DMB News June 2001
Diamond Mind Email Newsletter
June 29, 2001
Written by Tom Tippett
Welcome to the third edition of the Diamond Mind email newsletter for the year 2001. Through these newsletters, we will try to keep you up to date on the latest product and technical information about the Diamond Mind Baseball game, related player disks, and our ongoing baseball research efforts. Back issues are available on our website.
Topics for this issue
Office hours in July
Tony Muser on Rey Sanchez
Sports Illustrated on first-pitch strikes
Extended streaks and slumps
Passing the torch, part 1
Stealing against poor-throwing catchers
Version 8 update
Past season update
Passing the torch, part 2
On the nature of love
Our offices will be closed for the holiday on Wednesday, July 4th. We'll be open for business the rest of that week during our normal hours of 9-5 eastern, Monday through Friday.
From Wednesday, July 11, through Friday, July 13, our technical staff will be out of the office to attend the national convention of the Society for American Baseball Research in Milwaukee. Unfortunately, this means that our ability to provide technical support during those three days will be severely limited. If you have an urgent need for support during that period, contact us at DiamondMnd@aol.com or 800-400-4803. If possible, we'll have someone call you back or reply to your email. If you have questions of a non-emergency nature, please contact us before or after those dates. We apologize in advance for any inconvenience this might cause.
In the June 6 issue of Baseball Weekly, Kansas City manager Tony Muser was quoted as saying this about Rey Sanchez: "He's having an All-Star season. He's as dependable a shortstop as anybody in baseball. He probably means two to three runs every day to us."
At the time, Sanchez had compiled a 21-game hitting streak that pushed his batting average over .300, but he remained one of the weakest offensive players in the league at his position. Among AL shortstops with enough plate appearances to qualify for the batting title, Sanchez ranked fourth-last in OPS (on-base plus slugging), mainly because he had drawn only 7 walks and had no homers.
Defensively, our analysis indicates that Sanchez has been the best shortstop in baseball in recent years, combining great range with steady hands. But the notion that he saves 2-3 runs a day is ridiculous. If Muser really said that, he must believe that his staff's ERA would have been over 7.00 (instead of their actual 5.22 mark) were it not for Sanchez.
In reality, the best shortstop in the league generally makes about 30-50 more outs per season than a league-average fielder. And because almost all of those extra outs would have been singles, the number of runs saved is only about 15 to 20. Give him credit for starting extra double plays and the number of runs saved could conceivably go as high as 30 to 35. Per season. Or about one run every 5 days.
Even if Muser was misquoted, he's probably wrong. If he really said 2-3 runs per series or per week, he's still overstating the defensive impact of a top shortstop. If he really said 2-3 runs per month, he's understating the case a little. And I'm sure he didn't say 2-3 runs per month, because that doesn't sound like high praise, does it?
In my view, Sanchez's excellent defense is enough to overcome his weak bat and make him a solid everyday player, but there's no way he's an All-star ahead of guys like Alex Rodriguez, Derek Jeter, Cristian Guzman, and Miguel Tejada.
I know a manager has to back his players, so it doesn't surprise me to hear him talk Sanchez up as an All-star candidate in a year when both Jeter and A-Rod have made a bunch of errors. And I've heard other baseball people claim that a defensive player can save 100 runs a season. But if Muser really said 2-3 runs per day, and if he really believed it, it ranks as one of the more absurd statements I've ever heard about defense.
In the June 4th issue of SI, Michael Bamberger explored the importance of first-pitch strikes. Unfortunately, he began the piece with a series of errors that undermined (in my eyes, at least) his credibility.
In his second paragraph, he wrote that "for most hitters, the best chance to drive the ball is on the first pitch of an atbat, when a pitcher is especially determined to work in the strike zone." Later in that paragraph, he tries to back up that claim by pointing out that "when putting the first pitch in play ... the average for all of baseball was .336 -- compared with an overall major league batting average of .270."
Bamberger appears not to understand the difference between overall batting averages and batting averages when the ball is put into play. Last year, major league hitters (including NL pitchers) had 45,246 hits in 167,290 atbats for a .270 average. So far, we're in agreement with Bamberger on this.
But those results break down into two categories. Hitters had zero hits in 31,356 atbats for a .000 average when striking out. And they had 45,246 hits in 135,934 atbats for a .333 average when they were able to put the ball in play. So a .336 average when putting the first pitch in play doesn't look so hot any more.
My impression is that quite a few baseball writers and broadcast personnel fail to understand and/or articulate the distinction between results ON a certain count and results when passing THROUGH a certain count.
