+-------------+-----+------+----+-----+-----+----+------+
| LAST        | AB  | BA   | HR | RBI | R   | SB | $    |
+-------------+-----+------+----+-----+-----+----+------+
| Weeks       | 450 | .251 | 15 |  44 |  81 | 19 |  $ 7 |
| Mark Ellis  | 150 | .253 |  4 |  15 |  21 |  3 | -$20 |
| Weeks+Ellis | 600 | .252 | 19 |  59 | 102 | 22 |  $17 |
+-------------+-----+------+----+-----+-----+----+------+

+-------------+-----+------+----+-----+-----+----+------+
| LAST        | AB  | BA   | HR | RBI | R   | SB | $    |
+-------------+-----+------+----+-----+-----+----+------+
| Weeks       | 600 | .251 | 20 |  58 | 109 | 25 |  $20 |
+-------------+-----+------+----+-----+-----+----+------+

---

Posted by Derek Carty at 1:40am (16) Comments

Thursday, February 05, 2009

Designing the perfect keeper league

In trying to set up the first keeper league I've ever been a part of, I'm a little daunted by the challenges it presents. I've tried to stick to the most common settings out there (standard BA, R, HR, RBI, SB, W, SV, K, ERA, WHIP, $260 per team) to help increase my chances of being able to blatantly rip off someone else's rule set. But I've still come up empty so far.

In my eyes, the perfect keeper league must do the following:

1. Allow managers to get an advantage if they select players who outperform the consensus expectations
   
2. Ensure such an advantage is small enough that it isn't a deterrent to keeping other managers interested in competing in the future

That's it. Two constraints—should be easy, right? If only it were.

I've heard a number of ideas so far. One is that each manager can keep up to X players, and those players constitute his or her first X picks in next year's draft. If all but one manager keep at least two players, that final manager will have the only first round pick and the only second round pick, from which he can take any available player. The problem here is that if your 2008 draft contains a poor first round pick, you'll have a hard time keeping anyone else because your first player kept counts as your first round pick in 2009.

If you grabbed Prince Fielder in the first round of 2008 and Lance Berkman in the second round, and wanted to keep only Berkman in 2009, he'd occupy your first round slot. He's not a bargain there at all—actually a poor choice—but if you could keep him as a second rounder, you probably would. For this reason, I think Keeper Rule No. 1 is violated by this rule set, and therefore I don't see it as optimal.

Another classic keeper rule is that each manager can keep players for the next season's draft by using their draft pick one round sooner than they were picked in the current year. So if you picked Manny Ramirez in the fourth round in 2008, you could keep him as your third round pick in 2009. Might not be a bad idea, if you think he will be properly motivated this year. The problem, however, is that for truly low-round prospects that are gambled on, the payoff can potentially violate Keeper Rule No. 2.

In 2006, I drafted Jonathan Papelbon in the 14th round—a reasonable gamble, as he had very good call-up stats but it was uncertain whether he'd be the closer or a fourth or fifth starter. Even if it was a keeper league that forced me to draft him two spots sooner each successive year, he'd have been a 12th rounder in 2007, and 10th rounder in 2008, and probably would continue to be a bargain through at least the 2010 season. That, I believe, is a little too much value gained on an errant 14th round pick one season, and the advantage gained from that one fortunate draft pick would probably seem excessive to most.

What about auction drafts? Typical auctions allow for a budget of $260 to be spent on 26 players, with a minimum of $1 to purchase a player. One keeper rule I've heard is that the price to keep a player the following year is simply what you paid this year plus some fixed amount (such as $5). Much like the rule above, this ensures you won't get to keep a bargain forever. However, given the range of typical player values ($1 to about $50), it can take many years for one lucky pick to stop giving outlandish benefits to its manager. Papelbon may have cost $4 in 2006, thus making him a $9 steal in 2007, a $14 robbery in 2008, and so forth. This rule, like the one above, seems to me to violate Keeper Rule No. 2.

One could also auction each player each year, with the previous season's owner getting the right of final purchase. If you owned Alex Rodriguez in 2008, he would be up for auction amongst the other players in your league for 2009. Whenever that auction found a winning bid, you as the current owner would have the right to match that price to keep him. If you decide to do so, there is no further bidding; your price has been set by the "market" of other managers. If you decline, the highest bidder gets him.

In this rule set, I find it hard to imagine too many players would be kept, because an efficient market (the other players in the league) would typically be willing to pay a player's expected value each successive year. Or perhaps more, since in an auction the winner typically overpays for the good for sale; the current owner would have no reason to ever keep someone. This would violate Keeper Rule No. 1— benefit would be afforded to those making good decisions in the past.

However, one could add a discount to solve this. Imagine the current owner got to keep a player at 15 percent less than the other managers' final bid price. In such a case, it would almost universally be in the interest of the current owner to keep his player. Other ramifications would follow, however: managers could theoretically overbid for a player by 10 percent, knowing that once 15 percent is knocked off of the final value, the current owner is still getting a slightly good deal and would benefit from picking up the player at ever so slightly below his actual value. For this reason, I believe this rule set still violates Keeper Rule No. 1, in that it actually does not afford enough advantage to managers making good decisions.

Finally, we come to the actual league rule I've found to best mix the interests of Keeper Rule No. 1 and Keeper Rule No. 2. It's predicated upon an auction format with standardized categories for which many projections and price guides exist. The league decides on a publicly-available price guide, from which player values are derived. Each season, managers can keep as many players as they wish, with their next season's price being equal to the average of each season's price guide value during which he or she owned the player.

For players whose value doesn't change much over time—Albert Pujols for example—no advantage is derived by keeping him. He typically costs around $40, he's typically worth around $40, and his price guide price is typically around $40. For prospects, I believe the system still works. If you picked up Papelbon for $4 (his price guide value) in 2006, his price guide value before the 2007 season would reflect his new value as an elite closer with one season of great numbers - say, $20. Your price to keep him for 2007 would be ($4 + $20) / 2 = $12. When he put up tremendous numbers in 2007 once more, he would likely have been valued around $24 going into the 2008 season. Your price to keep him then would be ($4 + $20 + $24) / 3 = $16.

There's still a very large benefit to finding great players who are underrated for a season—as you can see, in 2008 Papelbon was likely a $20+ player who could be kept for $16—but he is one of the stronger examples of coming out of nowhere one season to become an elite player. And, as opposed to the system in which price goes up by a fixed amount each season, this system more quickly aligns a player's cost to his true value.

The point of this article was to present some ideas about how to construct the "perfect" keeper league. I'd love to hear some feedback about what you've seen or heard of that might work better than any of these—specifically if they're better than the final rule set, because that's what I'm leaning towards at the moment.