In the stock market, diversification is the key. If you asked any financial analyst worth his weight in cattle futures what investment advice he would give you if he could only utter one word, he would yell out, "Diversify!" You don't put all your money into companies that sell the same product because those companies' stock prices will tend to rise and fall together, and you'll experience wild swings in the value of your investment. This is what is meant by not putting all your eggs in one basket.

Diversification is important because it gives you consistency in the value of your investments. Veteran fantasy owners should understand instantly what I mean by consistency, because after projections, it's probably the most important factor we use in ranking players on our cheatsheets. It is why many of us stay away from Corey Dillon and Isaac Bruce and other players who have 2-4 huge games every year that inflate their stats, and then disappear the rest of the season. Many of us downgrade inconsistent players accordingly, and instead choose players who are projected to score slightly less but more consistently on a week-to-week basis.

Yet, despite this search for consistency, over the years I've seen many fantasy owners drool over the idea of having a stud QB and stud WR from the same NFL team. These owners long for that "double dip" that they get every time the two players hook up for a TD pass. There is a great deal of controversy about this strategy, however, and many fantasy vets reject it because, in their guts, they know this is a high-risk strategy that leads to inconsistent week-to-week performance. If Peyton Manning gets shut down, Marvin Harrison probably also gets shut down. If Terrell Owens misses time due to injury, Jeff Garcia's numbers suffer. If Daunte Culpepper can't hold onto the football, Randy Moss won't have as many red zone opportunities. In other words, your stud tandem is "boom-or-bust," and if there's one thing a savvy investor knows, it's that "boom-or-bust" investments are a very bad idea. After all, "boom-or-bust" is really just another way of saying "undiversified" - or "inconsistent."

Do statistics back up our gut instinct that these QB-WR tandems lead to greater inconsistency in our fantasy lineups? Can we debunk the "double dip" myth, and show that the "double dip" is nothing more than the boom component of a strategy that will eventually bust?

Let's take a look at top QB-WR tandems, and see how having them on the same fantasy team affects consistency. I have chosen tandems where both the QB and the WR finished in the top 12 at their positions for 2002, because I'm really only concerned with WRs that are worthy of WR1 status on fantasy squads. We'll use the following scoring system:

Passing TD = 4
Rushing/receiving TD = 6
25 passing yards = 1
10 rushing/receiving yards = 1
INT = -1.

As a measure of consistency, we'll use standard deviation. In a nutshell, standard deviation represents the amount by which we can expect a player's weekly scores to fluctuate around his average weekly score. A player's weekly scores will fall within one standard deviation above or below his average 68% of the time, and within two standard deviations 95% of the time. For example, if a WR averages 15 fantasy points per game, and has a standard deviation of 3, then we can expect 68% of his weekly scores to fall between 12 and 18. Likewise, we can expect 95% of his weekly scores to fall between 9 and 21. Thus, a lower standard deviation generally means weekly scores will be closer to the average - and it means a more consistent player.

It should be noted that just because one player has a larger standard deviation than another player, that does not necessarily mean that the first player is more consistent. If the second player has a higher average score, then it stands to reason that he will have a higher standard deviation. For example, a player with an average of 10 and a standard deviation of 4 is no more "consistent" than a player with an average of 20 and a standard deviation of 8. Thus, when comparing players, it makes sense to look at standard deviation as a percentage of average score.

Now that I've explained in excruciating detail the concept of standard deviation (quiz tomorrow), let's take a look at standard deviation for each player, based on that player's weekly scoring in each game played in 2002. Then we'll add up the fantasy points scored by each QB-WR tandem for each week of the 2002 season (ignoring weeks in which one of them did not play), and determine the tandem's standard deviation as a percentage of its average fantasy PPG. After looking at the tandems, we'll rearrange the pairings so that each QB is paired randomly with a WR from a different NFL team.

First, the QB-WR tandems:

Above, we see that the average standard deviation for the 10 individual players listed is 46.3% of their scores. When those players are grouped into pairs based on their NFL teams, the average standard deviation of the 5 tandems is 41.4% of each tandem's combined score. This means that by pairing a QB with a WR from the same team, we only reduce the standard deviation of one single player by a meager 4.9%. This small change is very disappointing because it suggests that the individuals in each tandem tend to rise and fall together on a week-to-week basis. An investor would say that adding the second player to the first achieves very little diversification.

As a basis for comparison, let's randomly match each QB with a WR from a different NFL team. Here are the results:

Here, we see a much different story. The average standard deviation of these 5 QB-WR pairings is a full 15% lower than the average standard deviation of the 10 individual players. A decrease from 47.5% for a single player to 32.5% for a tandem from different NFL teams represents a significant increase in week-to-week consistency in your fantasy lineup. In investor terms, you have succeeded in diversifying your investment by combining a QB and a WR1 from different NFL teams.

So, drafting a QB-WR tandem from the same team only reduces the inconsistency of one player by less than 5%, while pairing QBs and WRs from different teams improves consistency by 15%. When you take into account the other problems associated with QB-WR tandems - doubled injury risk, bye week problems, etc. - this seems to weigh heavily against targeting those stud QBs and WRs from the same team. Double dip? No thanks.

There remains, however, the issue of how much to downgrade a QB when you already have the WR1 from the same team. If you pick Marvin Harrison in the 2nd round, and then Peyton Manning falls to you in the 6th round, do you pass on him? Certainly not! Peyton Manning is a great value in the 6th round whether you have Harrison or not. But what you should consider is that you should bump Manning down just below those players who you have ranked within a few points of Manning on your cheatsheet. For example, if you have Manning projected at 300 points and Rich Gannon projected at 295, you might want to move Manning just below Gannon if you take Harrison early in the draft. A 5-point difference over 16 games is certainly within the margin of error. Or, if you are using VBD and you draft Harrison, and Manning's value is 50 and a potential RB for your team is valued at 45, you might want to consider getting that RB and waiting another round or three to grab your starting QB.