Evaluating Players and Statistics - The Key to Winning Roto
So, you want to win your Rotisserie league this year? To win, you will need good off-season preparation, good pre-draft preparation, good drafting strategy, a cool head during mid-season transactions, and some luck. Today's column will start the process of evaluating players - figuring out which players are "better" than others - from a purely roto standpoint.
Rotisserie Stats vs. Real Life Performance
The most important thing that you must do before you begin to evaluate talent for purposes of a Rotisserie league is to forget everything that you know about real life baseball and about the players. You must begin on a clean slate. You may consider a player to be a solid third baseman in some absolute real-life sense -- he may be a team leader, a clutch hitter, and a great defensive player. None of those qualities is relevant to Rotisserie baseball.
It is this aspect of Rotisserie baseball to which the self-proclaimed "purists" object. Rotisserie baseball is not "real." Of course it is not real. Get that notion out of your head! It is a game. In order to win the game, you have to have an appreciation of the rules and an understanding of the strategy. Then you have to execute that strategy well. That doesn't mean that playing Rotisserie baseball is divorced from being a baseball "FAN" or a fan of your favorite team. It simply means that, for the duration of the draft, and during any trade negotiations, you have to think in terms slightly different terms from an actual baseball fan.
An example may be helpful. Suppose you have a hand of five playing cards containing four aces and a King. Do you have a "good" hand? You can confidently say "probably" -- but you don't yet know what game we are playing. If it is Hearts, you do not hold a good hand, but in most card games your cards are great. The particular rules of the game you are playing determine the value of your cards.
It is so in Rotisserie baseball. If your pitching staff includes Pedro Martinez, Kevin Brown, Al Leiter, Mariano Rivera, and Rob Nen, you have got a great staff whether you are playing actual real-world baseball or Rotisserie baseball. Like the hand of playing cards above, this staff is probably "good" no matter what the rules of the game.
Now, suppose you have a hand of five cards with three jacks and two fives. If you are playing poker, you have a winner. If you are playing bridge, you may not. As the rules of the game change, so does your decision making process. In Rotisserie baseball, you may have Rick Reed, Jesus Sanchez, A.J. Burnett, Todd Jones, and Keith Foulke. Is that a good staff? That depends on what statistical categories you are looking at. You probably have an opinion about whether A.J. is a "good" pitcher or not; but do you have an appreciation of how much he will hurt your team if WHIP is one of your pitching categories?
We begin, then, with a clean slate. No player is necessarily better than any other until proven so by the stats. With that in mind, the question is: What is the best way of determining the value of a player for Rotisserie purposes? In evaluating the value of a major league player, you must determine how much that player will help your team in the statistical categories your league uses. A player who leads the league in doubles is not necessarily a valuable player if your league does not have a category that includes doubles. A more valuable player would be one who hits singles, and then steals bases.
So, you have two main data points: First, the statistical categories that count in your league, and Second the performance of each player in the upcoming season. Now, if the game were based on the players' performance LAST season, it would be a whole lot easier, since you would know in advance what the numbers are. Of course, that's part of the fun. You need to project what statistics players are going to put up this year. You can base that on last year, or an average of the last two or three years, or some weighted formula, or you can get "projected" numbers from a variety of published sources and web sites. You may try to figure out who will get extra playing time this year, and whose time will be cut back by the acquisition of another player. Who moved to a hitters' park, or to a pitchers' park? Who is in their free agent walk year? That's the fun part.
A few general rules, however, to keep in mind. First, players (especially hitters) tend to perform at or near their lifetime averages after a few years in the big leagues. If someone is a lifetime .250 hitter, he is probably not going to hit .280 this year. He might hit .265 or .235, but the odds of him performing outside that range is slight. Similarly, someone who has a lifetime average of 20 homers a year may hit 25, but by the same token is not likely to hit less than 15 unless injured. Even injuries tend to follow patters. If a player has a lifetime average of 116 games per season, you can expect that player to be hurt for some period of time, and his lifetime average numbers reflect that average number of injury days. So, if you have statistical sources available, look at a lifetime average for any player with more than 3 years in the league as a pretty good starting point.
Lifetime averages alone, however, are only part of the story. Young power hitters tend to increase their production as they mature, especially if they grow into the 3, 4, or 5 spot in the batting order. Aging power hitters tend to fall off quickly at the end of their careers. A particularly good year may indicate a batter who has changed his stance, while a precipitous fall-off in stolen bases may indicate a serious injury that will affect the player's stolen base totals for years to come. So, know who you are dealing with. Perhaps take last year's stats and compare them to lifetime stats - if last year was better than the player's lifetime average, then he's trending up, and vice versa.
Second, as a general rule, hitters are more predictable than pitchers. A hitter is much more likely to perform this year at roughly the same level as last year, while pitchers (particularly starters) can have wild swings of fortune and performance from one year to the next. With that in mind, it may be a good idea to focus on only the top flight pitchers (who tend to be more reliable) and proven hitters in the early rounds of your draft, leaving those unpredictable mid-range starters for later.
