The road to hell is paved with good intentions. And that just happens to be the way I feel about the red-headed stepchild stat that is the "hold," which has become a surprisingly popular addition to traditional leagues. I understand the attraction. Non-closing relievers represent nearly 25 percent of the typical major league roster, but remain the most underappreciated (and undrafted) type of player in fantasy baseball. Including holds as a category helps balance out this inequality.

In fact, even this humble scribe decided to include holds when it came time to start a league a number of years back. I know why. The reason was Duane Ward.

Remember Duane Ward? Ward was a reliever on the 1992 World Champion Toronto Blue Jays. That year, he posted a spectacular 1.95 ERA, 9.1 K/9 ratio and 2.64 SO/BB ratio -- not to mention a 1.135 WHIP and seven wins. Alas, he only had 12 saves and few knew who he was. Why? Because Ward would often pitch a sparkling one-two-three eighth, only to have the venerable Tom Henke come in and close the game. Other than saves, Ward's numbers were better than Henke's. Where was the justice? Was Ward's eighth inning that much less deserving than Henke's ninth? Thus, I became a believer in holds and had the best of intentions. (FYI, with Henke gone the following year, Ward recorded 45 saves and despite his ERA "ballooning" to 2.13, actually improved his other numbers.)

So, what is a hold? The hold was created in 1986 by John Dewan and Mike O'Donnell as a means to measure the effectiveness of the vast majority of relief pitchers who rarely have the opportunities that a closer has. Still, while middle relievers earn their share, holds are most often credited to setup men. But that barely scratches the issue of the hold.

A hold is awarded to a relief pitcher who enters the game in a save situation and leaves the game without his team having relinquished the lead. Therein, however, lies the problem.

Since the hold is not an official Major League Baseball statistic, the specifics can vary from stat provider to stat provider. For example, STATS, Inc. requires the pitcher to record at least one out for a hold, SportsTicker does not. This leads to discrepancies in hold totals depending on the source.

Furthermore, even a relatively ineffective performance can earn a hold. For example, a reliever that enters with a one-run lead, records one out, loads the bases, and is then pulled would still be credited with a hold. It is even plausible for a player to receive a hold and a loss in the same game. In the preceding situation, if the next pitcher immediately gave up a home run, the original pitcher would take the loss, as he was responsible for the runners on base. The SportsTicker definition makes it even worse. Suppose a pitcher comes in with a two-run lead, gives up a home run to the only batter he faces and leaves. He still would be credited with a hold.

Plus, unlike saves, wins, and even losses, more than one pitcher per team can earn a hold for that game.

I admit it, I made a mistake. While I thought it was a good idea at the time, I have since learned that the hold is not such a great stat.

But I know your league is not going to change the scoring system just because I think you should. So, if your league does count holds, who should you look to get?

First let's look at the list of relievers who recorded more than 20 holds last year.

This should tell us a few things:

1) Grab the guys that have done it before

What a novel concept! But when you're trying to project a category that is as unpredictable as holds, it pays to look at those who have done it consistently. Marmol and Shields once again are amongst the leaders this year as well. Even with something as fluid as bullpens, it should be no surprise that Shields, Okajima, and Romero were all in the Top 20 in holds in both 2007 and 2008. Shields was also in the top three in both 2006 and 2005. So, if you're in a league that counts holds, get Scot Shields. Of course if you're in a league that counts holds, your leaguemates know his value and good chances he's already taken. For those getting their first whiff of a holds league however, it is obligatory that I mention it.

2) Nothing wrong with closers!

I have yet to hear about a league that includes holds that doesn't include saves. If you go after closers, two possible scenarios can unfold: One, you end up getting saves instead of holds -- is that such a bad thing? (You can always pick up a holds guy later); the other is that your saves guy loses the gig and goes back to being a set-up guy and lo and behold, he gets holds.

3) Remember: Ratios!

Carlos Marmol was the most valuable holds specialist of any reliever who failed to reach double-digit saves (yes, even more valuable than Shields). Why? A 1.23 ERA and .95 WHIP can't be hurting you in those two categories!

4) Playing the Role of Reliever tonight...

There is not a perfect correlation to the number of appearances made and holds recorded. However, Feliciano, Ohman, and Marmol were the top three in appearances made last year and what do you know, they happen to be in the Top 20 in holds too. Coincidence? I think not.

5) Special K

Remember Duane Ward's exceptional K/9 ratio above? It should come as no surprise that Guardado and Bradford are the only ones on the list above with a K/9 ratio of less than 6.00. It is a lot easier to leave the game with the lead when you can strike people out two-thirds of the time than when you can't.

I've now given you a template of what to pursue in a league with holds. However, I know for some of you, it is not enough. For those of you too lazy to do your own research, here are three suggestions: Try Rafael Soriano (RP, ATL), Ryan Franklin (RP, STL) and Jason Frasor (RP, TOR). I'm confident those three should provide value ...y ou can even hold me to it!

SI Apps
We've Got Apps Too
Get expert analysis, unrivaled access, and the award-winning storytelling only SI can provide - from Peter King, Tom Verducci, Lee Jenkins, Seth Davis, and more - delivered straight to you, along with up-to-the-minute news and live scores.