, it's perfectly legitimate to "define away" any argument against Blyleven that is based on his winning percentage. Bert did his job of preventing runs better (considering his career length etc.) than all but a handful of all those who have ever pitched in the major leagues, and thus deserves to be in the Hall of Fame. End or argument. If you (not you personally, Larry, just a general you) want to argue that he's somehow less great based on the number of times his offense happened to score more runs than he gave up, I think the burden is pretty squarely on you, since you're arguing against logic. Analysis of the reason for his low-ish winning percentage would be interesting, but it's hard to see how it could have any impact at all on his Hall of Fame case one way or the other.

Thanks for the comment!

Well, no. You're proving my point. You're making assumptions about the causes of his relatively poor winning percentages - assumptions based upon accurate perceptions about pitchers in general, but assumptions which may or may not obtain in this particular case. Or, to put it another way, ignoring win/loss records entirely is (almost) as blinkered as relying on them as the primary determinant of pitcher success. Obviously if it can be shown that his teams gave him poor run support over the course of his career, you would be 100% correct. But that hasn't been demonstrated, and the seasonal run data of his teams (at least to the extent that I've looked at it so far) doesn't support such a conclusion.

A much as "pitching to the score" and "clutch" ability are generally a myth, it is entirely possible that they come into play in some (rare) cases. If Blyleven did indeed have the level of run support that one would expect from his teams' seasonal run production, then he was losing more games than he should have. You can't put ALL of the blame on the team. Especially given that there is a pattern throughout his career (more on that below). It's consistent, from year to year. I'm not a big fan of the "it's a pitcher's job to win games" argument for the very reasons you cite. But if a team is scoring an average number of runs for a pitcher (if they were ... as we both agree, more research is needed) and that pitcher is winning only half those games ... over a course of years ... sorry, that's an average pitcher (which of course is valuable in itself, but we are talking about the all time greats.

I mentioned the first 6 years of his career. His teams scored about 4.15 runs per nine innings; Blyleven allowed about 3.2 runs per hine innings. And he was barely over .500. There were other years like that - for example:

1977 - 14-12, ERA+ 151, team scored 767 runs ... with a league average of 732.

1985 - 17-16, ERA+ 143. Played for two teams, one of which was right about league average in runs scored, the other of which was a little (but not much) belwow average.

There were plenty other years like those. Sorry, that demands at least further research. He was either the unluckiest pitcher in the history of the game (possible) or ... something else was going on, that detracts (somewhat) from his HOF argument (though as I said I'd still vote for him).

Maybe there was something unusual about his performance profile ... I could speculate, but I won't*. The point is that there is a red flag there (not present as far as I can see with any other 200 plus game winner) - and it should be studied, not ignored.

*Here's the kind of profile I mean. In a 4 run per game environment, which pitcher is going to win more games:

A: allows 2, 3, 10, 3, 1, 4, 5, 2, 3, 7 (4.0 RA/G)
B: allows 0, 5, 7, 6, 1, 5, 6, 1, 4, 5 (4.0 RA/G)

Same RA/G, both 'average" pitchers, but the first guy allowed 4 runs or less 7 times, the second guy 4 times. I'll take pitcher A. That's an extreme example, and of course this kind of thing evens out more than 99% of thetime. Maybe Blyleven was the rare instance when it didn't.

Hmm, I hadn't realized that baseball-reference.com had game logs going back that far. Plenty of data available; I don't have the time to study it at length, alas. But just picking one year at semi-random, 1973. 20-17 despite an ERA+ of 158 and decent (roughly league average) run support. Eyeballing the game log ... on the one hand, he did have a number of tough losses (7 losses where he allowed 3 runs or less. On the other hand, that still means 9 losses allowing 4 runs or more, and his pattern of runs allowed was somewhat like pitcher B in my previous comment, albeit at a higher level.

Looking at the Cy Young voting that year (and yes he should have been higher than 7th)... Sure, most of the people above him had more run support than he did - but a couple of those guys kind of leap out - Ryan had a slightly better record despite about a half a run LESS run support per game. And Palmer had more run support. but not enough to account entirely for his better record.

