Ground Ball/Fly Ball Ratio: An Often Overlooked Pitching Metric
Robbie Ross - Photo courtesy of Jason ParksThe Baseball Cube doesn’t list it. Neither does Baseball Reference nor FanGraphs. If you want to know the ratio of ground balls to fly balls being produced by a minor league pitcher, you have to go to the player pages located at MILB.com or Minor League Splits for what I consider to be the third most important stat in predicting major league success behind strikeout and walk rates. What is so important about GB/FB, you ask? Well, pull up a chair my friend and take a look.
In a study published in The Hardball Times Baseball Annual 2007, Dave Studeman used batted ball information from 2006 to estimate the run values of strikeouts (-0.113), walks (0.30), ground balls (0.045), fly balls (0.192), and line drives (0.391). Studeman concluded that fly balls are more than four times as valuable as ground balls, even though the batting average on ground balls was slightly higher than fly balls. The discrepancy in value derives from the fact that fly balls are more likely to go for extra base hits and ground balls are more likely to produce double plays.
To appreciate the difference between fly balls and ground balls, consider two pitchers who neither walk nor strike out anyone. One pitcher is an extreme fly ball pitcher (i.e. all balls hit into play are fly balls) and the other is an extreme ground ball pitcher (i.e. all balls hit into play are ground balls). Assuming the batting average against for ground ball pitcher is .300 and the fly ball pitcher is .225, the ERA for the ground ball pitcher would be 1.74 and for the fly ball pitcher would be 6.69.
Obviously, there are no pitchers who produce only ground balls, though it is worth noting that there are three times as many ground ball pitchers (i.e. pitchers who produce more ground balls than fly balls) as there are fly ball pitchers among major league starters. To get a more realistic view of how GB/FB affects the ERAs of major league pitchers, the chart below plots expected runs per nine innings using the run values calculated by Studeman. Key assumptions for the chart below are that 18 percent of balls in play are line drives, 8 percent are infield fly balls, and that the batting average against for non-strikeouts is .300:
The difference in ERA between a high strikeout-low walk/ground ball pitcher and a low strikeout-high walk/fly ball pitcher is substantial (3.15 vs. 5.90). Assuming similar walk and GB/FB rates, the difference between a strikeout pitcher (9.0 K/9) and a non-strikeout pitcher (5.0 K/9) is nearly 1.5 runs per game. Assuming similar strikeout and walk rates, the difference between a fly ball pitcher (0.75 GB/FB) and a ground ball pitcher (2.00 GB/FB) is approximately 0.75 runs per game. Assuming similar strikeout and GB/FB rates, the difference between a pitcher with good control (2.0 BB/9) and relatively poor control (4.0 BB/9) is 0.6 runs per game.
Highlighted in gray near the top center of the table is the projected ERA for a solid-average major league starting pitcher (7.0 K/9, 3.0 K/9 and 1.00 GB/FB). The other gray highlighted cells show how pitchers with fewer strikeouts, more walks, and lower GB/FB ratios can produce the same ERA by being better than average in other areas. Perhaps most interesting is the observation that a pitcher with 5.0 K/9 can actually have an above-average ERA by limiting walks and inducing twice as many ground balls as fly balls.
RANGERS PITCHERS
2009 GB/FB, strikeout, and walk rates of Ranger pitchers and prospects. For prospects who have pitched at multiple level, the statistics listed are for the level where they have pitched the most this season:
[ERA* -- Estimated ERA using GB/FB, K/9, and BB/9.]
12 Comments | in Analysis, Prospects, The Season
Reader Comments (12)
Looks like all of the Rangers' ERA* is greater than thier actual ERA similar to FIP.
With all of Bleier's excellent peripherals, it's a wonder to me that he often gets overlooked among the Rangers minor league pitching prospects. Care to comment?
Rob - The fact that most of the Rangers' pitchers have ERA's that are superior to the thier FIPs supports the notion that the team's defense has significantly improved. However, I don't think that is the full story since the pitching staffs of other teams with excellent defenses don't seem to be benefiting to the same degree. I think that there might be a Mike Maddux effect at work, but it is still too early in my analysis to be sure. Hopefully, there will be something worth posting in teh next week or so.
rocky371 - Count me among those who have failed to get excited about Bleier. He's 22 years old in A+ and striking out fewer than 7 per 9 IP. He could be Doug Mathis, but in a system with 20 or so very talented pitchers under the age of 25, it will be very difficult for him to get more than a cup of coffee.
