An Early Season Reality Check
Tom Tippett
May 1, 1998
Sigh. It's a month into the season, and your favorite team is limping along a few games under .500. And your fantasy team isn't shaping up the way you planned before the draft. Have you misjudged the talent, or is this just a run of bad luck? Just how much can we read into the results from the first month?
To examine these questions, I'm going to take a look back at the best and worst April performances from a year ago, see how these players responded the rest of the way, and explore the luck factor in some depth.
And I'm also going to consider whether April results can be heavily influenced by the strength of the opponents each player or team has faced in the early going. So I'll take a look at the 1997 Cubs (who led off with 14 straight losses) and the 1996 Red Sox (who started 8-19). And I'll see if I can find any clues about the 1998 races.
1997 Hitters
There were 231 players who had at least 50 atbats in April last year, and 201 of them went on to get at least 250 atbats for the season as a whole. Of the thirty who didn't make the 250 mark, quite a few were injured (Bobby Abreu, John Jaha, Marc Newfield, Brian Jordan, Kevin Elster), some were sent down (Todd Walker) or released (Ruben Sierra, Kevin Mitchell), and some were role players who just happened to pick up a little extra playing time in the early going.
A large number of hitters finished 1997 with a batting average substantially different from the one they left April with. Here's the distribution:
Change in Average #players Examples Up by 141+ points 1 John Valentin Up by 101-140 pts 1 Tom Goodwin Up by 61-100 pts 19 DeShields, Burks, McLemore, Zeile Up by 21- 60 pts 50 many Less than 20 pts 51 many Down by 21- 60 pts 55 many Down by 61-100 pts 21 Blauser, BAnderson, LWalker, Delgado Down by 101-140 pts 3 Tucker, DSanders, Gagne
Overall, 96 players improved and 105 were lower by the end of the year. It's no surprise that there were more players who fell back than who gained ground, because some of the guys who started slowly lost their jobs before they had a chance to recover. Most managers had the patience to give their slow starters ample opportunity to rebound, especially if they were dealing with an established player in mid-career, but the prospects and older players usually don't get the same chance.
Extended slumps are pretty common throughout the baseball season, but the ones that are most visible are those that occur at the start of the season. And there were plenty to choose from in 1997. John Valentin batted .160 in 94 April atbats before roaring back to finish at .306. Tom Goodwin lifted his average from .143 (in 84 AB) to .260. Ellis Burks batted .195 in April en route to a .290 finish. Albert Belle batted only .206 in his first 102 White Sox atbats before rallying to end the year at .274. Todd Zeile started at .175 in 80 atbats and ended at .268. Bret Boone's .223 average was hardly impressive, but it was 90 points better than his .133 April mark.
Twenty-one players posted an April batting average under .200 in at least 50 atbats. Six of them were denied the chance to get 250 atbats. The fifteen who were given a shot at redemption responded with an average gain of 79 points. And because their final averages include those awful April numbers, that means these players batted about 95 points better from May to September.
On the other end of the scale are guys like Michael Tucker, who plummeted from a .418 April mark to end the season at .283; Deion Sanders, who lost 110 points off his gaudy .383 start; and Greg Gagne, who finished at .251 after hitting .356 in the opening month.
Twenty players went into May with averages of .350 or higher. Only Eric Davis (who battled cancer for much of the season) failed to make the 250 AB mark for the season. The other nineteen players saw their averages drop by 70 points per man before the year ended. By the way, Tony Gwynn was the only player in this group who end the year with a higher average (.372) than he put up in April (.367).
So if you're wondering what to do about a slow starter in a 1998 fantasy league, start by asking how safe his job seems to be. There's no point in holding onto a player who's heading for the minors or the waiver wire. If you think his job is safe, be patient. There's a very good chance you'll be happy.
And if you've been lucky enough to get a career month out of an ordinary player, beware. Barring injury, he will play a lot, but the odds favor a return to earth. If someone in your league covets one of these players, you might do well to trade him.