Just a little more data - not random, but still interesting ...

The 7 years I've looked at so far - through 77, excluding 76 when he played for 2 teams - his support was about the same as his teams's overall run production.

1977 was really pretty odd - 14-12 despite allowing just over 3 runs per game while the team scored almost 5 runs per game for him - a few tough losses, but also 9 losses where he allowed 4 or more runs. Not a BAD performance by any means,but not what you would expect from a 151 ERA+ either. Jim Palmer, 2nd in the Cy Young race, was 20-11 with a comperable (but slightly lower) ERA+ and MUCH LESS run support.

Larry, this is interesting stuff, but I still don't see how it's a "red flag" or "demands further research." What is the big troubling possibility here? That he kind of reverse-pitched to the score: was great in blowouts and let up in close games, or something? I don't think that's a serious argument, and I guess I just don't get what this would prove.

If you want to label it as me being "blinkered," I guess that's OK, but: no matter how it happens, I just don't think it makes any sense to attach any meaning to a pitcher's win/loss record when we already know how good he was at the one and only thing he could more or less control -- preventing runs. I think the sense in that should be self-evident, and that the burden of proof is on whoever wants to argue against it. What if it's the offense, and not the pitcher, that is inconsistent? If he gets average run support for the year because they scored 20 for him once and 2 or 3 a game the rest of the year, that's not going to help his W/L record much.

And you have to go beyond just the "tough losses." In 1973, I see eight games where he pitched 8 or 9 innings (OK, I'm counting one where he went 7.2), gave up 2 or 3 runs, and lost. Those are a killer -- you'd expect maybe six wins, a loss and a no-decision from that. But beyond that, in any game in which you can reasonably see a pitcher either winning or losing, he lost. Four runs in 7 IP, 4 in 6 1/3, 4 in 6 2/3, 4 in 8. You'd expect him to win at least one or two of those, but he lost them all. Looking at his wins, he didn't get a single cheap win: the very worst he pitched in a win in 1973 was 3 runs in 6 2/3, and we'd expect a pitcher to win at least half the time he did that, probably more. Bert could easily have won 28 or 30 games with average luck in 1973, run support be damned.

So anyway. I'm almost tempted to look into this more, but I think I need a little more understanding of why it matters. I still say that if he was among the best at preventing runs, he was among the best pitchers, and I don't see a reason it shouldn't end there. But maybe you can help me understand.

Also, I think the extent to which he did win less than we'd expect is overstated a bit. I wish I knew an efficient way to figure this out for myself, but I read somewhere that Bert's teams played .495 ball in games when Bert was not on the mound, about 40 points below Blyleven's own winning percentage. In contrast, Nolan Ryan had a .526 WP and his teams without him were about .503 (or, again, so I've read).

I did look at some relatively easy to figure, mostly single-team pitchers a while back, and seem to remember that Bob Gibson and Warren Spahn's winning percentages were about 50 points above their own teams' without them (they just pitched for much better teams). So while I don't think it would matter anyway, I'm also not totally convinced that Bert actually won that many fewer games than he should have given the teams he was on.

"That he kind of reverse-pitched to the score: was great in blowouts and let up in close games, or something? I don't think that's a serious argument, and I guess I just don't get what this would prove."

Why isn't it a serious argument? It's true or not true. If it's true, it means that the team lost more games with him pitching than they would have lost with a "normal" pitcher of his ability (as measured by runs allowed). Winning is after all the bottom line. We discount it for starting pitchers most of the time for the reasons you cite - because MOST of the time runs allowed (or other metrics) are a much better gauge of pitcher's contributions to winning. But if we can show that in THIS CASE Blyleven's runs allowed don't accurately reflect his contribution to his team's wins and losses, why wouldn't that be relevant?

I think we're so conditioned to the generally true proposition that pitcher winning percentage is not a good gauge of a pitcher's contributions to team wins that we forget sometimes that ultimately what we are still trying to measure when we evaluate players is ... contributions to their team winning.

