The Stranded Ones
Lost in the Kansas City Royals’ Yuniesky Betancourt-fueled 5-3 extra-innings win over the Cardinals yesterday was that Alex Gordon tied his franchise record with five walks in a single game. Yeah, I’m as furious as the rest of you that this was overlooked. But hey, at least this time he at least scored a run. When Gordon originally set the record (as I am sure you all remember) back on July 30, 2008 against Oakland, he did not score once. (I remember that game well, as not long before that I had an argument with someone who said that Jose Guillen was the Royals’ only “feared” hitter, unlike that loser Gordon. Guillen hit right behind Gordon in this game. FEAR.)
It was an amazing feat, in a way, but not nearly the most times on base without scoring. In fact, since 1918, there have been 73 players who have gotten on base six or more times without scoring in regular season games. What follows is a look at the most extreme cases.
While FanGraphs does have a handy play-by-play database going back to 1974, some of the cases of being stranded happened before that, so I have utilized Baseball Reference’s Play Index and game summaries for those. Also, I have used the “official” definition of times on base: hits + walks + hit by pitch (where available). Yes, I think that reached on error should be treated differently, but it has not been in baseball’s record keeping, and I wanted to be consistent.
Since 1918, there have been seven cases of players reaching base seven or more times in one game without scoring. Six of them involve getting on base seven times. Here they are:
Paul Waner, September 22, 1931. The notoriously hard-drinking “Big Poison” was in the midst of his Hall-of-Fame prime in 1931, so some of his five recorded walks might have have been intentional. However, intentional walks were not recorded distinctly at this time. The Pirates still managed to beat the Phillies 3-2 despite their inability to drive Waner in.
Ron Northey, September 24, 1944. Northey quite probably had the least impressive career of any player on this list. Sure, he had four singles, and the 1944 Phillies were pretty terrible, but it is still kind of hard to believe the Cardinals walked him three times. The Cards won, anyway, 4-3.
Reggie Smith, September 13, 1974. Smith arguably had a Hall of Fame-quality career, and it is really hard to believe that a player of his skill moved around so much in his prime (which might have cost him). He had a 150 wRC+ for the Red Sox in 1973, so naturally they traded him to St. Louis, where he regressed all the way to a 151 wRC+. Well done. This game went 17 innings, so Smith came to the plate nine times, collecting one hit, five walks, and a hit by pitch. Despite getting stranded all seven times, the Cardinals managed to beat the Phillies 7-3.
George Brett, June 6, 1991. Alex Gordon may hold the record for most walks in a game, but he’ll never be the next George Brett, so BUST. The Royals won this 18-inning affair against the Rangers, but Brett not only did not score any of team’s four runs, but he did not drive in any of them, either.
Eric Young, June 30, 2000. He was more of a “Northey” than a “Brett,” but Eric Young did have a some good seasons as a lead-off hitter. This game was deadlocked at four at the end of nine, and the Cubs won it by scoring three in the top of the 15th. However, despite a hit, five walks (only one intentional), and a hit by pitch, Young scored none of them.
Prince Fielder, April 10, 2007. Prince had a nice season in 2006, but 2007 was his first “monster” year, as he finished with a 152 wRC+ and 50 home runs. The Brewers one this one in 13, but Fielder was stranded despite getting on base all seven times he came to the plate.
Those are each interesting, but one player got on base eight times in one game without scoring:
Rod Carew, May 12, 1972. Maybe you’ve heard of this guy. This was actually one of Carew’s lesser seasons (by his standards), as he only hit .318 (122 wRC+). What is amazing is that while Carew got on base eight times, he actually had ten plate appearances during this 22-inning marathon. He was walked three times (twice intentionally) and had five hits. The Twins’ failure to drive him in even a single time cost them, as they lost the game 4-3.
I’m inferring that these were some very long games. Players don’t usually bat seven times in games with scores like 3-2 and 4-3.
We need a corresponding rotographs piece talking about most-stranded hitters this year!
Eric Young might me the worst ESPN analyst ever. And I don’t say that lightly.
I used to think Harold Reynolds was embarrassing himself, talking in non-sensical circles every night – now I think he really has something to say.
Reynolds has 6.5 years on Eric Young – announcers may not peak until their 50′s.
This illustrates why stats like wOBA and WAR can be extremely misleading. In each of these cases, these players performed very well in terms of wOBA, but in actuality, they were pretty worthless with respect to actually helping their teams win (unless they drove in runs).
The real problem, and one that is never acknowledged, is that wOBA flattens events into their likelihood of producing runs, but it does so by throwing away the variance in the events. Because the variance between events is not equal, you lose a lot of important information. Home runs produce runs with less variance than do walks. A player a who achieves a wOBA of .350 through walks is not the same as a player who achieves a wOBA of .350 through home runs. The former is more likely to either exceed or underperform their expectation. The latter is a sure bet.