A batter cannot strike out if he doesn't already have two strikes on him, so his results ON less-than-two-strike counts look pretty good. And all of his strikeouts are built into his results ON two-strike counts, so those results almost always look awful.
Here's how major league hitters (excluding NL pitchers) performed in the 2000 season ON the various counts:
AVG OBP SPC AB H 2B 3B HR RBI W SO
On 0-0 .341 .345 .552 22031 7508 1471 154 955 4099 0 0 On 0-1 .327 .335 .519 14405 4715 897 85 565 2317 0 0 On 0-2 .168 .178 .248 12016 2013 331 44 184 941 0 5448
On 1-0 .346 .346 .585 14500 5019 1020 117 737 2718 0 0 On 1-1 .329 .334 .524 15339 5052 1002 109 590 2448 0 0 On 1-2 .183 .189 .274 23025 4208 786 92 380 1969 0 9832
On 2-0 .369 .367 .675 5544 2045 451 45 386 1254 0 0 On 2-1 .343 .343 .565 10776 3696 752 65 503 1884 0 0 On 2-2 .198 .202 .309 22770 4501 885 94 487 2172 0 8883
On 3-0 .428 .957 .825 383 164 39 4 35 231 4859 0 On 3-1 .345 .689 .632 4824 1663 356 40 316 1112 5437 0 On 3-2 .237 .483 .393 16104 3817 773 95 518 2227 7701 5124
Among other things, this table shows that Bamberger was wrong about the first pitch providing "the best chance to drive the ball." In reality, compared with first-pitch results, hitters posted higher batting averages and slugging percentages any time they were ahead in the count, with the advantage growing the more they were ahead. That won't come as a shock to anyone who follows baseball.
(By the way, the results on 3-0 counts are skewed to the high side because only the best hitters are given the green light to swing at those pitches.)
If Bamberger really wanted to explore the importance of first pitch strikes, he would have focused on the results when passing THROUGH certain counts. That's because taking a first-pitch strike increases the chance that the hitter will eventually strike out, and these results include those now-more-likely strikeouts.
Here's how major league hitters (excluding NL pitchers) performed in the 2000 season when passing THROUGH the various counts:
AVG OBP SPC AB H 2B 3B HR RBI W SO Thru 0-0 .275 .350 .445 161632 44381 8759 944 5656 23353 17969 29255 Thru 0-1 .243 .289 .382 73632 17889 3476 371 2015 8752 4245 18964 Thru 0-2 .185 .217 .281 27658 5119 923 107 509 2476 917 11537
Thru 1-0 .288 .411 .480 66007 18997 3814 419 2686 10515 13734 10306 Thru 1-1 .252 .329 .403 63348 15978 3198 352 1881 8095 6970 15168 Thru 1-2 .195 .249 .304 41929 8197 1618 176 861 4047 2731 1677
Thru 2-0 .307 .539 .537 19748 6058 1253 129 1009 3663 10104 2571 Thru 2-1 .270 .414 .446 32498 8790 1738 188 1202 4771 7975 6600 Thru 2-2 .209 .315 .335 33641 7033 1393 163 837 3684 5108 12400
Thru 3-0 .308 .776 .546 3437 1060 235 26 176 777 7288 460 Thru 3-1 .293 .605 .512 10054 2947 621 66 484 1827 8029 1606 Thru 3-2 .237 .483 .393 16104 3817 773 95 518 2227 7701 5124
As you can see from this table, the first pitch does make a difference. The hitter's OPS (on-base plus slugging) rose from .795 for all appearances to .891 when the first pitch is a ball and fell to .671 when the first pitch was a strike. Those are very big differences, and more than enough to make Bamberger's point.
Bamberger also stated that pitchers are "especially determined to work in the strike zone" on the first pitch. That's not true either.
In recent years, approximately 62% of all pitches were recorded as strikes. But the strike percentage is only around 56% on the first pitch. As the pitcher gets further ahead in the count, he goes out of the strike zone more often in the hope that the hitter will chase a bad pitch. And as the pitcher falls behind in the count, he no longer has the option to waste pitches, and his strike rate rises. Contrary to what Bamberger wrote, pitchers "work in the strike zone" most often on 3-1 and 3-2 pitches.
Before leaving this topic, I want to comment on the nature of the data we are working with and the implications that has for the study of pitcher-batter behavior.
Our analysis is based on pitch-by-pitch data that we license from STATS, Inc. They record the result of every pitch using codes that indicate whether the pitch was a called ball, swinging strike, foul ball, foul bunt, put in play, and so on.
