Scoring and Not Scoring the Runner From Third
I don’t mean to pick on Jose Abreu. Abreu has had a fine start to the season, and he is currently under the weather. But, to this point, Abreu has batted five times with less than two out and a runner on third. In zero of those five chances has the runner been driven home. Twice, Abreu has hit into a double play. Once, he’s popped out. Once, he’s struck out. And once, he’s lined out. He’s better than this, of course, and eventually the RBI will be there, but from a fan perspective, few things are more frustrating than such a wasted opportunity.
Every opportunity that doesn’t work out feels wasted. And in large part I think it’s because people don’t really know what normal is. Of course teams can’t convert every opportunity, but, it’s just moving a runner up 90 feet, right? It sounds like it should be easy. In a way, it’s just like bunting. You feel like everyone should be able to do it, but it’s surprisingly challenging to execute. It’s helpful to look at the league-wide numbers. It’s the only way to establish the proper context.
From this point forward, just about everything comes from Baseball Reference. As I write this, the Cubs have batted 62 times with less than two out and a runner on third. That runner has scored 22 times. Is that good? Is that bad? How bad is that? The Phillies have batted 53 times with less than two out and a runner on third. That runner has scored 33 times. Same questions. Good? Bad? I can tell you, the Cubs’ result is bad, and the Phillies’ result is good. The current average success rate is about 49%. Last year, it was about 51%.
Think about that. I don’t know what you expected to be true, but for many of you, you probably thought there would be a greater rate of success. About half the time, that runner isn’t driven in. Now, this isn’t exactly true — it’s not a bad thing to, say, draw a walk. The worst are the unproductive outs. Baseball Reference doesn’t narrow this down further, but what we have still works as a proxy. And this allows us to examine baseball history.
Strikeouts are way up. You know this. Big swings are also en vogue. A common opinion is that modern-day baseball players aren’t very good at the fundamentals. That would be, relative to the past. And how much more fundamental does it get than driving in a runner from third with less than two out? We all have an inclination to romanticize the past, and it’s easy to forget about what older baseball was actually like. Still, there is this plot of strikeout rates. These are the league-wide strikeout rates — since 1950 — with a runner on third and less than two out.
The increase is unmistakable. It’s also unsurprising, if you’ve been paying almost literally any amount of attention. There is less contact now than there’s ever been. On the other hand, let’s look at this again, but while also including the overall league strikeout rates.
You can see a separation start to develop in the 80s and 90s. The strikeout rates used to be more or less identical. Now there’s a split, with batters whiffing less often in the given run-scoring situation. This reflects a shift in strategy. Batters are still striking out more often with a runner on third and less than two out, compared to their previous baseline, but there is a sign they’re prioritizing contact. More strikeouts happen under other circumstances.
So that’s one look. But even more to the point, we can look at success rates since 1950. That’s plate appearances with a runner on third and less than two out that result in the runner on third scoring. These days, it’s somewhere around 50%. Where has it been before?
I wouldn’t pay much attention to 2018 yet — we still have another five months to go. Relative to, say, the peak of the steroid era, we have seen a bit of a drop. But we’re talking one or two percentage points, max, and there was a huge, huge dip in the 60s and 70s. As a different but related look, I pulled all plate appearances with a runner on third and less than two out, and then calculated runs per PA. This would also count other runners scoring, and not just the guy on third. Every run counts the same in the box score.
