|ESPN.com: Draft Kit||[Print without images]|
Remember when "targets" seemed like an exotic, over-fancy concept? I do. In the early 2000s, it was nearly impossible to find a list of the NFL pass-catchers who had the most footballs thrown their way. Stats like "yards after contact" and "yards at the catch" were once a glimmer in the wise pro football fan's eye, and now they're readily available and referred to ubiquitously. In fantasy circles, the same is becoming true for "Value-Based Drafting." In fact, when you Google that term, my name comes up on the first results page, so I guess I've done my part to popularize the concept. But make no mistake: The idea isn't mine. ProFootballReference.com has been using it for years, and they credit the wise souls at FootballGuys.com for its formulation.
Simply put, VBD developed as a way to retroactively compare the fantasy usefulness of players at different positions. It seeks to establish a "baseline" player at each position, and then measure the degree to which the best players at each position exceeded their respective baselines. It may have a fancy acronym, but it's the furthest thing imaginable from fancy. In fact, it's kind of a ramshackle, kludgy tool that may get more credit for prescience than it deserves. But I still really like it.
My friend and colleague Tristan Cockcroft and I recently had a revealing conversation about VBD. Tristan doesn't have a problem with my narrow definition of VBD above. What gets his blood boiling about the promulgation of VBD to every Tom, Dick and Harry fantasy web site is the way in which some folks try to turn it into a finely-tuned predictive tool. "Hey, the word 'drafting' is right there in the name!" goes the logic. "Of course it's supposed to be a tool that tells you who to draft!" Except Tristan's right: It's really not. VBD's biggest sin is probably its name. Call it RPA (Retroactive Positional Analysis). Call it JHGWRGBSA (Just How Good Was Rob Gronkowski's Breakout Season, Anyway?). In my opinion, VBD exists to help us process the season that's just passed, so that we can philosophically (not statistically) implement the lessons of that season moving forward.
How does VBD work? It's all about establishing that "baseline" player. Calculating the difference between each player's absolute fantasy point total and the point total of a baseline player at the same position gives us a relative number, which can then be compared across positions. Logically, this allows us to look at past-year performances, determining which positions tended to justify higher draft slots. After all, if we only judged by absolute point totals, we'd all be drafting quarterbacks in the first round, since for example in 2011, eight of the top 10 highest raw fantasy point totals were scored by QBs.
Intelligent folks have different ways of thinking about how to determine baselines for each position. Some like to think of baseline as "replacement-level," as we sometimes do in baseball, and in a 10-team league, for example, that leads us to the No. 11 QB as our baseline (since the top 10 would presumably start for each of the 10 fantasy squads). Frank DuPont (who wrote this e-book) figures baseline players by adding up the total number of man-games your fantasy league requires to fill out a full season at each position, then calculating the average number of players at each position it usually takes to produce that many man-games. FootballGuys prefers a "Top 100" methodology, whereby we establish how many players at each position tend to get drafted in the overall top 100 of your draft, and use that number as our baseline for each position. And there are several other approaches, often created by smart fantasy owners who simply tinker with stats inside their own leagues.Personally, I tend toward the "10 Round" theory (which, to be fair, is simply an offshoot of the "Top 100" methodology): I look at the number of players at each position that typically gets drafted in the first 10 rounds of any league's draft, and however many QBs that is, that's my baseline QB. However many RBs that is, that's my baseline RB. And so on. So if you're in a 10-team league, you'd look at the top 100 players drafted. If you're in a 12-teamer, you'd look at the top 120. Why do I feel 10 rounds works well? To some extent, it's simply trial and error (me Excel pretty good). But philosophically, for me 10 rounds is the point where fantasy starting lineups are set, the top rookies and hyped sleepers we hope could work their way into being fantasy starters have been drafted, and we're now looking for backup (or "replacement") type players. Hey, I said it's kludgy, right?
The truth is, though, that any infighting about establishing the best baseline methodology feels a bit like a Talmudic inquisition, where the principals spend time parsing semicolons. I've run most conceptions of VBD baselines for multiple years' worth of data, and the ranks that result usually look quite similar. No matter how you do your baseline, there's room for you under my big VBD tent, because our results are going to bear far more than a passing resemblance to one another.
Here's what the top 20 raw fantasy point totals looked like at the end of last season:
Any fantasy veteran will tell you that any system that tells you Mark Sanchez retroactively should've been a second-round pick probably isn't a very good system. Because so many QBs are relatively similar to one another, there's far less scarcity among QBs than at other positions, so you can afford to wait on drafting your signal-caller, goes the logic. What VBD does is inject sanity when comparing players at different positions. For a 10-team league, recent history tells us that the positional breakdown in the first 10 rounds of the average draft goes like this: 16 QBs, 38 RBs, 33 WRs, 9 TEs and 4 D/STs. So for a 10-team league, our baselines for '11 will be the No. 16 QB (Joe Flacco), the No. 38 RB (Toby Gerhart), the No. 33 WR (Malcom Floyd), the No. 9 TE (Brent Celek) and the No. 4 D/ST (Seahawks). We'll subtract the raw fantasy point totals that each of these guys scored in '11 from the point totals of all the men above them at their respective positions. (For example, Aaron Rodgers' VBD total in a 10-team league is his own raw point total minus Flacco's raw point total.) And we wind up with this.