“in actuality, they were pretty worthless with respect to actually helping their teams win”
Actually, the players hitting behind them were pretty worthless with respect to actually helping their teams win.
Actually, both were.
@Jason H
Your response to me seems in line with everything else you have posted, such that I will only post one general response to add to the correct points that LTG has been making…
Blaming someone who walks for not scoring because the players behind him failed (i.e. when variance means a walk does not lead to runs), is silly, because (obvious claim alert) that player’s not scoring has NOTHING to do with his walk.
Saying that “In reality, his walk did not lead to a run” is missing the point. Getting on base and avoiding an out so that another player can have a chance at the plate in a positive run-scoring environment IS IN ITSELF helping said player’s team win. His teammates happened not to capitalize on the opportunity. That is NOT the walking player’s fault.
You are confusing real possibility with epistemic possibility. Real possibility tracks the likelihood that one event follows another. Epistemic possibility tracks how confident we are that one possibility follows another. For each walk we might not know whether it will produce a run (unless the bases are loaded), whereas we do know that a homer will produce at least one and exactly how many. But what we know is not relevant. Only what happens is. And what happens after a walk is a certainly probably that a run is going to score given the game state. We can’t get more fine-grained than that without relinquishing the goal of finding a player’s skill.
Also, you are cherry-picking examples. In statistics, outliers are not counter-examples.
I don’t understand this. Maybe you could explain it again.
I’m not cherry picking examples. Walks have greater variance in producing runs than home runs. These outliers are illustrations. In each case, the player had a great game. However, the result was not useful. Sometimes that happens. Its more likely to happen with walks than with home runs though.
Yes, a home run is more useful than a walk. But that is reflected in the linear weights of each event, which are used to calculate all manners of newfangled stats like wOBA. No stat except OBP treats walks and homers as if they were equal. The extra variance you are trying to point out is not real variance but just a variance in our knowledge of the particular result of a given event (whether it leads to a run). Here’s where you shift to epistemic language: “The former is more likely to either exceed or underperform their expectation. The latter is a sure bet.”
And, yes, these examples are cherry-picked because you are using them to motivate an objection to the theory. But if the model predicts outliers, as any statistical model does, then you have to show how the outliers undermine the explanatory power of the model. Losing one game doesn’t show a lack of explanatory power because there is too much randomness to a one-off test. You would have to show that teams with high wOBA due to walks and average pitching did not tend to win as many games (at a statistically significant level) as teams with equally high wOBA due to HRs.
By the way, this is why Black Swan events are not just outliers in economic theory but real objections to the efficient market hypothesis. First, they happen more frequently than predicted. Second, the severity of the lost capital belies the explanatory power of the model: it is one event with as much loss as years of gain. This is not to say that we ought to throw away all of the mathematical models of economics. We just need to make some adjustments both to the theory and the ideology that has enveloped the theory.
LTG,
“The extra variance you are trying to point out is not real variance but just a variance in our knowledge of the particular result of a given event (whether it leads to a run).”
This is wrong. The variance is real and has to do with the rules of baseball. Players that hit a home run always score. There is zero variance in that outcome. It doesn’t matter what team you play for, where you hit in the lineup, how many outs there are, etc., you always score. A player that walks either scores or he doesn’t. There is variance in that outcome. Whether he scores or not is dependent upon many things that are largely out of his control.
As you correctly point out, the linear weights account for the difference in run expectancy between walks and home runs such that a home run is not treated the same as a walk. However, wOBA does treat X number of walks the same as a home run (where X = hr/bb and hr = home linear weight and bb = walk linear weight). Looking forward, treating these the same may be reasonable because you EXPECT them to produce the same amount of runs (albeit the confidence intervals will be smaller around the HR than around the walks). However, looking retrospectively, these are rarely actually the same. The greater variance in the run expectancy of the walks means that the OBSERVED will deviate from the expected to a greater degree. Sometimes all the walks will score. Sometimes none will. Sometimes everything in between.
Further, the linear weights used in wOBA are calculated from league averages. No player actually plays on a league average team with the league average hitting behind him. A Redsox player who walks is probably more likely to score than a Mariner who walks. A Redsox player that hits a homerun is equally likely to score as a Mariner that hits a homerun (however he is probably more likely to drive in additional runs).
wOBA is a great tool for a simple evaluation of talent. wOBA is not a great tool for evaluating what actually happened in baseball games. I’m not trying to undermine this stat as you seem to think. I’m just pointing out its limitations, and the limitations are especially obvious when you look backwards at what actually happens in games.