The plots mirror one another, of course. And while, on a year-to-year basis, you see the lines bounce around, this table splits everything by decade:
| Decade | Success% | R/PA | 3rd, < 2 Out K% | Overall K% |
|---|---|---|---|---|
| 1950s | 50.1% | 0.64 | 10.9% | 11.4% |
| 1960s | 47.5% | 0.61 | 14.4% | 15.0% |
| 1970s | 48.6% | 0.61 | 13.3% | 13.5% |
| 1980s | 51.7% | 0.64 | 13.0% | 14.0% |
| 1990s | 52.3% | 0.66 | 14.4% | 15.9% |
| 2000s | 51.8% | 0.66 | 14.8% | 17.0% |
| 2010s | 50.7% | 0.64 | 17.3% | 20.0% |
It’s true, beyond any doubt, that modern hitters strike out a lot. Strikeouts with a runner on third are frustrating. When you watch one, you figure all that was needed was a measly ball in play. But, there are plenty of unsuccessful balls in play. And in baseball’s earlier decades, there were even more unsuccessful balls in play. You might think there would be a strong relationship between strikeout rate and success rate — or failure rate — in these cases, but that just isn’t what we see. Batters aren’t meaningfully worse at driving the runner from third home. They’re still producing runs in those plate appearances. The league’s collective success rate isn’t extraordinarily high, but it never has been. It topped out at a hair under 54%, in the year 2000. Did baseball have the most fundamentally-sound players in the year 2000?
Used to be, baseball games featured more balls in play. And that might well be a good thing by itself. So many of those batted balls, however, were effectively useless. And it’s the useless batted balls that have been replaced by the strikeouts, in a sense. When there’s a runner on third and less than two out, a hitter in 2018 will find it challenging to drive the guy home. It’s been challenging all along.
Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.




Are the numbers for driving in a runner from third with less than two outs just counting the runs from the ABs with less than two outs and not subsequent ABs (i.e., if a guys strikes out with one out, but guy behind him gets a single to drive in the runner at third, does that count as success)?
I think it’s just AB with a runner on third so it probably counts all AB. So if a guy strikes out but a guy behind him drives him in it’s counted as a strike inn This graph but also counted as a success
I was thinking the same thing. I’m curious to know “percentage of time a runner scores after being on third with less than 2 outs”.
Looked up all scenarios for scoring a run any time after man on third with less two outs on Tango’s run expectancy charts, the lowest percentage was 63%. As such, I assume Jeff’s results are for that plate appearance only and not inclusive of subsequent plate appearances in the inning.
Confirming Joe Joe’s inference, the numbers cited from B-Ref are strictly for the defined situation — batting with a runner on 3rd and less than 2 out. No 2-out results play any role; for that, you’d want a run-expectancy table.
The increase in strikeouts would be offset to some degree by a decrease in GIDP. Has the league seen a decrease in GIDP in the <2 outs, runner on 3rd situation?
I came here to ask this as well. Would love to see an analysis of this. Strikeouts might feel like a “wasted out”, but with less than 2 outs in the inning and a man on first, it may be safer to K than choke up just to make some soft contact
In five year chunks, SO/AB vs GDP/AB:
YEARS SO% GDP%
1984-1988 16.5% 2.21%
1989-1993 16.8% 2.16%
1994-1998 18.7% 2.24%
1999-2003 18.9% 2.29%
2004-2008 19.1% 2.33%
2009-2013 21.2% 2.22%
2014-2018 23.3% 2.24%
[edit]Year-to-year correlation, -0.007
You are assuming a decrease in GIDP, which I don’t think is a given. I am not sure that it would be all that significant. The infield would be up a bunch of the time and not even trying to turn a DP. I get a bit of a feel that you want to show that modern hitters are better… they probably hit more weak fly-balls which are truly useless. I would imagine that scoring a runner from third is pretty constant – no need to dig. From the hitter’s perspective it is just another at bat – really not all that different. I know they like to think of it as different, but if they just treated it like any other at bat it wouldn’t change much. Well hit balls score runs.
Good research!
I don’t think it’s out of bounds to suggest that the increase in fly balls have led to more sac flies than unproductive grounders. I wonder what the grounder rate was in the 60s/70s compared to today.
Unproductive grounders move runners, which is what this entire article is about. Unproductive grounders also induce errors, which look like zeros in the data but are back-breakers in reality. Yes, I am a ground-ball-apologist.
