Thanks a lot, Dak Prescott. Now more people will think the fourth round is still a gold mine for quarterbacks, but the data says otherwise. The update to our quarterback draft study for 1994-2016 shows little has changed: finding a good QB is really hard.
01 Aug 2005
Guest Column by Dan Lewis
The big news came across the wire five minutes past midnight on March 17th -- the report that made me jump out of my chair and yell "OH... YEAH!" like the Kool-Aid guy. No, you probably did not see it as a headline on ESPN.com or Yahoo! Sports. Unless you read the AP wire itself -- hi, my name is Dan, and I'm a dork -- you would have missed it.
Baltimore running back Chester Taylor signed a one-year, $3 million offer sheet with the Browns.
March 17th, 12:05 A.M. The first game of the NCAA men's basketball tournament was just over twelve hours away. Spring training was already in full swing. And here I was, excited over a back-up running back signing a contract -- no, a mere offer sheet -- with another team.
Your fantasy football league cannot claim to cause that.
Welcome to the next generation of fantasy football.
Your league probably resembles the rules found on, say, Yahoo Sports. Twelve team league, 15-player rosters. You start one quarterback, two running backs, three wide receivers, a tight end, and a defense. Your schedule is head-to-head, running from Week One until Week 14. The top four teams make the playoffs, with the Fantasy Super Bowl played in Week 16, with Week 17 idle.
Sure, you may have your peculiarities, be it 10 teams; four points for thrown touchdowns or a flat three for a field goal; you start only two wide receivers; top six teams make the playoffs; etc. But the gist of the rules is the same.
And most of them have to go.
Most. You should keep your scoring system. You probably expected me to advocate for using DVOA or DPAR in your fantasy league, adding a big slice of accuracy to an otherwise inaccurate game. Not happening. One of the great allures of fantasy football is the instant gratification one gets when his quarterback and wide receiver hook up for a touchdown. When Tom Brady hits Outsider fave Deion Branch for a score, I know that I just got a quick 12 points. Yahtzee! Break out a calculator to figure out how many DPAR points I got? Uhh ... no thanks.
A scoring system should be simple. Six points for a touchdown, fine. Four points for a TD tossed, fine. Want to penalize Ahman Green for fumbilitis? Whatever you fancy. Add in some bonus points for yards rushed, passes caught, yards from scrimmage, etc., that's fine too. Makes reading the box score fun.
So, keep your scoring system or tweak it if you would like. Just remember the primary goal of fantasy football -- have fun.
All your other rules go buh-bye.
First, let us develop a mantra. Make sure you repeat it five or six times. Ready?
This is not real.
Whenever you put in a rule because you want your game to parallel the one going on the field, you have violated the mantra. This game is fantasy football. It is, as my wife calls it, fake football, and she is right. F-a-k-e fake. Trying to make it consistent with reality is impossible and, more importantly, foolish. This is not real.
Therefore, the first rule to go is the head-to-head format.
The real problem with this sadly typical style is that the results -- a simply W or L -- do not reflect skill. When assembling your team, your goal is simply to amass as many points as possible. But when you use a head-to-head system, it changes; suddenly, the goal is to get at least one point more than the other guy.
True, this is how real football works -- a last second field goal to edge your opponent is as much a win as a 27-point blowout. But, remember our mantra! Yes, the NFL uses a head-to-head format, but in that universe, one has the ability to develop a team suited toward that format. You, the fantasy owner, cannot do that. No matter how hard you try, you cannot make your fantasy league into the real deal.
Even worse, by employing a head-to-head system, you add a significant degree of luck to your game. We have all been in the situation where our team puts up the second or third most points for the week -- but happens to face one of the few teams that put up slightly more. Meanwhile, some team that barely breaks 40 points wins because he was matched up against the Kyle Boller/Fred Taylor powerhouse.
There is another way. No, I'm not advocating for Rotisserie Football; if anything, that style causes the same problem that using a DPAR-based league would. But more importantly -- and as you will see later -- RotoFootball does not work well with the rest of my suggested framework.
Employ a fantasy point system. Add up the points like you would normally -- six for a TD, four for 100 yards passing, whatever. Whomever gets the most over a stated time period wins.
The "time period" thing -- that is a bit tricky..
The two obvious choices -- one week or the whole season -- both suffer from the same problem. After a few weeks, teams start to fall out of competition. Do you want the bad teams to lose interest after week nine? Hardly. I advocate strongly for a keeper league, but that simply replaces the dead-team problem with an obvious second one: talent dumps.