A Brief Exercise in Probability
Every week, the newspapers give us lists of the ten best and worst hitters for the past week, and the averages are typically over .400 on the good list and under .100 on the bad one. Various books and online services provide month-by-month batting stats, and it's not unusual to see some pretty big variations. So most fans have an intuitive awareness that baseball is full of streaks and slumps.
In 1997, for example, Jeff King's monthly averages were .314, .170, .355 (with 10 homers), .146, .215, and .262. Imagine how frustrating it was for fantasy leaguers who owned this guy in July and August, when he batted .179 with only 5 homers in 196 atbats. Could his skills have risen and fallen this much, this often, in a six-month period? Is it really possible that these variations are due to luck?
Suppose you had a special coin, one that comes up "HIT" 26% of the time and "OUT" 74% of the time. Tossing this coin a million times would produce something like 260,000 HITs, for a simulated batting average of .260. Of course, you wouldn't always get exactly a .260 average. Sometimes it would be higher, sometimes lower. But if you did this often enough, the long-term average would be .260.
Now suppose you tossed this coin only 85 times, to simulate the number of atbats a typical player gets in a month. The most likely result would be 85 * .260 = 22 HITs. But you won't always get exactly 22 HITs. And this is the basis of our little experiment. We wrote a very simple computer program that simulates the flipping of this coin, and we asked it to generate 5000 different sequences of 85 atbats. Here's what we found:
- the simulated monthly average was between .240 and .280 about 30% of the time.
- the simulated monthly average was between .200 and .320 about 82% of the time
In other words, the monthly batting average was more than 60 points removed from the player's long-term batting average about 18% of the time, or about once every 5 months. So King's performance was unusual, because he was more than 60 points away from his final average in 4 of 6 months last year.
By the way, it doesn't matter that I chose a .260 hitter for this example. The important point is the variation, and that's about the same whether we're modeling a .260 hitter or a .300 hitter or a .235 hitter. We're still going to see the monthly averages stay within 20 points of the player's true ability about 30% of the time.
But there's something even more interesting here. Suppose we apply the results of our simple coin-tossing program to a population of 201 players. How many of those players could be expected to have an April that was 20 points, or 60 points, or 100 points different from their true ability? And how do these predicted numbers compare to what we actually saw in 1997? Here are the results:
Actual Predicted Change in Average #players #players Up by 141+ points 1 0 Up by 101-140 pts 1 3 Up by 61-100 pts 19 13 Up by 21- 60 pts 50 52 Less than 20 pts 51 59 Dn by 21- 60 pts 55 54 Dn by 61-100 pts 21 15 Dn by 101-140 pts 3 4 Dn by 141+ points 0 1
To explain these numbers, let's look at the fourth line in this table. In real-life 1997, there were 50 players who raised their average by 21-60 points from April 30th through the end of the year. Another way to say the same thing is that these 50 players had April averages that were 21-60 points below their full-season averages. The last column indicates that our coin-tossing program predicts that a sample of 201 players would typically produce 52 with a monthly batting average that is 21-60 points below their real long-term average.
As I noted in the last pair of bullets, if luck is the only factor creating a variation in batting averages over an 85-atbat stretch, we'd expect to see 30% of the players within 20 points, and 82% within 60 points, of their final averages. In the real 1997 season, 25% of the April averages were within 20 points, and 78% were within 60 points, of their full-season counterparts.
There was actually a little more variance in the real-life numbers than can be explained by chance, but the differences are not large, and it's reasonable to conclude that luck is the major factor in month-to-month variations in batting average.