Absolute valuation of projected performance -- the Diamond formula
No matter what method you use to project next year's stats, you'll need a method for comparing players against each other. The Diamond formula for determining the absolute value of a player for Rotisserie purposes is a secret. I will tell you, however, how it works. The philosophy of the Diamond formula is to evaluate talent based on the over-all stats, rather than looking only at individual statistical categories. A player who steals 60 bases may win you that category, but if the same player bats .265 with little power, he may not be worth the investment. The Diamond system tends to value good over-all players rather than players who are great in a few categories.
For each statistical category, I assign an arbitrary level of performance as a value of zero. This is intended to reflect the "average" performance. This is not intended to be the average major leaguer's performance, but rather the "average" performance that you expect from your Rotisserie players. If the average is assigned at .275, then a player whose 2001 projection in batting average is .275 receives a score of zero. For every batting average point above .275, the player earns one point. For every point below .275 the player earns a negative point. Thus, a player whose projection is .290 receives 15 absolute Diamond points in the category of batting average. A player whose projection is .240 receives a Diamond score of -35. Categories for pitchers work the same way, with an "average" value assigned for most categories, and points earned based on projecting better or worse than the average.
For each hitting category, the maximum value is 50, and the worst score is negative 50. Thus, the formula tends to flatten out those players who have one category where they are far above the average, while boosting players who are consistently above average in all categories.
Thus, after performing the Diamond analysis, you have a projection of the probable 2001 performance of each player, and an absolute Diamond value for those projections. Based on this formula, if you assume that 2001 performance will be exactly the same as 2000 performance, you can say that Alex Rodriguez has a higher absolute value for than Ivan Rodriguez. But, should you draft A-Rod first, or spend more money on him in the draft? What about comparing a pitcher to a hitter? If the formulas for hitters and pitchers are different, how do you compare them? Well, there is a way . . .
Relative valuation of players by position, the Z-score
The absolute value of a player is only half the equation when evaluating Rotisserie talent. Equally important is determining whether the shortstop with the highest absolute value among shortstops is a better pick than the outfielder whose absolute value is higher, but where there are more outfielders with very high values than there are shortstops with very high values. Thus, knowing who to draft depends on your drafting strategy and also on the relative values of players whom you could select now versus players whom you may be able to select later. In other words, is Barry Bonds so much better than Andrew Jones that you would rather take him and pass on Derek Jeter, or is Jeter so much better than the next best shortstop that you can live with a lesser outfielder knowing that you got one of the top shortstops?
The way to unscramble the statistical puzzle lies in a statistical vehicle called a Z-score. The Z-score is based on a comparison of an actual number against the average of a group, then dividing by the "standard deviation" of all scores. The Z-score will assign the average score a value of zero. The Z-score then evaluates the standard deviation of the group of numbers; that is, by how much does each number deviate from the average? This number is often reported as a "percentile." A score in the 67th percentile is 1 standard deviation above the mean (average) in a perfect bell-shaped curve. A score in the 92nd percentile is 2 standard deviations above the mean. A score in the 99th percentile (usually the highest score) is 3 standard deviations above the mean. Anything more than 3 standard deviations above the mean is basically off the charts. A raw Z-score simply reports the number of standard deviations above or below the mean, without reducing that number to a percentile. A Z-score of zero is the average. A Z-score of 1.5 means 1.5 standard deviations above the mean.
It is not necessary to understand the formula from which the Z-score is derived. It is enough to understand its significance. A player whose Z-score is 1.27 is .27 standard deviations better than a player whose Z-score is 1.00. By examining the difference in Z-scores, you can see how much better one player is than the next best player within a group (e.g., by position). So, if you evaluate each position, and compare shortstops to shortstops, and outfielders to outfielders, you get a pretty good idea of not only who is better than whom, but by how much as compared to the "average" player at each position, which is the only way to evaluate talent since it is necessary to fill all your positions.
So, who is the best player in the draft?
It may come as no surprise that, using 2000 stats as the basis for 2001 comparisons, and using a standard 5x5 roto format, the best player in roto ball is: Pedro Martinez. Pedro is so far above the rest of the pitchers in the league that his Z-score is an astounding 4.64 - which essentially means that he is more than 4 times better than the "average" starter. Randy Johnson, with a Z-score of 3.65 is not far behind, although Pedro is still a full standard deviation better than the Big Unit. (By contrast, Kevin Brown at 2.88 is the next best, followed by Greg Maddux at 2.65 - a much closer call.)