In any event, these problems are easily avoided by using a makeshift "quarter" system like my current league uses. Weeks 1 through 3 are the first quarter. Weeks 4-7 are the second, 8-10 the third, and 11 to 13 the fourth. Each quarter winner advances to a single elimination playoff. Yes, there is one four-week quarter, but that is unavoidable. (Because the NFL season is 17 weeks, and because my current league opts to reserve weeks 14-15 and 16-17 for the playoffs and league championship, respectively, we end up orphaned week, and append it to the second quarter.)
Say you have 10 teams in your league. All 10 would compete for the quarter 1 ("Q1") title. The winner would advance to the playoffs to face the Q2 winner. The other nine teams would have another three chances to advance to the playoffs. The nicest part is that, after week 10, the season starts anew. Even if a team was dead last for each of the first 10 weeks, it can rally in weeks 11 to 13 and qualify for the playoffs. What other rules system can claim that?
You may think this disadvantages the truly superior teams -- say, the Q1 winner. Hardly. He can win other quarters as well, to ease his path to a title. If he wins, say, quarters 1, 2, and 3, he effectively has earned two shots at the championship.
By removing the head-to-head nonsense, we reduce the luck factor. By starting anew four times in the same season, we give all teams a chance to compete even as the season is winding down. Finally, we reward the elite teams by easing their path to the chalice. These are the hallmarks of good rules design.
At one point last year, Ron Dayne, officially speaking, "started" for the Giants. On their first play from scrimmage, he -- not Tiki Barber -- was in the backfield. Fictionalizing a bit -- okay, a lot -- let us postulate that Tim Carter was in and Amani Toomer was sitting next to Tiki on the bench; that Kurt Warner was in at QB; and that for some reason Jeremy Shockey was also sitting this one out.
After the first play, Tiki comes in. Amani's hammy feels better -- Carter goes back to three-receiver duties. On the first passing down, Shockey comes in for good. And when the Giants fall behind 17-0 after the first quarter, Coughlin gives the ball to Eli.
If this were a fantasy game, the Fantasy Giants starters -- Dayne, Warner, Carter, and whoever the blocking tight end is -- would put up a paltry score. Luckily for the Giants (not that it mattered in '04), they are not a fantasy team.
Still, it seems silly that your fantasy team only gets points from its starters.
My guess is that the whole "name your starters" system is an anachronism. Once upon a time, six to 12 guys, probably armed with pen, paper, stats, and beer, decided "Hey, that whole fantasy baseball thing we're doing? Let's do it for football."
But applying one sport's framework to the other does not work. Tapping certain people as starters is pretty obvious if you are playing baseball, as reserves (with apologies to Dave Roberts) typically do not make much impact on the game. This is not true in football, as evidenced by three-receiver sets, the Minnesota Vikings backfield, and any quarterback in the Denny Green regime.
As historical error is not a good reason for your league to use a rule, get rid of it. Like the above section, there are a number of other ideas. I know of one league whose bench players earn points at a 50% discount compared to starters. Another allows you to use your bench for the purposes of breaking ties. But it is rare to see a league that does what, ideally, one should: ignore the "starter" nomenclature entirely.
There are practical reasons for doing this. In one keeper league last year, I inherited a team featuring (for lack of a better word) Jeff Garcia. Realizing he was a one-year solution, I drafted Carson Palmer. When Garcia went down, I luckily was unperturbed, having grabbed Kerry Collins earlier on.
But in earlier weeks, I was left with a Faustian choice. I could read my tea leaves, sacrifice a virgin, or break out the tarot cards, but choosing the right quarterback each week was impossible. Sure, I could go with what appeared to be the best matchup, but even that is unreliable. My best bet was to go with my gut and hope for the best.
This whole "matchup" shtick adds another layer of luck that skill games (such as, hey, fantasy football!) should avoid. So, do not employ it. Start virtually everyone.
At any given point in time, my long-standing keeper-league team has anywhere from two-thirds to four-fifths of its roster active. No, not all the players are eligible to earn points. It is much more elegant than that. On a weekly basis, my top scoring kicker and quarterback each start -- top scoring based on that week's results. My top two running backs and wide receivers also start. (Tight ends count as wide receivers and defenses are not used.) In the end, six players earn points for me.
Here is an example:
In the final week of our regular season, I had three kickers -- Phil Dawson, Lawrence Tynes, and Adam Vinatieri -- active. Dawson kicked a lone extra point, earning my team one point. Vinny had a much better game, to the tune of six points. But Tynes stole the show with a big fat 10.
