Introduction to these player comments
Index of player comments
Glossary of terms
Kurt Warner career statistics

I play in two leagues each year. One is a keep-7 league in which I do not own Warner. Further, I know I don't want to pay whatever it would take to pry him away from his current owner. The other league is a redraft with rules that I believe de-emphasize QBs relative to RBs and WRs. Nonetheless, that league contains some QB-lovers, so I know Warner (and the other top shelf QBs) will be long gone before the slots where I'd consider them good value.

As a result, I don't spend any time thinking about Warner. It would be only a mild exaggeration to say that he's not even on my draft board. So I won't be astonishing you with my insights into what Warner is likely to do this year.

What I'll do instead is examine some data on the more general question of how top QBs hold their value compared to the top guys at other positions. Take a look at this chart. Explanation follows:

                 Median finish
   Group           next year
-------------------------------
qb 1 - qb 5            8
qb 6 - qb10           16  
qb11 - qb15           15  
qb16 - qb20           23  
qb21 - qb25           28  
qb26 - qb30           33  

This includes all players from 1970-2000. For the first line, I looked at all QBs who ranked first through fifth. So we have five guys each year, times 31 years. That's 155 player seasons being considered in that first group. I then tracked those 155 players the next year. Their median finish was #8, which means that half of them finished #8 or better and half finished #8 or worse. Likewise, if you look at QBs who finished 6-10 in a given year, their median finish the next year was #16. And so on.

This doesn't mean much unless we compare it to the RB, WR, and TE data, so here goes:

                 Median finish
   Group           next year
-------------------------------
qb 1 - qb 5            8
rb 1 - rb 5            8  
wr 1 - wr 5           10  
te 1 - te 5            6  

So top 5 RBs tend to hold their value about as well as top 5 QBs do. WRs are a little worse, and TEs are a little better.

                 Median finish
   Group           next year
-------------------------------
qb 6 - qb10           16  
rb 6 - rb10           18  
wr 6 - wr10           21  
te 6 - te10           14  

Much the same picture here. QBs and RBs about the same (QBs a little better), WRs a little worse, and TEs a little better. Let's finish out the top 20s:

                 Median finish
   Group           next year
-------------------------------
qb11 - qb15           15  
rb11 - rb15           26  
wr11 - wr15           21  
te11 - te15           18  

                 Median finish
   Group           next year
-------------------------------
qb16 - qb20           23  
rb16 - rb20           29  
wr16 - wr20           29  
te16 - te20           29  

Frankly, I'm a little surprised by how well the QBs come out looking here. I would have guessed that RBs would be more reliable year-to-year than QBs, but these data don't support that notion.

Let's check out a more modern (though necessarily smaller) data set. This time, I'll run the same numbers, but only include the years 1995-2000.

                 Median finish
   Group           next year
-------------------------------
qb 1 - qb 5           11
qb 6 - qb10           19
qb11 - qb15           15
qb16 - qb20           21
qb21 - qb25           26
qb26 - qb30           35

rb 1 - rb 5           11
rb 6 - rb10           14
rb11 - rb15           23
rb16 - rb20           20
rb21 - rb25           38
rb26 - rb30           46

wr 1 - wr 5           10
wr 6 - wr10           21
wr11 - wr15           20
wr16 - wr20           38
wr21 - wr25           23
wr26 - wr30           44

te 1 - te 5            5
te 6 - te10           15
te11 - te15           21
te16 - te20           33
te21 - te25           29
te26 - te30           30

There's a lot to absorb here, so I'll just let you poke around at the data and draw your own conclusions if you're interested.