Grigs Allmoon said:
bostonfred said:
Adam Harstad said:
The average VBD of the top 3 RBs over the last five years (non-PPR) is 157, 161, 134, 134, and 154 (overall average of 148).
The VBD of the #4 RB over the last five years (non-PPR) is 113, 130, 108, 103, and 102 (overall average of 111).
In other words, the typical "top 3 RB" has been 28% more valuable than the typical "4th best RB" over the last five years. Based on that, the gamble starts making sense at much less than 95% confidence levels. If the only options were "top 3 or bust (i.e. 0 VBD)", the gamble would make sense if you were just 78% confident in your ability to identify a top-3 RB (assuming that "top 3 RB" would be equally likely to turn out as RB1, RB2, or RB3). Since the options aren't really "top 3 or bust"- since it's possible to wind up with RB12, or RB18, instead- the real confidence levels have to be much lower than that. I'd say even as low as around 50% confidence, a strong case could be made for rolling the dice.
Why is that?
I've been meaning to go back over that math in detail, myself, but until then I'll just say that even 78% seems impossible. I'd bet even the most accurate prognosticator can't even get 2 out of the top 3 from year to year. (I'd have to do some research to come up with an actual estimate. Maybe later...)
ETA: My thinking on that last sentence is a bit twisted, as you have to pick one to finish in the top three, as opposed to picking 3 and hitting on 78%, but it's kind of similar.
I would agree that
78%75% would be an impossibly high confidence level to achieve if we're being honest with ourselves.
But, again, the 75% figure assumes that any back who misses the top 3 automatically provides 0 VBD, which is obviously absurd. If we assumed that the two possibilities were either getting a top-3 back who provided 148 VBD or a decent RB2 who provided 40 VBD, then suddenly the break-even confidence level becomes 65.7%. If we opened the door to the whole array of possibilities (a much more complicated math problem than I feel interested in doing), my guess is that we'd find a break-even point somewhere between 50% and 60%.
Also, remember that this assumes that the value of each point of VBD scales linearly, which simply isn't true. A guy worth 200 VBD is not just twice as valuable as two other guys worth 100 VBD each. 148 VBD might be 33% more than 111 VBD, but that doesn't mean that a 148 VBD player is just 33% more valuable than a 111 VBD player. If we go through the complex mathematics to determine true value, we might find out that he's actually 50% more valuable. In that case, the break-even confidence level drops to 66% in the "studs or bust" scenario (where the only two possibilities are 148 VBD or 0 VBD), and 55% in the "studs or RB2" scenario (where the only two possibilities are 148 VBD or 40 VBD), although that math again simplifies the scenario and assumes that VBD increases linearly from 40 to 111. Hopefully I'm managing to illustrate just how thorny the concept of a "break-even probability" really is. I wouldn't even attempt to calculate it with anything resembling accuracy. I consider myself an educated lay-person when it comes to applying math like that, but that kind of problem would be way above my paygrade.
And even the concept of a "break-even confidence level" ignores the fact that fantasy football favors high-variance strategies. If you had one strategy that gave you a 50% chance of finishing 1st and a 50% chance of finishing last, and another strategy that gave you a 100% chance of finishing second, you should go with the first strategy. The second strategy will, on average, provide substantially better outcomes... but the point isn't to maximize your average finish, it's to maximize your number of first-place finishes- a subtle but important distinction. So we should be willing to roll the dice even if we fall short of the truly calculated "break-even confidence level", simply because of our predilection for variance.
Adding together all of the crazy variables in play, and I could see rolling the dice if your confidence in your ability to land a top-3 RB was somewhere in the 50-66% range. Can someone reasonably hope to achieve that level of confidence? That's a completely different (but still very interesting) discussion. I, personally, am not that confident in my ability, so I'd prefer the "sure thing" RB4... but I think it's not at all unreasonable for others to feel differently.
Maybe if I have time later I'll take a look at pre-season ADP and see, historically, what the odds have been of the preseason RB1 finishing among the top 3 at the end of the season. Or another approach would be to look at the average VBD of #1 draft picks. There are lots of ways to approach the problem, any one of which will offer a passable approximation of the hypothetical, even if none of them can definitely provide an answer.