I analyze 1973 a little differently than you do - again, contrast his year with Ryan's, who managed a slightly better record with much worse run support - but I'm not saying that he had a lousy year by any means, just that he didn't have quite a good a year as you would think just looking at his runs allowed. But for 1973 you have something of a point, he did have a bunch of tough losses. But what about 1977? Sorry, his 14-12 record that year was a better indication of how well he pitched than his ERA+. He got plenty of support, he just coundn't convert the support to wins.

" I still say that if he was among the best at preventing runs, he was among the best pitchers, and I don't see a reason it shouldn't end there."

Well, if all you are interested in knowing is that he was "among the best pitchers" then sure. But if you want to know where he ranks among the best pitchers, then I think this stuff is very important. Put it another way: his HOF argument is compelling enough that none of the stuff I'm raising is going to undermine it. But if we want to know whether he is a close to borderline (but in) HOFer (my take), or an inner circle HOFer (arguably so if we ignore winning percentage and award voting), then I think the stuff I'm raising is very interesting & valuable.

The context here also is that, over a very long career, he appeared to (that's where we need some more research, to see if the appearance was real)consistantly "underperformed" his win expectation (considering support and runs allowed. Over hundreds of games for several different teams, that seems ... unlikely ... to have been the sole fault of his teams. Over the coruse of a season you can put it down to luck or poor distribution of supporting runs. But over a career?

Really, though, I guess I need to try to finish the study I started (and re-do the part already done to consider actual run support rather than predicted support) and see how much of a discrepancy there really is. If it's, say, 10 or 15 wins over his career, that can be put down to bad luck. But if it's more like 30 wins - well, that's something that we should know when evaluating him.

Comparing win % with him as starter versus win % with others as starters is IMO not the best way to look at it - it doesn't control for the quality of the other pitchers on the team.

The reason it's not a serious argument is that there's just no reason to believe that that would ever happen, as a systematic thing. At least with "pitching to the score," there's a kind of logic behind it -- you can see why a kind of dimwitted pitcher might want to save his best stuff for tight spots and float a few meatballs up there if he's up by five or six. But the other way around? There's no reason to believe Bert folded in pressure situations or anything like that. He did well in postseason games, threw a no-hitter, had all those 1-0 games to deal with. Attributing this tendency (assuming it exists) to anything other than dumb luck seems like quite a stretch to me, and I'd have to see some pretty strong evidence that Bert had something to do with it.

I guess where we disagree is that, yeah, I do think that it's entirely possible that any discrepancy between his W/L and "expected" W/L is just bad luck, even over his entire career. Because you're not talking about 22 seasons and 695 games, we're talking about a handful of wins on the very margins of that much larger sample. If he'd gone 27-10 instead of 20-17 in 1973, for instance, which I think I showed earlier was entirely possible, he's got 294 and a .547 percentage. Give him a couple wins in 1972 for when he threw 10 shutout innings and got a no-decision and then threw 10.1 one-run innings and took the loss, another one for ten shutout innings and a ND in 1971...a few more here and there, and he's right where it seems like he should be. We're not talking about a huge sample size at all. It's absolutely possible that Bert was victimized by some bad luck in his run support distribution (and in fact, I can't think of any other plausible explanation for the "underperformance," assuming again that it exists at all).

I’ve looked more closely at the data. Conclusions:

(1)As expected, the “poor offensive support” for Blyleven meme is mostly untrue – his teams scored 4.25 R/G for him versus a league average of 4.31. That tiny disparity would likely have disappeared if I had adjusted for Blyleven’s varying workload.* Throughout his career, Blyleven had average run support. That said, while I didn’t study it, my sense is that most of the all time starting pitchers enjoyed somewhat better than average run support.

(2)Using Blyleven’s actual RA data, his actual run support, and the Pythagorean theorem, I calculate that Blyleven had an “expected” win total of 300 for his career, as opposed to an actual win total of 287. Interestingly, that “deficiency” (if it is that) is concentrated entirely in his early career – he is 19 wins below expectation for his first 8 years, and a little above his "expected" win level for the rest of his career.