1997 Starting Pitchers
Of the 99 pitchers with at least 25 innings in April, 79 proceeded to rack up at least 125 innings by year end, with the other 20 succumbing to injury (Langston, Osvaldo Fernandez) or losing their spot in the rotation (Dennis Martinez, Scott Sanders, Fernando Valenzuela, Steve Avery). Here's what the 78 qualifiers did:
Change in ERA #players Examples Down by 1.75+ 5 Bullinger, Burba, Trachsel Down by 1.26 to 1.75 7 Drabek, Hampton, Hershiser Down by .76 to 1.25 5 Neagle, Cooke, Mussina Down by .26 to .75 8 Hentgen, AlBenes, Nagy Less than .25 9 Mlicki, Cone, Holt Up by .26 to .75 13 Nomo, KBrown, Erickson, Rueter Up by .76 to 1.25 11 Lieber, Dickson, Maddux Up by 1.26 to 1.75 10 Cordova, Prieto, Burkett, Glavine Up by 1.75+ 11 Witt, Karsay, MLeiter, Belcher
Managers were less patient with starting pitchers with bad Aprils than they were with position players. As a result, 48 of these pitchers saw their ERAs rise and only 31 improved their record by year end. Quite a few of the guys who started poorly were banished from the rotation and didn't get enough innings to make our list.
Of the twelve pitchers with an April ERA of six or higher, the eight who hung around long enough to get 125 innings lowered their ERAs by an average of 2.62 runs.
All nine of the pitchers who entered May with ERAs under 2.20 stayed healthy enough to make the 125 inning cutoff. (In fact, they averaged over 200 innings each.) Their year-end ERAs were an average of 1.49 runs higher than their April marks.
1997 Closers
There were 42 men who pitched in April and were given enough shots at the closer role to compile at least 5 saves during the season. (A few more closers, including Karchner and Escobar, weren't on the roster in April.) A typical April workload for this group was 12 innings, so their ERAs were naturally quite volatile. Even so, I was surprised to see that 13 (almost a third) entered May with an ERA over five, and several prominent closers were in the sixes and sevens.
Managers were very patient with this group, and were rewarded in almost every case. Robb Nen ended April at 7.36 and was down to 3.89 by year end. Aguilera went from 7.15 to 3.82, Hoffman from 6.48 to 2.66, Mesa from 6.43 to 2.40, Roberto Hernandez from 5.19 to 2.45. Troy Percival went from 20.25 to 3.46, though his April record was based on only 2-2/3 innings. Even Todd Worrell, who never really pitched well, lowered his ERA by almost a run, from 6.17 to 5.28. Of the closers who ended April at 4.00 or higher, only Norm Charlton got worse. The average improvement among the group of 4.00-plus closers, even including Charlton, was two runs per game.
Two things are clear. First, today's closers almost never lose their jobs no matter how bad their numbers look in the early going. They might be moved to a setup role for a while, but their almost always given at least one more shot at closing. Second, because closers get so few innings, one or two bad outings can send a closer's early-season ERA skyrocketing. So don't panic if an established closer has an April ERA of six and a half. If he's healthy, it probably won't last.
It's the Other Guy's Fault!!
I've made the observation that luck can explain much of the difference between April and full-year stats. So we need to be careful when we try to interpret April results. It would be easy to make the mistake of attributing something to a real change in ability (a new batting stance, a new pitch, a better attitude, clean bill of health) when the player has merely been lucky.
But there are two real factors that seem to be overlooked -- the ability of the guys you've been playing against, and to a lesser degree, the nature of the parks you've been playing in. Over a full season, these things average out for the most part. But there are schedule imbalances that can play a meaningful role in the first few weeks of the season.
Everyone who follows pro football knows the NFL schedule isn't balanced. Division winners play a very tough schedule the next year, while the last-place teams are rewarded with extra games against the other cellar-dwellars. It's not uncommon for a team to parlay a fifth-place schedule into a near-.500 record that puts them in the playoff hunt the next year. As a result, assessing the strength of the opposition is a major part of football analysis.
