To paraphrase Lenin, "Sabermetrics, yes, but for whom, in the service of what?" It is certain that any deployment of sabermetrics requires the proper balance. But what should this balance reflect? The nature of victory in a hypothetical baseball game? This seems odd, since rotisserie scoring is based on the year-long stats of the league, setting the numerical logic of such value at a remove of potentially x162 for position players and x30 for pitchers. Balance required for a team to go the postseason? This seems more reasonable since it is a season long metric. This brings us to the topic of my discussion - in what sense is titration in baseball statistics a cognizable system?
In order to reflect upon this question we must in turn ask ourselves a series salient related questions. Fantasy baseball appears to present us with a dialectical system, that is on first glance, it would appear that fantasy statistics have a literal and geometric relation to the statistics of the MLB, upon further reflection it then appears that they have perhaps a more symbolic and functional relation, and then exhausted, we must admit that in some complex sense both are true. The result of this experiment should rather delight us than cause gnashing of teeth - for what is titration but a dialectic process of truth seeking?
Salient Questions:
Salient Questions:
A baseball line-up card is filled out with 10 players (9 in the National League) at any given time in the game. Naturally, there are more players on the roster than typically participate in a single given game. From April-September MLB rosters are set at 25 players, in September the number rises to 40. There are several senses in which these numbers are irrelevant to fantasy statistics. The most obvious is that nearly 1/3 of the players on a standard roster will be statistical dwarfs because they A.) are platoon/back-up players, B.) are pinch hitters C.) are middle relievers. Do these players have value to the team? Certainly. Is their value the often the difference between victory and defeat? Perhaps. Is their value reflective of the overall liklihood of a team making the playoffs. Probably not.
A second difference is that players on a fantasy team are, obviously, not necessarily members of the same team. While all position starters will have the opportunity to play 162 games, health & management permitting, they will not play them on the same days. Should it be possible to acquire "extra" statistics by closely managing one's roster to fill in all potential gaps, giving one potentially 180 games of statistics. There are several ways to manage this through titration.
1. waiver moves may be limited (last year we limited them to 25, or 1 per week), thus rewarding the successful pick-up but discouraging game to game wire-plug-ins.
2. roster size may be limited, forcing franchises to drop valuable or injured players if they want a temporary fix.
3. valuable players may be "franchised" so they may not be dropped even though under-performing or injured.
4. the waiver wire may be monetized.
Although baseball does not lend itself to so clear a waiver-bid set-up as football, we should not reject the monetization of waivers out-of hand but rather examine what is required for them to function as a fluid economy. A.) a competitive bidding structure (i.e. a day in which all free agents become available for bidding). B.) an active market with enough engaged players who understand the mechanics of the wire to make effective and realistic bids. C.) Ratio of fantasy league size to league pool. For example, last year we played with 6 franchises in an American League only set-up. With 17 player rosters that's 17x6 = 102 players out of 25x15 = 375 players. If we assume that aproximately 1/3rd (125) of those players are statistically invisible at 102 are taken at any given time that leaves 148 players who may potentially be of value on a week to week basis. Thinking practically, we may assume that likely only 1/10th of these will actually be statistically of interest. Still, 14 players in a 6 team league would likely make monetization impractical. A 12 team league, might be a different story.
5. Scoring categories could be set to discourage constant line-up alterations. That is, we could remove scoring categories that record "base" statistics like hits, runs, and errors, and record only statistics that reflect multiple statistics.
So, we should say tentatively that a fantasy team should reflect the general shape of a baseball team. That is, it should retain all positions relevant to baseball in relative proportion to the size of the pool from which the league draws. We should affirm that as players that are statistically un-interesting are of little value, so should statistics which "tell us little" be restrictively valued. And we should agree that the ideal of the league is to "win the season" and thus our categories should be primarily oriented to favoring those statistical measures which most reflect a player's individual contribution to victories over the season.
Should certain statistical categories be multiply titrated?
A second difference is that players on a fantasy team are, obviously, not necessarily members of the same team. While all position starters will have the opportunity to play 162 games, health & management permitting, they will not play them on the same days. Should it be possible to acquire "extra" statistics by closely managing one's roster to fill in all potential gaps, giving one potentially 180 games of statistics. There are several ways to manage this through titration.
1. waiver moves may be limited (last year we limited them to 25, or 1 per week), thus rewarding the successful pick-up but discouraging game to game wire-plug-ins.
2. roster size may be limited, forcing franchises to drop valuable or injured players if they want a temporary fix.
3. valuable players may be "franchised" so they may not be dropped even though under-performing or injured.
4. the waiver wire may be monetized.