I can't begin to tell you how important this type of data is in our work as game designers. We've used it in the design of our computer manager by studying how managers factor the ball-strike count into their decisions about when to bunt, hit and run, steal, or swing away. Based on the results in the above tables, DMB increases the chances that the batter will hit for average and power when he's ahead in the count and reduces his effectiveness when he's behind. And we've used the pitch result data to adjust how often pitchers throw strikes and how often the batter swings at pitches on various counts.
But it's important to remember that the data are results-based. What do I mean by that? I mean that we know whether a pitch was put in play, fouled off, swung at and missed, taken for a ball, and so on. We don't know whether the pitch would have been called a ball or a strike if the batter didn't swing at it.
So when we say that pitchers throw more strikes when they are behind in the count, we're certain that's true, but we don't know exactly how many more. Because the batter's behavior changes with the count, too, and when the batter swings more often, even at pitches out of the strike zone, the pitcher gets credit for throwing more strikes.
So the count that produced the highest strike rate, 77%, is the 3-2 count. That's because the three balls give the pitcher an incentive to throw strikes and the two strikes give the batter an incentive to swing at anything close. On 3-1 counts, the strike rate was still quite high, at 70%, but not as high as on 3-2 counts because the batter takes a lot more close pitches on 3-1 counts, and some of those pitches are balls.
Getting back to the Bamberger article, it's true that first-pitch strikes are very important. It's unfortunate, though, that all of the claims he makes in his second paragraph are just plain wrong.
I'm an analyst first and a writer second, and I'm sure anyone on the SI staff could take just about anything I've written in the past four years and make it twice as good. And I'll continue to read SI because I enjoy good writing and want to stay on top of what the national media is saying about baseball. But this isn't the first time I've seen an SI writer make serious analytical errors, and I hope they're a little more careful in the future.
We recently posted a new article called "What a difference a month makes". It compares April and May performances to highlight some of the lengthy streaks and slumps we've seen so far in the 2001 season, and it contains links to three related articles that we posted in 1998. Those three articles take a deeper look at streaks and slumps, and their conclusions are still valid even if the data they contain is three years old.
Can you guess the identity of the following two players?
AB R H 2B 3B HR RBI BB SO SB CS AVG OBP SLG OPS A 341 68 120 17 6 3 34 13 17 26 6 .352 .381 .463 .844 B 332 51 112 16 4 4 34 32 18 19 7 .337 .396 .446 .842
See below for the answer.
Every so often, perhaps two or three times a year, I receive an email from someone who complains bitterly that a very good or excellent base stealer was thrown out by a poor catcher in a Diamond Mind Baseball game. I got one of those messages a few weeks ago, this time with reference to Todd Hundley.
I honestly believe there are a lot of misconceptions about how easy it is to steal on catchers who earn our Poor rating. Here are a few facts that I've dug up when looking into Hundley's real-life performance in the 2000 season:
- last year, Hundley caught 700 innings. In that amount of playing time, the average catcher faced 68 steal attempts, but Hundley was challenged 100 times. That high attempt rate was the main reason we gave him a Pr rating.
- the average catcher allowed 68 runners out of every 100 to steal successfully; with Hundley catching, 76 runners stole successfully, so the difference between Hundley and the average catcher was only 8 successful steals per 100 attempts.
Some DMB gamers look at a Pr throwing rating and assume that you can challenge him several times a game and be successful almost every time, especially with the top-rated base stealers. But last year, Hundley was challenged only 1.3 times per nine innings versus a league average of 0.9 times. Even with the advantage of having a poor thrower behind the plate, enemy runners were quite selective.
Part of the reason is that the pitcher has a lot to do with the ability to get a good jump and steal a base, and I believe that some DMB gamers don't put enough emphasis on the pitcher's hold rating when making base-stealing decisions. So if a guy like Hundley throws out an Ex base-stealer, my first question would be, "what was the pitcher's Hold rating?"
I also compiled some data from the 2000 season for several of the base-stealers whom we rated Ex for stealing (their Jump ratings vary), focusing on their opportunities and attempts with Hundley behind the plate:
CBiggio 8 opportunities, 0 attempts ABrown 3 opportunities, 0 attempts MDarr 9 opportunities, 1 attempt, 1 SB CFloyd 7 opportunities, 1 attempt, 1 SB TGoodwin 2 opportunities, 1 attempt, 1 SB WGuerrero 2 opportunities, 0 attempts LHarris 4 opportunities, 1 attempt, 1 SB DJackson 10 opportunities, 2 attempts, 2 SB
I could work my way through the rest of the alphabet, but that wouldn't change the basic point. It's true that all of these guys were successful when they tried to steal, but it's also true that they made only 6 attempts in 44 opportunities. There are, of course, a lot of reasons why a runner might stay put in a particular situation -- the team could be ahead or behind by a large margin, the next hitter might be walked if first base is opened up, or the pitcher has done a good job holding the runner. But I believe it's also true that these runners know that there's a real danger of being thrown out, even by a guy like Hundley, if they don't pick their spots well enough.