As my league uses one slot for kicker, I earned ten points -- the ten by Tynes. Vinny's and Dawson's went to waste. Of course, the week before, it was Dawson who ended up as my "starter" (12 points to Vinny's 10 and Tynes' five); and one week before that, Vinny's nine bested Larry (seven) and Phil (one). Over a three-week period, I ended up with 31 points. Had I been forced to name starters, the most points I could have earned is that 31. But the fewest is a ridiculous seven. The swing there is immense and must be avoided.
One could argue that the odds of me getting only seven points is slim to none, as I would have almost certainly started Vinatieri each of the three weeks. True -- if forced to, I probably would have, and he would have earned a robust 25 points. But those of you who bask in this realization are missing the other big advantage of this system: In this league -- nine teams -- my third-string kicker actually has value. This goes against the conventional wisdom set by the run-of-the-mill leagues, wherein your top kicker is, by and large, a fungible asset. It is almost impossible to imagine a situation in the standard league format where you would be scrambling to draft a kicker; if anything, you would pick up a guy like Tynes or Dawson from the waiver wire. In my suggested format, every player counts.
By now, you have probably figured out that our roster sizes are a bit bigger than you would expect. First, you know that lackluster fantasy players like Phil Dawson and Chester Taylor are on my roster. Second, you know that I activated three kickers for one kicker spot, and, extrapolating that three-to-one ratio for my six point-earning slots, I have to start around 18 players. (Actually, it's 20.) Finally, I said that I start the majority of my lineup -- as little as two-thirds -- so a good guess would be a roster size of about 27.
That's fine. We'll stick with it, even though my league actually has roster sizes averaging 29. Anywhere in that ballpark is fine. But aim big. Why?
To start, as a friend of mine said, rationalizing his day-too-early selection of Barry Bonds: "[Fantasy] baseball is better than football -- because one injury does not mean the end to my season." Under typical rules, he is right. Anyone who banked on Michael Vick in '03, only to lose him to a leg injury, knows the feeling of watching your season flush down the bowl. Countering that is reason enough for exceptionally deep rosters. But it is not the best reason.
The primary reason for such a deep roster is because there are a lot of fantasy-relevant players out there, especially when your peak performer is the one earning you points. Every starting quarterback is worthy of a roster spot. All 32 kickers are too. The Warrick Dunn/T.J. Duckett backfield becomes two players of value instead of the typical zero. Many teams have three (or more!) wide receivers worth taking. A back-of-the-envelope calculation suggests that there are an absurd 200 players with value in this format, and that does not include prospects or backups (both of which tend to be owned, for good but different reasons).
What this does is make for a very active and interesting trade marketplace. Yes, it is just as difficult to trade a guy like Peyton Manning or LaDanian Tomlinson in this format as in the standard. (Then again, LT was traded before last year, so it can be done.) But in standard leagues, there are few opportunities to truly improve your team via trade -- giving up, say, Kevan Barlow all but requires getting a guy like Chris Brown in return. And that is not much of a deal.
Here is a list of some of the lesser players who were traded in my league last year:
Josh Reed, Keary Colbert, David Boston, Josh McCown, Rex Grossman, Drew Henson, Greg Jones, William Green, Matt Stover, Maurice Morris, Ernest Wilford, Dominic Rhodes, Eddie George, Troy Hambrick, and Justin Gage.
The majority of leagues out there would look at that list and wonder who those guys are and what they are doing on fantasy teams. We wonder what type of fantasy owner reacts that way -- good ones see a list like that and can understand that each of these players, at one point or another, rang of opportunity. This layer of complexity adds more skill to a game that thrives on it.
No one in my league entered the season with Reuben Droughns on his roster. In a league this deep, the demand for him was high -- very high. How does one determine who gets him?
Because of how many players are already taken when one enters the season -- in our case, 175 players are already spoken for -- there are going to be few free agents worth taking throughout the year. With that in mind, we (mostly randomly) decided that each player should get four free-agent tokens that are good throughout the year: they are awarded after our pre-season draft (discussed later) and expire before the Super Bowl.
The tokens work as such: After the pre-season draft, any players still left unclaimed become free agents. Using the pre-season draft order, we run a free-agent draft. Any owner with at least one free-agent token can, when it is his turn to pick, spend a token and select a player. If an owner lacks a token or does not wish to select someone, he passes. When everyone has passed at least once, the draft ends. Between the draft and the beginning of the first quarter of our season, it is a free-for-all, which has thankfully not turned into chaos.