For comparison's sake, if you're in a 12-team league, the baselines are somewhat different (remember: I use the average number of players at each position drafted in the first 10 rounds, which in this case should total 120 players): 17 QBs, 43 RBs, 40 WRs, 12 TEs, 6 D/STs and 2 Ks. So for 12-team leagues, last season's most valuable 20 fantasy players were:
Clearly, these two sets of ranks for '11 are pretty darned close. In each case, we see five QBs (all in the top 10, which is unprecedented) and four wide receivers, with Michael Bush and Jimmy Graham trading places.
What conclusions can we draw from the '11 lists above? Not, I don't believe, that you suddenly must have an elite QB, or that Rob Gronkowski is automatically a second-round fantasy pick. I honestly don't believe you can use VBD as that kind of projection tool. The NFL changes so much on a yearly basis -- because of scheme changes, because of injuries, because of sheer blind luck -- that assuming extraordinary seasons will repeat tends to be a losing game.
No, I think what this analysis does for us is assess where excellent seasons tend to fall, regardless of which player submits them. We've just seen Gronkowski and Graham produce two of the greatest TE fantasy seasons ever, with the main difference between them being Gronk scoring 18 TDs and Graham finding the end zone "only" 11 times. VBD tells me that the going rate for this kind of extraordinary TE production still shouldn't be a first-round draft pick. And because I'm dubious that Gronkowski will approach 18 TDs in '12 -- it seems to me that a low-double-digit total is much more reasonable -- the earliest you should consider taking him or Graham is late in the second or early in the third of a 10-team draft.
Do I think Jordy Nelson or Victor Cruz will appear on my VBD top 20 list once the '12 season is over? I don't; I think each player was helped by unsustainable big-play rates in '11. But for me the larger VBD WR point from the past several seasons is how much of a "little brother" the WR position continues to be. Here's the number of RBs and WRs who've wound up in the VBD top 20 of a 10-team league in the past five years.
2011: 10 RB, 4 WR
2010: 15 RB, 2 WR
2009: 10 RB, 6 WR
2008: 12 RB, 3 WR
2007: 10 RB, 6 WR
To me, this trend means that other things being equal, I still want to sniff out the best early-round running backs rather than the best early-round wide receivers, if I can.
And finally, there's the question of what to do with quarterbacks like Rodgers, Brees, Newton, Brady and Stafford, who as a group outpaced their baseline QB (Flacco) by an unprecedented distance. In the past 10 years, there hadn't been a season where more than two quarterbacks wound up in the VBD top 10, and last year we had five? Perhaps this can be explained away by high-profile injuries (Peyton Manning and Michael Vick come to mind) skewing the baseline downward, or perhaps this reflects a schism in the NFL, where play-calling grows ever more aggressive for teams with elite QBs and ever more conservative for teams without them. Perhaps the most fascinating storyline of '12 will be whether these top quarterbacks can continue to post running back-like VBD numbers. Based on recent history, I find the notion somewhat dubious.
Any system that seeks to compare relative values between positions needs to start from a baseline of statistical projections, and NFL statistical projections are basically impossible to get right. (Take it from someone who helps create them for a living.) Yes, we can do a passable job with yardage totals for players who don't suffer unexpected injuries or depth-chart pratfalls. But so much of fantasy football hinges on touchdowns, and touchdowns are impossibly difficult to predict from season to season (let alone week to week).
For the sake of thoroughness, however, I'll plug ESPN's statistical projections for all NFL skill position players into the VBD spreadsheet, for a 10-team league, and show you what I get. Take this list with the world's largest grain of salt, because it's based in large part on trying to guess how many times players accidentally fall into the end zone.
As you can see, according to the letter of the VBD law, ESPN's projections think those same five QBs will continue to be awesome (and Eli Manning will join them), while Gronkowski and Graham will fly high and be justifiable as second-round picks in 10-team leagues, and the WR position may be deep but a bit flat near the top, after Megatron.
But of course, the rub is that while VBD is philosophically pretty good talking about, for instance, how many running backs we might expect to wind up among the 20 most valuable players in fantasy football circa '12, it is basically useless at telling us specifically which running backs. VBD doesn't account for which players are major injury risks, or which could be subject to losing their starting gigs. It doesn't account for week-to-week safety and stability, or potential platoons. And these are the things excellent fantasy owners (and analysts) must interpret year-round, which is why in the end projecting for the '12 season is as much art as science.
Come draft time, when your turn to pick arrives, you'll be faced with your own personal pressure points. Your gut might tell you that drafting great running backs wins fantasy championships, but where's the breaking point at which it no longer makes sense to take a shaky RB? Where should you consider drafting a rock-star quarterback, or a proven and safe wide receiver, or a tight end coming off an all-time-great season? That's where an analysis of past seasons via the lens of VBD is helpful. For example, if I'm drafting late in the first round of any-sized league, I might look at Rob Gronkowski's massive '11 campaign and say, "Man, if he can even come close to that kind of production again, I'm totally justified in grabbing him right now!" But VBD tells me that isn't so. Can VBD by itself tell you which player to draft? No, because in the end all it's doing is folding together multi-positional data into one list. And because touchdown projections are so difficult to make, it tends to be a "garbage-in, garbage-out" situation: Your VBD ranks will only be as smart and accurate as your stat projections. My advice, therefore, is to heed the lessons of past seasons that VBD teaches us. "This is what a great tight end season looks like," VBD says. "When quarterbacks separate themselves from the pack, this is where you're justified in taking them. This is why you should still feel a default pull toward taking a running back." Those are useful lessons, but of course, they're just a few arrows in your fantasy football quiver.