(3)What does #2 mean? Well - the 13 win difference is lower than I expected (because I started with the early career data where the “deficiency” is much more pronounced). Is it statistically significant? I didn’t run a test; eyeballing it - maybe. It’s borderline. It could be mere chance. If his early career pattern persisted, I’d say there would be STRONG evidence of something non-random going on. But the early pattern did not persist. That said, there's a huge difference between 300-233, versus 287-250.

(4)I still think further study is needed. Unlike you, I can think of several mechanisms for the data to be non-random. I don't want to speculate further, however, without more evidence.

In any event, he should be in. I thought so before the study, and I think so now.

*I simply averaged the yearly data together; I did not account for varying number of games/innings on Blyleven’s part. If I did adjust for that, the tiny disparity between his support and the league average would likely have disappeared, as the seasons when he suffered below average support tended to be the years when he pitched less.

I will add this: I tend myself to be one of the stat nerds. But what bugs me sometimes is that some of them (not you necesarily, but I think it effects the "conventional wisdom" have an overly strong level of certainty on some issues which I don't think is justified by the evidence, and which IMO sometimes interferes with good critical analysis.

The complete dismissal of the relevance of win/loss data for pitchers is IMO one of those blind spots - though I agree that most of the time (probably the vast majority of the time) other metrics tend to do a much better job of pitcher evaluation.

Without speculating on whether any of them are true, what are some of those "mechanisms" you can think of? I'm just very curious, because I really don't believe there are any.

I agree that statheads in general get too certain of things. That really bit them (/us) with the defense thing -- ten years ago, a lot of statheads would have told you that a team with nine Jason Giambis would destroy your average 100-win team with a real shortstop and second baseman because defense didn't matter that much.

But one thing I am pretty sure of is that pitchers can't control how many runs their teams score or save their best pitching for certain score differentials or anything like that. I assume you'll agree with me that if a pitcher gives up a single run in nine innings, the quality of his performance isn't affected in any way by whether his offense scored eight runs or zero. I can't think of any reason the same principle shouldn't apply across 4000 innings.

Well, yeah, giving up one run is always a (very) good performance. But Blyleven won most of those games anyway; those are not the games we are concerned about.

I'll give you a couple hypos - not the only ones I can think of.

(1) Maybe a certain pitcher's "stuff" or makeup is such that they avoid getting rocked very often. A good thing, you say? Sure - but ERA+ will "credit" that, and maybe OVER "credit" it. Short version of the argument: giving up 10 runs is not signficantly worse than giving up 6 runs (in terms of win probability). Either way, you lose most of the time. But the guy who gives up 10 runs every now and then is going to look a lot worse using ERA+ than the guy giving up 6. Conversely, I'll pick the pitcher who gives up 3 runs in each of two games over the pitcher who gives up 6 runs one game and zero the next. In other words, there may be patterns of pitching performance that are more condusive to winning for a given level of ERA+.

(2) Why are we so dismissive of the "bear down" kind of argument? 100 years ago pitchers "bore down" only in certain pressure situations. That's changed over the years to the point where now most pitchers bear down on almost every pitch. But even in Blyleven's day that wasn't always the case. Maybe some pitchers are better able to (or less able to) take it up a notch in crucial situations. or maybe some players have more of a tendency to lose their composure in pressure situations.* The fact that this argument, when deployed, is almost always bullshit, doesn't make it always bullshit.

Do either of these apply in Blyleven's career? Heck if I know.

*Let me just say in this context that the evidence I've seen regarding clutch performance has indeed made by sceptical of such claims, but I think it isn't strong enough to justify the almost complete dismissal of such claims by statheads. Certainly in our experiences in life, some people do indeed perform better under pressure than others. So I think that, absent the statistical evidence, you would expect "clutch" peformance to exist. The burden of proof is on the people dismissing such performance. A burden that has been partially, but not completely, met.