But it seems to me that there's much less talk in baseball about the other team. Yes, it's common to tip the cap to a starting pitcher that's just shut us down in one game. But, more often, if we are in the midst of a losing streak, it's because we're making bad pitches, we're not hitting in the clutch, we're kicking the ball around, or we're not playing fundamental baseball. Similarly, when things are going well, it seems as if we take all the credit. We're seeing the ball well, we're tightening up our defense, and we're making great pitches in critical situations.
Doesn't anybody think that our pitching might look good because we just played two of the weakest hitting teams in the league and had a third series in a great pitchers' park? Or that our batting prowess might have something to do with having back-to-back series in Wrigley and Coors? Of course, when a team is winning, it's unreasonable to expect players and managers to talk much about the quality of the opposition. Why give away the credit when you can use the winning atmosphere to build confidence in your own club? Why disparage the other team and give them additional motivation for their next visit to your park?
So it's ok that the players and managers don't talk this way. But I'd like to see the media spend a little more time looking at the qualities of the opposing teams when trying to explain a run of wins or losses. I'll give you a couple of examples.
The 1997 Cubs
Last year at this time, the Chicago Cubs were 4-18, having started the season with 14 straight losses. Their team batting average was a league-worst .214 and they were averaging only 3.2 runs per game.
But they opened the season with 10 games against Florida and Atlanta. They faced Kevin Brown, Al Leiter twice, Alex Fernandez twice, Denny Neagle, Greg Maddux twice, John Smoltz, and Tom Glavine. It's no wonder they weren't hitting! The worst of those pitchers allowed opposing batters to hit .242 last year.
It got better, but not by much. In their next 12 games, they faced seven more pitchers who held opposing hitters at or below the league batting average. (Technically, it was six, but one was Roger Bailey, and he would have been below the league average if he wasn’t pitching in Coors half the time. And the Cubs played the Rockies in Wrigley, so Bailey's road batting average is more relevant anyway.) They scored about twice as many runs per game in this stretch as they did in the first ten.
Don't get me wrong. I'm not saying they were a good team. Just that they were made to look worse by a steady diet of very good pitchers. By the end of the year, their team batting average had rebounded to the league average of .263.
The 1996 Red Sox
The 1996 Red Sox began the season with a terrible 8-19 record. They weren't doing anything well, but the biggest surprise was the lack of offense. Nobody expected them to pitch well, but they were expected to score a lot more runs. But it wasn't all their fault.
To assess the quality of pitching they had faced, we took all of the starting pitchers in the AL and grouped them into three tiers. The top tier consisted of staff aces like Randy Johnson and Mike Mussina and the better #2 starters. The middle tier was made up of mid-rotation starters. The bottom tier included everyone else who had started a game that season. We put about the same number of pitchers in each group.
We then examined the Red Sox performance over the first 27 games. Because the groups are of equal size, we would have expected the Sox to face about nine from each category in a 27-game stretch, with a small bias toward the top group because the #5 starter doesn’t pitch as much early in the year. But the Sox had faced 14 top-tier pitchers, 8 from the middle group, and 5 from the bottom group, and their record reflected it -- 2-12 and 3.7 runs per game against the top group, 2-6 and 4.5 runs per game against the middle group, and 7.4 runs per game and a 4-1 record against the bottom group.
At the time, we predicted that they would make some noise in the wild card race and be among the offensive leaders once the schedule started to balance out. They won 85 games, staying in playoff contention until the last four days of the season. And they scored 928 runs, or 5.72 per game, to finish tied for fourth in the league in scoring.
Looking at 1998
Early in the season, I like to scan the team-versus-team grid (USA Today publishes them every Tuesday and Wednesday) to see who's played who so far. When I do this, I'm looking for teams that have played an unusually high or low number of games:
- against the better teams
- against the better hitting teams
- against the better pitching teams
- in parks that favor either hitters or pitcher
Most teams don't stand out in any of these ways. But some do. And you can often find clues about the future performance of these teams. Here are a few things I noticed in this week's listings:
- in the AL, through April 26th, neither the Angels nor the Tigers had played a single game against a team with a losing record. This is a little misleading, because they've both played Tampa Bay several times, and while the Devil Rays stood at .500 on the cutoff date, they're not likely to remain there for the full season. Nevertheless, both teams have faced tough schedules. The easiest? Minnesota had played only two games, the Yanks three, against teams that were at .500 or better.