I’ve always relied on Tango’s Run Expectancy charts, which suggest a better than 80% chance a runner on 3rd will score with zero outs and better than 60% chance with one out. But I’ve become increasingly concerned with the relevancy of the data used for those tables, as the most recent set ends in 2015 — right before the new air-ball / juiced ball era began. And with Tango now cocooned inside Statcast wrapped by Baseball Savant inside MLB, that table isn’t going to be updated. Unless Baseball Savant publishes a new version (we can hope, right?)
This data can still be true and relevant, since run expectancy takes into account potential future plate appearances – Jeff’s tables are about isolating each individual plate appearance’s success rate.
It might be interesting to look as well as the hit profile that results when contact is made with less than 2 outs and a man on third.
Like batted ball stats? I would also be interested in that. I bet you EV is significantly worse.
Fantastic read as usual, Jeff. It’s got me thinking of ten different tangents around the way we value players and how we measure defensive value in particular.
More specifically, we do not have a good way to reconcile “defensive environment” across leagues, parks, years, etc in the same way that we do with pitching and hitting.
Since defense makes up a pretty substantial portion of a player’s career WAR, it follows that WAR really isn’t great at comparing players across history.
Which is more impressive – Bob Gibson’s 1.12 ERA in ’68, Pedro’s 1.74 in ’99, or Maddux’ 1.56 in ’94? If we use ERA- we can see they are all equivalently amazing. We can even index some of the underlying numbers – K%, BB%, HR/FB, etc. In fact, FIP- is simply taking all those peripherals and folding it into a single easily-indexed reference point. (spoiler alert: vote for pedro). Is DIPS theory perfect? No, but it works.
Same with hitting . We have mountains of data on individual players dating back decades, and it’s easy to create league- and park-adjusted indexes for lots of peripherals. We can also model the context-neutral value of individual hits – a HR is worth more (but not 4x more) than a single, etc. We can even measure this down to the negative impact of GIDP vs a strikeout (2013 mvp ptsd triggered). Are linear weights perfect? No, but it works.
But with fielding? We have _none_ of this underlying data. In fact we barely even have the primary data (total zone is not great, bob). This article is a really good example of why that matters.
How do we compare the relative difficulty of fielding a ground ball in 2018 vs 1998 vs 1968? When K% was at an all time low, presumably the contact was weaker and therefore defensive opportunities were easier (and more frequent). How much easier were they? Does that matter? Are park adjustments made for defense? Fielding a grounder in the astrodome “grass” certainly isn’t the same as yankee stadium in may.
Defensive metrics like UZR and Total Zone are already limited by how few opportunities an individual fielder might see. DRS attempts to correct for this somewhat. If public statcast data ever becomes available that will be a gold mine for doing this stuff. None of that applies historically though because the data simply doesn’t exist.
Why does this matter? Take Andruw Jones. 66 career WAR, kind of on the bubble for the HOF. But 2/3 of his career value comes from defense. How accurate are those defensive numbers? The way fangraphs delivers the information, you’d believe it’s accurate to the 10th-of-a-run. Do we know Andruw Jones’ career value as 66.90 +/- 0.01 WAR? Hell no! Maybe +/- 10 WAR is more realistic. Was he a 50 win player? or an 80 win player? Honestly, you could convince me either way.
What about manny ramirez? 66.3 career war, just like Andruw Jones. Jones had +280 runs of defensive value, ramirez had -280. Ramirez lost a lot of that value while playing LF at fenway park, which is notoriously difficult to capture in zone-based metrics.
My point is not “Manny Ramirez is better than Andruw Jones”. They both had incredible careers in their own way. My point is not “WAR is bad!!!” If you really dig into it, WAR tells the story of these two players quite well – a generational bat and a generational glove.