“As you correctly point out, the linear weights account for the difference in run expectancy between walks and home runs such that a home run is not treated the same as a walk. However, wOBA does treat X number of walks the same as a home run (where X = hr/bb and hr = home linear weight and bb = walk linear weight). Looking forward, treating these the same may be reasonable because you EXPECT them to produce the same amount of runs (albeit the confidence intervals will be smaller around the HR than around the walks). However, looking retrospectively, these are rarely actually the same. The greater variance in the run expectancy of the walks means that the OBSERVED will deviate from the expected to a greater degree. Sometimes all the walks will score. Sometimes none will. Sometimes everything in between.”
This is only true with arbitrary end points, namely one game or one inning. But it is terribly unclear how to assign credit to individuals at the level of one inning or one game because the results of a game are so dependent on the interactions of all of the players and other environmental factors.
To return to your initial point that walking 7 times but not producing any runs means having a bad game, this requires specification. If I walk 7 times but could have swung at a couple of pitches that would have been HRs or XBHs, then maybe I did have a bad game or not as good a game as I could have. But if I walk 7 times and saw no pitches for which the previous counter-factual is true, then in what sense did I have a bad game? And how are we to tell whether that counter-factual is true? Even if I saw some fat pitches, who knows whether swinging at them would have produced a hit or an out?
You take it as obvious that we can relate results to performance after we know what happened without considering the probabilistic framework behind stats like wOBA. It certainly isn’t obvious. And I doubt we can do it with much confidence… yet.
“This is only true with arbitrary end points, namely one game or one inning.”
Those are not arbitrary end points!!! Those are the rules of baseball!!! I think you are losing site of the fact that the purpose of baseball is to win games, and not to evaluate skill individual skill level.
“To return to your initial point that walking 7 times but not producing any runs means having a bad game.”
This was not my initial point. You are misunderstanding. Walking 7 times is a good game for the player. Always. It is does not, necessarily mean much for the team though (if you never score for example). A single solo home run with six k’s would have been better for the team (even if it produced a lower wOBA for the player). My initial point was that some events always help the team, and some events only help the team some times.
If I’m a GM and I have a choice between two players both with a .350 wOBA and exactly equivalent defense and position, I take the guy who gets his wOBA through walks if I already have a power hitting team, and I take the guy who gets his wOBA through power if I have a light hitting team. Why? Because the guy who walks depends on his teammates more than the guy who hits home runs. The guy who walks may very well exceed their expected performance on a good team by a large margin. On the otherhand, the guy who walks is very likelihood to actually produce less value than expected on a poor hitting team (think of how few runs Ichiro scores). The guy who hits home runs is the safe bet since he relies on teammates less.
It’s also important to point out that you can help your team win, offensively, without scoring or driving in a run. The obvious way is, for instance, to walk between two singles, with the second hit scoring a run. Less obvious ways range from increasing the pitcher’s stress and pitch counts to providing more at-bats for your teammates.
Reggie Smith reaches base seven times, only to have his teammates win without his (apparent) help in the 17th inning – but those players may not have even had those at-bats until the 19th or 20th without his help. Alternatively, consider how important it is if you manage to do nothing all game but draw one walk – and subsequently get stranded – but your team comes from behind on a two-out walkoff.
These are great points.
“the limitations are especially obvious when you look backwards at what actually happens in games.”
It isn’t attempting to say exactly what happened in games, but rather show the value of what happened independent of what actually occurred afterwards. If a person walked 5 times in a game and scored 5 times, would you want the stat to say those walks had the value of 5 solo home runs?
That, of course, depends on the question. If the question is “how will the player perform going forward?” then, of course not. If the question is, “how valuable to the team was the player?” then perhaps I might.
But this misses the point of attempting to isolate skill (relative to MLB players) from other environmental factors, which is why we use predictive metrics as substitutes for True Talent.
If all you want is to know what results from various players playing with other players in particular games, just look at runs scored and RBIs (although RBIs already include judgments about whether the batter deserved credit for the run scoring).
At any rate, if what you mean by “what actually happened” is who deserves credit for the results on the field, then something like wOBA is still better than looking at traditional stats or just observing whether event E led to + or – runs. Individual players deserve credit for having dispositions that tend toward more runs for and fewer against.
Of course, at a certain level of description you are right that wOBA cannot tell us what actually happened. But who wanted it to give us that level of description? And at the level of description at which wOBA fails, you won’t be able to assign credit to players with any more confidence than you could with wOBA.
“But this misses the point of attempting to isolate skill (relative to MLB players) from other environmental factors.”
But that misses the point of trying to win ballgames?
In all seriousness, I certainly understand that particular statistics are useful for particular questions. However, this site frequently advocates using predictive statistics to evaluate past performance. This site generally argues that the player with the greatest WAR should win the MVP. We don’t give the MVP to the player with the greatest skill. If we did we could assign it to Albert Pujols before the season starts. We know he has the most skill (of course this has recently become debateable).