Looks like the metric used to predict ERA overestimates by about 1 run. How did you calculate the predicted ERA ratio?
Also interesting that Hunter's GB/FB ratio went the other way from minors to majors; his K/9 declined slightly; yet his ERA dropped significantly this year from AAA to Texas...... perhaps a testament to great ML defense?
We appreciate your work....
Will Kasey Kiker ever make the Rangers 25-man roster? His GB/FB ratio, BB/9, and pitching efficiency suggest that he will struggle mightily in the MLs.
The predicted ERAs listed for each of the Rangers pitchers was calculated in two steps:
Step #1 - Estimate the number of flyballs, groundballs, line drives, popups, strikeouts, and walks that should occur per nine innings based upon each pitcher's GB/FB ratio, K/9, and BB/9
Step #2 - Multiply the Studeman's run values for each hitter outcome (strikeout, walk, line drive, fly ball, groundball, popup) by the estimated number of times that each should occur in a game
The fact that the predicted ERAs are primarily higher than the actual ERAs suggests that the Rangers' defense is better than the average major league defense in 2006 since the run values were derived from 2006 batted ball information. As noted in my comment to Rob, I think that there is more to this than simply the Rangers' defense, but I need to test a couple of ideas before being sure.
Hunter's GB/FB ratio in 2009 is schewed due to two games in late July/early August when he gave up 24 fly balls and only 5 groundballs against Boston and Seattle. Hunter's minor league career GB/FB is 1.54 and I expect him to approach that number as a major leaguer.
Regarding Hunter's 2.64 ML ERA in 2009, I think that he has had his fair share of good fortune. His BABIP (.253), LOB % (79.7), and HR/FB (6.6%) have all been significantly better than league average. Don't be surprised to see more of his starts go like his last one. Long term, I think Hunter will be a 6 K/9, 3 BB/9, 1.30 GB/FB, 4.75 ERA pitcher. Assuming he remains durable, that will be extremely valuable to the team.
The more I look at Kiker's performance numbers, the less convinced I am that he is going to be a productive major leaguer. He's still young, so he has time to develop command of a low 90's fastball which will be necessary for him to be considered anything more than a fringe prospect.
David: What is it about Ross' stuff that induces so many ground balls? (I know that's too simplistic, but I'd love to hear more about this kid.)
Rockwall - You can check out Robbie Ross' most recent minor league outing at http://www.boisehawks.tv/vod2.asp?081709.wsx. My takeaways from that game are that Ross:
(1) throws hard,
(2) has excellent command of his fastball and slider,
(3) has good downward movement on each of his pitches,
(4) does an excellent job of using angles.
Based on this one game, it appears that Ross' high groundball rate results from consistently working low in the strike zone and inducing weak contact after getting ahead in the count.
Regarding fangraphs, they do post percentages of each type of batted ball and GB/FB ratio, but apparently for major leaguers only, sadly. Since they have added many awesome features the past year I keep hoping they'll add that for prospects, too.
I looked up Ross on milb.com and it showed his ratio as 3.41, not 4.14, is that a typo or did you use a different source?
Good morning, t ball - The opening sentences of the article were directed at minor league databases. I usually use FanGraphs for batted ball information on major leaguers. Regarding Ross' GB/FB ratio, I used Baseball America to pull the 4.14 ratio a couple of days ago. I guess that the 5 GB/6 FB game that he had last night dropped his season ratio to the 3.41 that you mention. I suspect that his problems last night stem from the extra pitches he made in his previous start that were made necessary by the 4 errors committed by his teammates.
David, do BA and milb.com figure GB/FB ratio the same way? It doesn't seem like a 5/6 night should pull the ratio down that much. I noticed firstinning.com has his GB rate as 63%, and that doesn't seem to matchup with a 3.41 rate, either, but I think firstinning might not separate out infield flies...
Sometimes it's frustrating trying to interpret these sites' stat data. I suspect, though, that in a few years, even most minor league teams will have more complete databases, and pitch f/x type stuff. I would think that most teams would see the benefit in that, I just hope they don't keep the info to themselves.
Terrific write-up, Dave. Very illuminating.