But if you want to use WAR to compare, say, Manny Ramirez to Miguel Cabrera? Or Andruw Jones to Kevin Keiermeier? I just don’t think it’s that simple. And if you start including players from further back – Jim Rice? Dwight Evans? Mickey Mantle? Willie Mays? I don’t know, WAR just doesn’t cut it for me.
Anyways this is getting more than a little rambly, I just thought this was a great article because it really made me think about how I am comparing the value of players across generations. I always love when an article makes me think!
I have said it a million times before, WAR is not a good tool for using across generations. For no reason other than the data points will always be changing due to new metrics and technology/accuracy. The other problem is that the values of on-field skills change. Some eras de-criminalize Ks and some were hesitant to take walks. WAR is good for comparing similar players in the same era only – everything else is highly questionable. Only a tool like WAR could argue that Vladimir Guererro was a borderline HOF player… which is exactly what the majority of commenters made clear a year ago – which was fueled by WAR and WAR-link metrics. Great players don’t fit into the same containers as everyone else – that is what makes them unique. Depending on what makes the player unique, an algorithm may penalize a player or reward him depending on the algorithm, which will likely reflect the current generation. That is where things start to get messy. The people who create the statistics have a great deal of interest in making it reflect highly on current players – otherwise, nobody will want to hear it.
Muy buen artículo.
This is an excellent article and it certainly opened my eyes. I have been under the assumption that the increased K rate would have affected the percentage of success in getting a man in from third with less than 2 outs much more than it apparently has. The feeble two-hopper to short gets the job done but the K doesn’t. It is logical to think that many of these “useless batted balls” weren’t actually useless and it would not have surprised me to see a 3-5% drop in the success rate in the last ten years. With that in mind, could anyone give me the success rate of suicide squeeze attempts and the percentage of times the defense botches the suicide squeeze completely, which is not an inconsequential number.
That is a problem with bunting outcomes in a nutshell… to a computer. Bunts induce lots of errors, which look like outs or FCs statistically, which is far from what actually happened.
But one of the things that you have to remember is that the two-hopper to short doesn’t always get the job done – sometimes the infield is in, and the runner doesn’t go (contact play not on) or the runner goes and gets thrown out. And sometimes the two-hopper is to 3rd, and the runner is even less likely to go, or more likely to be thrown out. Or the batter pops out. There are lots of ways a non-strikeout can still not result in getting the run in.
And I would think these numbers already account for instances when the run scores on an error, via suicide squeeze or otherwise. Although I wonder whether than includes runs that score via wild pitch (or balk), since this is not a batter-driven outcome. I’m sure it’s accounted for in the overall run expectancy numbers, but not sure if it is here, in the runs with less than two-out stats.
interesting read Jeff. Thank you
One thing that would be useful here is adjusting the scoring rate (i.e, the latter two plots) in some way for the league-wide scoring rate. I’d be interested in knowing whether the probability of the runner from third scoring is less/greater than some baseline likelihood in the presence of no strategic changes to get him in.
Good work, Jeff, tackling a common misconception. Many or most observers assume that the high-K era, which includes a decline in all small-ball strategies, would bring less success in this situation. Most of them, including a shocking number of commentators, just don’t bother to check the numbers.
Scoring a runner from third isn’t “fundamentals”. Fundamentals are things like smart base-running, squaring around early to bunt, not dropping your hands below the ball when bunting – they are things that anyone can do – not skill-based. Scoring a runner from third isn’t fundamentals – a hitter need to hits the ball well… which is what they need to do every other time at bat as well. Its really not all that different than the rest of the at bats. Its not like they just need to do something simple, which is what I would call fundamentals.
“So many of those batted balls, however, were effectively useless” – only useless to those who lack context, like machines and anyone else who isn’t well-versed in on-field baseball. I imagine there will come a time when some sense of accuracy will be applied to useless balls in play – perhaps a penalty for strikeouts with situational weights. We should spend more time examining the limitations of data, machines and algorithms – it would be the best thing anyone could do to advance sabermetrics. It would be really painful in the short-term…