I absolutely think runs and RBI are a more useful way of determining MVP than wOBA. Baseball is a team sport. I have no problem with the fact that the rest of the team’s performance plays a role in how these stats are accrued. I don’t want to isolate a player’s performance from the team’s when I am trying to determine value to that team. For it is precisely this that is valuable in terms of actually winning ballgames.
Now, if I am trying to determine a player’s skill level relative to his peers than of course I want to isolate his performance.
“I don’t want to isolate a player’s performance from the team’s when I am trying to determine value to that team. For it is precisely this that is valuable in terms of actually winning ballgames.”
Yes, winning ballgames involves multiple players interacting in certain ways. If player A goes 5-5 with 5 triples, but does not score (because the following players keep striking out), then those triples ultimately happened to be useless in that game. But, for player A, he added a whole great deal of value in terms of run-scoring opportunities and giving the following hitters and easy opportunity to plate runs. Your thoughts seem to hold something against player A. Why demean his amazing performance, where he put his teammates in an amazing spot to produce runs, just because they happened to fail to capitalize on that opportunity?
“But, for player A, he added a whole great deal of value in terms of run-scoring opportunities and giving the following hitters and easy opportunity to plate runs.”
No, he added the chance for value. In your scenario it ultimately proved not to be valuable. Sometimes that happens. Unlucky.
“Your thoughts seem to hold something against player A.”
No, of course not. He did a great job trying to help his team win. It just didn’t work out for him.
“Why demean his amazing performance, where he put his teammates in an amazing spot to produce runs, just because they happened to fail to capitalize on that opportunity?”
Its not diminishing his performance. It just happens to be the case, given your scenario, that his amazing effort didn’t much help his team win the game. To me, that is part of the beauty of baseball.
There is this bizarre tendency in the SABR community to try to remove the stochastic elements of the game while looking back retrospectively. The stochastic elements of the game are a big part of the enjoyment and beauty of the game. Without them, we would be able to predict the outcome of games with certainty. If you follow the SABR logic to its logical conclusion, we really could assign awards before the season begins. If you remove all of the stochasticity, all you are left with is individual skill.
In practice, it is not possible to remove all of the stochasticity. The stochasticity related to players playing on different teams and batting in different lineups has been removed with linear weights. This really is better for looking at skill. However, it ignores a lot of what actually happened in the games and replaces it with what happens on average.
But wOBA does not remove all of the stochasticity. Not even close. All of the BABIP related randomness, for example, is still there. Again, this site advocates using WAR to determine MVPs. Why do they advocate removing randomness associated with teams and lineup spots, but not randomness associated with luck on balls in play? There is no good reason for this.
If I want to look back and say who had the best season, I want to know what actually happened in the games. I dont want to remove the randomness. …If I want to know who has the most skill, that’s a different story.
“No, he added the chance for value. In your scenario it ultimately proved not to be valuable. Sometimes that happens. Unlucky.”
No, he left his team in an EXTREMELY valuable situation. His teammates made in non-valuable…that has nothing to do with our original value provider.
“Its not diminishing his performance. It just happens to be the case, given your scenario, that his amazing effort didn’t much help his team win the game. To me, that is part of the beauty of baseball.”
I agree baseball is beautiful. But, I disagree that his performance did not help his team win a game. Given the awesome performance we are assuming for player A, A has greatly increased his team’s chances of winning. You can’t blame A for the fact that OTHER players have hurt his team’s chance of winning. After each A PA, A’s team had a much higher win probability than before each A PA, yet those following A have squandered that advantage. The advantage A gave is significant and demonstrable, even though the lack of success off others may harm A’s awesome impact.
RationalSportsFan,
You are arguing against things that I don’t think. Once again, I don’t want to take anything away from a player for having done his job. I don’t know if I can be more clear about this. You are addressing a different question than I am. You are talking about skill, and I am talking about what actually happened in games. What actually happened in games is a combination of skill and randomness.
For example, you don’t know that the players hitting behind the mega-walker didn’t do their jobs. Suppose the player hitting behind him actually hit 7 home runs, except each time Torii Hunter pulled the ball back over the wall to rob him of the HR and make an out. You seem perfectly happy to leave that randomness in. And you should be, since that is actually what happened. Why, then, looking back would you want to credit a player with runs scored that never scored and take away extra runs scored (because that is what wOBA does. It makes everyone average)?
Nobody gives a shit about “What Actually Happened in Games.”
Note that at least 5, and perhaps the other 3 of these games, went extra innings. That would tend to make them low scoring games, thereby increasing the chance that baserunners would be stranded.