The process repeats itself after every quarter, with the recently-completed quarter's results used as the draft order (worst to first). In Droughns' case, the team that finished last in Q1 selected him with the first pick in the subsequent free-agent draft -- and, being an expansion team with little chance of success in the current season, promptly traded him to the team that won that quarter for Dominic Rhodes (see above) and a draft pick.
Note that when adding a player to your roster, you do not cut someone -- roster sizes are fungible. There is simply no need -- in effect, you are dropping your token instead. (Yes, that means there is no maximum roster size -- in fact, rosters in my league ranged from 27-31 players at the close of last year.) Therefore, it follows that free-agent tokens are tradable commodities like anything else. Once again, we have the trend of making everything into a bargaining chip. He who can make the best deal is rewarded properly.
As adding as many types of trading chips seems to be in vogue in this piece, it should surprise no one that I am a strong advocate for making any league into a keeper league. We already have the ability to trade current assets (players) and nearly-realized assets (free-agent tokens) -- it only makes sense that one should be able to deal away (or for) future assets as well. In this case, prospects and draft picks.
It goes without saying that keeper leagues are difficult to pull off -- mostly because you need a large number of dedicated owners. This format makes it somewhat easier, with credit going to the high number of players on each roster. First and foremost, this format cannot support 14 or even 12 owners -- the range seems to be eight to 10. So you do not end up seeking out a lot of guys to fill your spots, and even more importantly, the marginally interested need not apply. Secondly, because of the (at least heretofore) uniqueness of the format, and certainly the possibilities a 29-player keeper league opens up, attracting a new owner to replace a leaving one is not very difficult.
How to set up the keeper aspect is more the issue. In order to keep guys like Rex Grossman and even Dominic Rhodes relevant, one needs to ensure that a lot of the 30-man roster is carried over into the next year. In practice, we have found that about 70% seems right. The breakdown we use? We keep up to 22 players for the next season. We then have a three-plus round pre-season draft. Finally, we have the four free-agent tokens previously discussed. (That is where the number 29 comes from.)
If you cut down to fewer than 22, you get an extra pick at the end of the draft for every cut deeper than the 22nd. For example, before the 2000 season, I somehow managed to cut my roster to 17 players -- and received five picks at the end of the draft (eight rounds total). Whether or not you want to employ that rule is to taste -- in our experience, it works well; then again we have never tried the hard-line "cut to exactly 22" stance.
The draft order is determined by looking at each team's average finish in the year prior, with the playoff teams and eventual champion picking at the end. If one team (eight-team league) finishes in fifth, sixth, fifth, and eighth in quarters one through four, respectively, that team had an average finish of sixth. Assuming that was the worst score, that team would pick first in each round of the draft.
The draft itself occurs right before the regular season. Oftentimes, there are a few players who fell through the cracks the previous year and of course the recently cut talent (or lack thereof) is out there for the picking. But most importantly, this is the first time recently drafted (NFL draft) rookies are eligible for the league. You can picture some of the wild trades that occur because of this -- a subpar team that owns a high draft pick and has a guy like Curtis Martin. It could rebuild its QB corps by dealing Martin for another #1 and taking both Alex Smith and Aaron Rodgers if the fates may have it. The amount of flexibility is incredible.
And with that, it starts all over again. We started this league as middle schoolers. In August, it enters its 12th year. In 1994, eight of us sat around and enclosed porch and drafted huge rosters -- 25 players apiece. Two of the original owners have left, but they were both replaced without much difficulty. One of those replacement owners also left; he was replaced with relative ease. Last year, we expanded to nine.
My team's Achilles' heel is its running game. Some teams have duos or trios of bonafide workhorses like Priest Holmes/Jamal Lewis/Ahman Green or Deuce McAllister/LT/Clinton Portis. My squad is headlined by Rudi Johnson with a supporting cast of Kevin Jones, Chris Brown, and, er, DeShaun Foster; my desire for a fifth potential starter is unmatched. So, when Chester Taylor -- my mid-season free-agent token acquisition -- was signed to an offer sheet on a team where maybe, just maybe, he would have a chance to start, well, you can understand why I was excited.
Don't you wish you were, too?
Dan Lewis is a Manhattan-based attorney and contributor to Pro Football Prospectus 2005.
72 comments, Last at 09 Aug 2005, 4:20pm by BknGen