- in the NL, through April 27th, Montreal and Pittsburgh led the league in games against .500+ teams with 22 and 19, respectively. Montreal is a bad team, and Pittsburgh is probably destined for last place in their division, but both have played a tough schedule so far. Atlanta has played only two games against teams at or above .500, so their strong start is a little suspect.
- only six teams (Red Sox, Rangers, Angels, Mets, Brewers, and Padres) have winning records against .500+ opponents.
- don't be surprised if the Reds start to make up some ground in the NL Central. Two-thirds of their games have been against clubs that are .500 or better. Barry Larkin's back, and he's likely to get better. They're 0-7 in one-run games, and while part of this is due to a bullpen problem that might persist, some of it is bad luck that should reverse itself. True, they traded away their #1 starter for a player, Sean Casey, who's been out with a freak injury, so they've been weakened. But I still think a lot of people, including some folks in the Reds front office, are underestimating this team.
- Aaron Sele leads the majors with a 5-0 record, and his 2.00 ERA is nothing to sneeze at. But his five starts have come against the White Sox (twice), Tigers, Devil Rays, and Royals. I'd like to see him do well, partly because I think he got a bum rap here in Boston, but I'm staying off his bandwagon until I see how he does against teams that can really pound the ball.
- there are some unfamiliar faces near the top of the AL team pitching rankings. The Red Sox are third, and with six games each against Cleveland and Seattle, who rank 2nd and 3rd in scoring, they've earned their high standing. The Twins lead the league despite facing Seattle seven times already. Sure, they've also had five with the Blue Jays, but the average strength of their opponents is only a little below the league average.
- Anaheim's stadium might turn out to be a very good pitchers' park this year. In 15 home games, the Angels and their opponents scored only 4.03 runs per game, more than a run less than the league average. Scoring is noticeably higher in their road games. And the visiting teams -- Yanks, Indians, Red Sox, Devil Rays, Orioles -- are among the better offenses in the league. It's awfully early, and it may be that the Angels attack isn't as good as we thought, or their pitching is better than we thought. But if it is a very good pitchers' park, we'll have to remember this as we evaluate their pitchers and hitters in the coming weeks.
Conclusions
When you look over the results for the first month of the 1998 season, keep these things in mind:
- Long streaks and slumps are not unusual, so don't put too much weight on April stats, especially for established players who are likely to keep their jobs.
- The slow starters are likely to bounce back in a big way if they're given a chance. But before you blindly follow this advice and hang on to every slow starter on your fantasy roster, keep in mind that some of them will lose their jobs and the improvement for others might not be enough. If a starting pitcher has an April ERA of 7.25, a 2-run improvement still doesn't make him a particularly valuable player.
- The fast starters will get to play a lot, but don't expect them to keep it up. Fast and slow are relative terms of course. A .360 average is a very fast start for Mickey Morandini, but it's normal for Tony Gwynn.
- Be especially cautious about closers, for their lack of innings can lead to highly-volatile ERAs that disguise how well they're really pitching.
- If a team has surprised you so far in 1998, look for reasons. If they're not hitting, scan the boxscores to see if they've been facing more aces than they normally would. If they're pitching well, make sure they're not just feasting on the weakest hitting teams in the league. If their win/loss record is a surprise, take a quick look at the team-versus-team grid to see how many .500+ teams they've faced so far. Often, their success/failure has more to do with who they’re playing than their own work, and they'll revert to form once the schedule evens out.
Copyright © 1998. Diamond Mind, Inc. All rights reserved.