Although baseball does not lend itself to so clear a waiver-bid set-up as football, we should not reject the monetization of waivers out-of hand but rather examine what is required for them to function as a fluid economy. A.) a competitive bidding structure (i.e. a day in which all free agents become available for bidding). B.) an active market with enough engaged players who understand the mechanics of the wire to make effective and realistic bids. C.) Ratio of fantasy league size to league pool. For example, last year we played with 6 franchises in an American League only set-up. With 17 player rosters that's 17x6 = 102 players out of 25x15 = 375 players. If we assume that aproximately 1/3rd (125) of those players are statistically invisible at 102 are taken at any given time that leaves 148 players who may potentially be of value on a week to week basis. Thinking practically, we may assume that likely only 1/10th of these will actually be statistically of interest. Still, 14 players in a 6 team league would likely make monetization impractical. A 12 team league, might be a different story.
5. Scoring categories could be set to discourage constant line-up alterations. That is, we could remove scoring categories that record "base" statistics like hits, runs, and errors, and record only statistics that reflect multiple statistics.
So, we should say tentatively that a fantasy team should reflect the general shape of a baseball team. That is, it should retain all positions relevant to baseball in relative proportion to the size of the pool from which the league draws. We should affirm that as players that are statistically un-interesting are of little value, so should statistics which "tell us little" be restrictively valued. And we should agree that the ideal of the league is to "win the season" and thus our categories should be primarily oriented to favoring those statistical measures which most reflect a player's individual contribution to victories over the season.
Should certain statistical categories be multiply titrated?
Another question from this question might equally well be posed "Is there a way not to reward certain statistics multiple times?" For example, if we used the categories AB, BA, and Hits. Theoretically, we would be simply doubling the points for this category, because BA = Hits/AB. However, if we took out the hits and at bats, then batting average could be skewed by picking a player with a very high average and a small number of at bats. Ideally, we could simply require a number of at bats for the season, for the team and be done with both hits and at bats. But it calls into question the value of individual statistical categories. Are hits valuable, of themselves? Or are they only valuable as situational elements? Logic tells us that hits are valuable in themselves, but that they are amplified by such a variety of situations as to make their naked statistical value almost useless. So, we might say that although there is no way to reduce the statistical overlap in BA, OBP, SLG, BA-RISP there is no need to muddy the waters with walks, singles, doubles and triples. Runs, interestingly may be another matter which brings us to the next question.
Are certain categories valuable through rarity or contribution to victory above and beyond how they are recorded in other metrics?
The popular imagination holds home runs at the pinnacle of the baseball statistical world. And certainly for sheer spectacle it is difficult to top the long ball. But is it deserving of its own categorical measure. And beyond fan experience, there is a some reason for this. A home run is the best a batter can do in a single trip to the plate (i.e. advance four bases). It is also, in an interesting way a backwards way of recording RBI's, if we remove RBI in favor of batting-average-with-runners-in-scoring-position, we should also count HRs for an effective measure of how the batter created runs with a swing of the bat. On the pitching side we might say that complete games are also difficult to entirely capture though other statistical measures. In today's game a complete game is a rarity for many pitchers, and many teams are simply averse to them as a standard practice. But still it's an important question to ask: is completing 9 innings substantially more valuable than completing 8. Should freak occurances like no-hitters, perfect games, and cylces be kept as a bonus point system?
Categories
For reference: here are the categories we used last year:
Are certain categories valuable through rarity or contribution to victory above and beyond how they are recorded in other metrics?
The popular imagination holds home runs at the pinnacle of the baseball statistical world. And certainly for sheer spectacle it is difficult to top the long ball. But is it deserving of its own categorical measure. And beyond fan experience, there is a some reason for this. A home run is the best a batter can do in a single trip to the plate (i.e. advance four bases). It is also, in an interesting way a backwards way of recording RBI's, if we remove RBI in favor of batting-average-with-runners-in-scoring-position, we should also count HRs for an effective measure of how the batter created runs with a swing of the bat. On the pitching side we might say that complete games are also difficult to entirely capture though other statistical measures. In today's game a complete game is a rarity for many pitchers, and many teams are simply averse to them as a standard practice. But still it's an important question to ask: is completing 9 innings substantially more valuable than completing 8. Should freak occurances like no-hitters, perfect games, and cylces be kept as a bonus point system?
Categories
For reference: here are the categories we used last year:
| Batters Stat Categories: | Runs (R), Hits (H), Home Runs (HR), Runs Batted In (RBI), Caught Stealing (CS), Walks (BB), Total Bases (TB), Putouts (PO), Assists (A), Batting Average (AVG), On-base Percentage (OBP), Slugging Percentage (SLG), Net Stolen Bases (NSB), Double Plays Turned (DPT) |
| Pitchers Stat Categories: | Wins (W), Losses (L), Complete Games (CG), Shutouts (SHO), Saves (SV), Outs (OUT), Walks (BB), Strikeouts (K), Earned Run Average (ERA), (Walks + Hits)/ Innings Pitched (WHIP), Strikeouts per Nine Innings (K/9), On-base Percentage Against (OBPA), Walks Per Nine Innings (BB/9), Quality Starts (QS), Blown Saves (BSV) |
No comments:
Post a Comment