Some DMB managers are nowhere near as selective as this when they have an Ex stealer facing a Pr throwing catcher. They run more often and, as a result, get thrown out more often than they would expect.
For what it's worth, we just finished simulating the 2000 season five times, using the real-life starting lineups to ensure that playing time was comparable, and came up with the following results for Hundley (average of five simulated seasons):
Attempts Success rate Real-life 100 76% Simulations (avg) 98 76%
Both the real-life figures and the simulation results include CS that were the result of the pitcher throwing over to first and the runner breaking for second.
We expect to be able to release a version 8D patch in mid-July that includes some minor bug fixes and the ability to sit back and watch the commentary for one or more innings without hitting a key or clicking with the mouse before each play.
At this time, the development work on the so-called "passive play-by-play" feature has been completed. Although it appears to be working perfectly, we need a couple of weeks for internal testing and field testing to ensure that we didn't accidentally mess up any of the other play modes in the process of adding this one.
After we complete our testing and prepare the patch for delivery, we will notify version 8 owners via email and our web site, and we'll document the changes that are included in that patch.
We've been working very hard on the 1978 through 1981 seasons, and I'm happy to report that most of our research is complete. We still have plenty of work to do -- loading the stats and ratings, setting up manager profiles, and testing, testing, testing -- so it will be a few weeks before they are ready to go.
In our last newsletter, we said that we thought we'd have these seasons ready by the middle of the season, which is only a week or two away. Since then, we expanded the scope of this project to include the compilation of real-life transactions and game-by-game starting lineups. This is labor-intensive work that has added about 4 weeks to our schedule, pushing the likely ship date back into August, but we're convinced it will make for a better product.
By the way, this is the first time we've included transactions and lineups for past seasons, and we're planning to add transactions and lineups for other Deluxe Past Seasons later this year. We'll give you more information on the timing and upgrade pricing (including free updates for recent buyers of those seasons) when we have some of those updated seasons available.
Our plan is to start taking orders for the 1978-1981 seasons when all four seasons are ready to ship. As soon as we reach that point, we'll let you know via this newsletter and our web site.
A couple of pages ago, I asked if you could guess the identity of the following two players:
AB R H 2B 3B HR RBI BB SO SB CS AVG OBP SLG OPS A 341 68 120 17 6 3 34 13 17 26 6 .352 .381 .463 .844 B 332 51 112 16 4 4 34 32 18 19 7 .337 .396 .446 .842
Suppose we told you that both players bat left-handed, both play right field for West Coast teams, and both are superior defensive players.
You may have guessed by now that player A is Ichiro Suzuki of the Mariners. Those are his stats through June 27th, so we don't really know if the league will catch up with him or whether he'll sustain or improve upon that pace going forward. But if he does maintain that pace, it's interesting to see how similar his stats look when compared to those of . . .
. . . Tony Gwynn in his twenties. Player B is a composite of Gwynn's first six seasons as an everyday player, from age 24 to 29, prorated to match Suzuki's playing time. I picked those years because they straddle Suzuki's current age of 27 and I figured that would provide the best basis for comparison.
So far, Suzuki's average is a little higher, but Gwynn walked more often during that part of his career. Ichiro is on pace for 55 steals in 68 attempts this year; in Gwynn's best season (1987) he stole 56 in 68 tries at age 27. Gwynn won five Gold Gloves in six years from age 26 through 32; Suzuki is being talked about in a way that suggests he'll pick one up this October.
During the off-season, it was popular to compare Suzuki's skills to those of Johnny Damon, but with Gwynn announcing that he'd be calling it a (Hall of Fame) career after this season, I suddenly realized that the young Gwynn might make for a better match. So as Gwynn prepares to exit stage right and begin working on the acceptance speech he'll be giving in Cooperstown five years from now, Ichiro is providing us with a living reminder of what it was like to watch Tony fifteen years ago.
Someone recently sent me an email with comments made by children when asked about love. Here's my favorite:
"Love is the most important thing in the world, but baseball is pretty good too." - Greg, 8