The best hitter? Well, the highest raw Diamond score goes to Johnny Damon at 192 points. But, Damon's Z-score is 2.23. Now that's very good, meaning that he is more than 2 times better than the average, placing him between the 93rd and 99th percentile among all outfielders. But the highest Z-score in the league goes to: Ivan Rodriguez at 2.75. (This is based on a projection of Rodriguez over 500 at bats, and factors out last year's injury.) In fact, the top three Z-scores belong to catchers - Rodriguez, Mike Piazza, and Jason Kendall. Their scores are so high because they are so much better than the average catcher. If you don't get Damon, and settle for Andrew Jones later in the draft, the difference will be 41 Diamond points, and a drop of .52 in Z-score. But if you miss out on one of the big three catchers, you may end up with someone whose Diamond score is in negative numbers! That's the trick - choosing value based on performance relative to the other players at that position.
Simple, right? Well, maybe not so simple. But useful. Below are the top hitters and pitchers, based on a 5x5 roto format in a mixed league. In the weeks immediately before the start of the season, I'll publish all my stats, broken down by position, for all the players who had useable stats in 2000. For now, here are the top picks, meaning those players whose Z-scores are at least 1.5:
Ivan Rodriguez 134 2.75 Jason Kendall 133 2.74 Mike Piazza 123 2.52 Alex Rodriguez 180 2.47 Johnny Damon 192 2.23 Jeff Bagwell 186 2.22 Jeff Kent 156 2.20 Luis Castillo 155 2.18 Todd Helton 183 2.18 Chipper Jones 134 2.02 Roberto Alomar 145 2.02 Charles Johnson 87 1.98 Darin Erstad 164 1.90 Barry Bonds 161 1.83 Richard Hidalgo 160 1.82 Derek Jeter 131 1.82 Troy Glaus 119 1.81 Frank Thomas 156 1.77 Carlos Delgado 155 1.77 Nomar Garciaparra 124 1.73 Edgardo Alfonzo 126 1.72 Jorge Posada 71 1.72 Andrew Jones 151 1.71 Jeff Cirillo 112 1.71 Jason Giambi 150 1.70 Vlad Geurrero 150 1.69 Gary Sheffield 149 1.68 Jim Edmonds 149 1.68 Sammy Sosa 148 1.67 Jose Vidro 121 1.64 Eric Young 120 1.62 Travis Fryman 104 1.60 Maglio Ordonez 140 1.57 Brian Giles 137 1.53 Bobby Abreu 137 1.53 Shannon Stewart 136 1.52 Pedro Martinez
955
4.64
Randy Johnson 785 3.65 Rob Nen 472 2.94 Kevin Brown 651 2.88 Greg Maddux 611 2.65 Armando Benitez 398 2.27 Al Leiter 515 2.09 Tom Glavine 506 2.03 Chan Ho Park 504 2.03 Trevor Hoffman 371 2.03 Darryl Kile 491 1.95 Jeff D’Amico 449 1.7 Derek Lowe 335 1.7 Mariano Rivera 325 1.61 Keith Foulke 323 1.59 Mike Mussina 425 1.57 Billy Koch 318 1.55 Tim Hudson 418 1.53 Agree? Disagree? Please send me an email and tell me about it. (Remember, your email may be published, so let me know if you want your identity withheld.)
Roto Notes
Troy O'Leary - OF - BOS - He has no position to play, and a big contract. A bad combination. Look for the Sox to try to move him, but if he doesn't get a spot to play for another team, he's pretty much worthless. Take him late and hope for a trade if one doesn't happen by draft day.
Jose Valentin - OF - CWS - last year a poor fielding shortstop, this year an out-of-place Center fielder. He may bump Chris Singleton out of a job, and his OF numbers may not be nearly as valuable as they would be for a SS. If he qualifies as a SS in your league, he'll be a bargain - he won't be taken out of the games late for defense, and he won't have to struggle with his defense - unless he finds that he can't play the outfield.
Royce Clayton - SS - CWS - traded from Texas, should find a good home in the bottom part of a potent White Sox order.
Juan Gonzalez -OF - CLE - don't be lulled into a sense that last year's poor numbers indicate that Juan is on the down side of his career. He was unhappy in Detroit, and the trade to Cleveland should energize him. BIG numbers here.
Tim Salmon - OF - ANA - if he gets traded before opening day, look out world. He has a load of talent, and only his head is standing between him and big numbers (assuming he can stay healthy for a full season). Absent a trade, his contract/free agent status in Anaheim will likely dog him.
Johnny Damon -OF - OAK - he had a great second half last year, and the move to Oakland will help him - a lot. Look for 125 runs scored, 50+ steals to go along with 20 HR and 80-100 RBI. There will be a lot of base runners on at the bottom of the As lineup for him to knock home.
Miguel Tejada - SS - OAK - He's young and getting better. Look for him to bat 2nd, and see a lot of fast balls with Johnny Damon on base ahead of him. Upside is very high for him. Could put up numbers approaching the "big 3" at short.
DD
Next week: Strategy for your Roto Draft
Previous columns:
02/22/2001 - Preparing for your Fantasy Draft
02/12/2001 - Rotisserie® Baseball - What is it, and why should I play?