Look, I get upset myself by people who don't understand sample size calling (say) Barry Bonds a poor clutch player based upon a few post season ABs. But that instinctive reaction is no reason to completely dismiss the issue when dealing with larger sample sizes.

To maybe clarify a bit - the fact that Blyleven won 13 games less than expected is "indirect" evidence that his other metrics may (may) slightly inflate his value. Or it may, as you believe, just be luck.

But if evidence did exist that he was worse than expected in "clutch" situations* - that would be direct evidence, and it would be silly to dismiss it, whatever one thinks in the abstract about the existence of clutch "ability."

Again, none of this takes away from the fact that he does belong in the HOF. In fact, one might even argue that my anlysis butresses his case, in the sense that he won "only" 13 less games than "expected."

*eyeballing the data - man, baseball-reference.com has added SO much data over the past couple of years - Blyleven does seem to have perhaps performed a little bit worse in "clutch" situations - but I don't know what the norms are (maybe most pitchers do?), so I draw no firm conclusions from that. And it's not a LOT worse. (.684 OPS in late/close situations, versus .668 overall - the fact that almost all pitchers tire a bit late in the game probably explains most if not all of that. He also pitched much better in low leverage versus high leverage situations, but again I don't know what the norms are.)

One last thing and then I'll shut up. Just glancing at 2 season's worth of seasonal data - 1972 and 1977, two of the years in which he significantly "under performed" in terms of wins. In both years, especially 1972, he had pretty lousy late/close data. All the usual caveats regarding sample size apply, but I suspect that we do have part of the answer why he underperformed in those years.

FWIW - honestly, despite what I consider partial vindication of my concerns, on the whole this look at Blyleven has actually strengthened my over all appreciation of his career.

Thanks for all the comments, Larry. Your two examples are good ones, and sure, that stuff might exist. But I don't think there's any meaningful evidence of it with Bert at all. As I said above, you're talking about 13 games. If you flipped a coin 500 times, what are the odds that it would come up heads or tales 237 times or less? I don't have any idea, but I bet it's not terribly unlikely. It just seems more likely to me to be a slightly wider than expected distribution in the number of runs he was given. I don't think there's any question just from the years we've looked at that, in the years when Bert was at his best, he had an unusual tendency to throw his best games on days when his teams didn't score any runs at all. I can't come up with any feasible way of blaming that on Bert, aside from teh fact that he himself was unable to create a whole lot of runs at the plate.

And I think we can do away with any ideas that have anything to do with crumbling in pressure situations or the like. He won those 15 1-0 games, apparently more than anyone ever. He threw a no-hitter. He pitched very well in the postseason.

For this kind of effect to be real and observable, I think we'd need to see some kind of semi-regular pattern from year to year, and I think it would need to be a much larger number than 13 wins out of 537 decisions.

I could be totally wrong...I'm just not seeing it. Thanks again for the discussion.

just to throw a hay-penny in here, i noticed someone mentioned that pitchers can't control wins, but can control giving up runs. i understand the sentiment, but sentimentality does little good in the reality of baseball. the fact is the pitcher can control the runs he gives up over the course of a game, and while harvey haddix could have told you that perfection through nine... or twelve... isn't always enough, the idea that the only thing the pitcher controls is getting outs and not giving up runs is silly. baseball, more than any other game, is about timing and momentum. the pitcher controls the efficiency with which he gets outs and that next level of focus that keeps mistakes from occurring in critical situations. i'm sorry that i don't have any statistics, but baseball is baseball because statistics and probabilities can be defied by great athletes and great men.

in relation to blyleven, i'm not saying he wasn't clutch or he wasn't efficient or he wasn't consistent, because i didn't watch him enough to know. i'm saying that wins and losses should be first and foremost in analysis of pitching performance. the rest is nice, and a freak-talent-loser can be a hall of famer, but the guys on the borderline shouldn't be judged on their outs and runs allowed, but whether those outs and runs allowed were used effectively to win baseball games.