I believe historians date this adage back to 1962, when thousands of owners in AFL fantasy leagues got burned by taking flashy third-year man Bill Groman instead of old reliable Abner Haynes. Ever since then, it has been commonly believed that, while some amount of risk-taking is essential to winning your league, it should not be done in the early rounds. Especially not in the first round.

You can't win your league in the first round, but you sure can lose it.

In the first round, you take guys that you know are going to give you solid production. Even if it means passing on some players with tempting upsides, you pass on question marks in round one. Or so the saying goes.

Alright, if you can't win your league in the first round, when can you win it? Conventional wisdom says you win it by getting steals in the middle and late rounds. You win your league (in 2005) by getting Carson Palmer in the 6th, by getting Santana Moss in the 9th, and by getting Terry Glenn in the 16th.

Conventional wisdom is wrong. The single pick with the most upside potential each year is your first round pick. The next most potential for upside comes with your second round pick, and so on.

Don't believe me? Ask yourself this: how much value did you get out of Santana Moss or Terry Glenn last year? According to standard footballguys scoring and worst-starter VBD, Moss was worth 85 points of value last year, Glenn was worth 45, and Palmer was worth 58. Now that's a lot of value, no question. A large percentage of the guys drafted near Moss, Glenn, and Palmer earned zero value, so those who drafted Moss certainly got a big advantage over their opponents. Eighty-five points is a big deal.

But it's a drop in the bucket compared to the 180 points of value you got by drafting Shaun Alexander instead of Peyton Manning, whose ADP was almost identical, in the first round. OK, OK, you're way too much of a shark to have considered Manning. You might even argue that not making that choice would be an example of losing your league in the first round. What about this: taking Alexander over LaDainian Tomlinson would have gotten you about as much extra value as Carson Palmer was worth. In the second round, the difference between Tiki Barber and Marvin Harrison, who had a nearly identical ADP, was worth more than three Terry Glenns.

Winning and losing are, of course, two sides of the same coin, but I fail to see how taking the rock solid performers like Marvin Harrison and Peyton Manning could be called cases of losing your league in the early rounds. Rather, many who chose Alexander or Barber instead did in fact win their leagues in the early rounds.

Over the past six years, the average first round pick, if he stays healthy, has been worth about 100 points of value, yet first rounders routinely approach or exceed 200, and it's not always the guys at the top of the round either. So it's not uncommon to have 100 points of upside there for the taking when your first round pick comes up. It is, however, extremely uncommon for there to be 100 points of upside available for you when your 5th round pick comes up. Or your 6th round pick. Or any other pick besides the first or the second.

So yes, your first round pick always carries a lot of downside potential; if you get a goos egg, you're 100 points in the hole compared to your leaguemates. But it also has more upside potential than any other pick you make. You can win your league in the first round, at least as much as you can win it in any other single round. Ricky Williams in 2002, Priest Holmes and Ahman Green in 2003, Daunte Culpepper and Peyton Manning in 2004, Alexander in 2005, these are all first round picks that would have given you more "bonus" value than almost any mid-round steal.

Unproven players: are they overvalued?

Discussions about losing your league in the early rounds generally occur during discussions about players --- usually young players --- who have never proven that they are capable of the production their ADP implies. Obviously they have the potential or else their ADP wouldn't be that high, but they seem risky because they are not known quantities. I'm going to define a player to be "unproven" if his ADP is higher than his previous career best finish (fine print). This would obviously include all rookies. It includes players like Kevan Barlow in 2004 (ADP: RB12, best previous finish: RB17) and Kevin Jones last year --- these are the ones we remember well --- but don't forget that it also includes players like LaMont Jordan and Larry Fitzgerald from last year. This year's unproven players of interest include, but are not limited to, Ronnie Brown, Steven Jackson, Cadillac Williams, Reggie Bush, and Roy Williams.

So let's do an experiment. We'll look at the last six years of ADP data and do the following: we'll pretend that we are drafting from the ADP list and that, when our turn comes up, an uproven player is tops on the list. Then we'll scan down the list and look at the next-highest-ranked player who is not unproven. Then we'll see who did better (in terms of fantasy points per game) in that season.

For example, in 2000, the rookie Jamal Lewis had an ADP of RB18. The next running back on the ADP list was the venerable Ricky Watters, who was obviously not unproven. As it turned out, Watters would have been the better choice. He averaged 2.3 fantasy points per game more than Lewis did.

Here's how that will look in the data I'll present:

Name                  YR  PrevHi ADP  FPPG
-------------------------------------------
Jamal Lewis          2000     -   18  12.6
Ricky Watters                 3   19  15.0
DIFF:  -2.3

But one example means very little. That's why we're collecting six year's worth. Also, because our question is somewhat subjective and our algorithm is rote, some examples might even seem a bit misleading. That's why I've included all the data right here. You can comb through it and focus on only the subset that seems most relevant to you.

But the point is, if the market overvalues unproven players then we should be able to gain advantage, in the long run, by skipping over them and taking a lower ranked but presumably safer option. And if that's the case, then at least some evidence for it ought to show itself in this study.

Results: overall, there were 136 pairs. In 73 of them (53.6%), the unproven player outperformed the proven one. On average, the unproven player outperformed the proven one by 0.47 fantasy points per game. In the official sense, this is not anywhere near statistically significant evidence that unproven players are undervalued (especially since the comparison is slightly rigged in their favor), but that's not important. The important thing is that we see no evidence that unproven players are overvalued.

I have long believed that the fantasy football market is pretty efficient and this study is, in my view, more evidence in support of that view. There may be a few players each year who are incorrectly valued by the masses, but I have yet to find any identifiable group of players who are systematically over- or undervalued by the general fantasy football playing public.

I'll close with the positional breakdown:

Pos     Number of Pairs      Unproven was better    Proven was better   Avg Diff
--------------------------------------------------------------------------------
QB           28                     20                      8             +1.66
RB           49                     22                     27             -0.10
WR           37                     19                     18             +0.00
TE           22                     12                     10             +1.02

You might be tempted to infer from this that unproven quarterbacks are actually undervalued, but I'm not sure there is enough evidence here to conclude that. It's a pretty small sample. Still, it does at the least say that it's very unlikely that unproven quarterbacks are overvalued. If you like Philip Rivers, for instance, and are trying to build up your resolve to pull the trigger on him, this might help.

Fine print

1. I excluded players who had previously achieved an elite level of production, even if it wasn't quite as high as their ADP. For example, LaDainian Tomlinson has never ranked higher than RB3, but even his ADP is RB2, no one would think of him as unproven.

3. I only included the top 20 QBs, the top 30 RBs, the top 30 WRs, and the top 15 TEs of each season.

2. All ADP data is taken from myfantasyleague redraft league drafts conducted after August 25th of the given season.