The RotoViz Guide to Value Based Auctions – Part I
Auction drafts for fantasy football leagues may not be the standard format, but I think they’re a killer enhancement to the hobby. For the true fantasy football addict, you just can’t beat the idea of going into your auction and being able to create just about any team you can dream up. If you are the kind of player who just wants your guy? You got get ‘em! No worrying about whether he falls or not. No worrying about having to reach. Just get him. Or perhaps you’re the ultimate value investor. You love to sit and wait, lurking in the shadows until everyone else has spent too much money and the board is just littered with undervalued studs. You get to dominate the middle portion of the auction as your leaguemates just watch in horror while you stockpile talent. The possibilities are endless.
The flip side of the coin is that precisely because auctions have that ebb and flow that snake drafts do not, they can be very tricky. Where should you spend your money? Do you go with the stud QB? Should you stockpile running backs and wide receivers? What’s the best way to gain an edge over your leaguemates? They seem like difficult questions, but I have some answers for you.
If you’re doing the prep work yourself, you may have tried taking some pre-season projections and some average auction values from any of the sites out there and created a Points-per-Dollar calculation. This is a logical way to look at things, since you’re trying to maximize the points on your roster, given your budget. Here’s what the top players from each position look like on the points-per-dollar (PPD) metric. (To keep things standardized I’m using the projections and 10 team auction value data from FantasyPros.com that feed the Salary Cap App, and I’ve set the scoring to 0 PPR for a “standard” league.)
|Player||Projected Points||Proj Auction Value||Points/$|
|Aaron Rodgers (GB, QB)||350.62||$44||7.97|
|Adrian Peterson (MIN, RB)||274.35||$63||4.35|
|Calvin Johnson (DET, WR)||225.11||$49||4.59|
|Jimmy Graham (NO, TE)||170.66||$34||5.02|
The initial (and natural) reaction might be “I should be sure to get Aaron Rodgers because he’s the best use of my dollars.” I’m going to show you not only why that logic is flawed but also why Aaron Rodgers is actually the worst use of your auction dollars out of these four players.
You may know where I’m going with this, so rather than reinvent the wheel on exactly why the Late Round QB strategy works so well in fantasy football, I’ll point your attention to what I consider the definitive resource on the topic: “The Late Round Quarterback: 2013 Edition” by JJ Zachariason. In that book, Zachariason decomposes the success of the Late Round QB strategy into chapters about relative positional value, player projections, supply and demand, and the impact of weekly scoring. There are several formulas and tables in the book that are elegant in their simplicity yet do a fantastic job illustrating why the strategy works. There’s even a chapter dedicated to employing the strategy in an auction league. I encourage everyone to check out the book and the website to see for yourself why you should be using this strategy to dominate all of your fantasy leagues.
With that knowledge in your back pocket, I hope to expand on Zachariason’s work and show you the massive advantage you can gain over your leaguemates when you employ the equivalent of the Late Round QB strategy in auction leagues. I like to think of it like this: What you’re actually doing by “waiting” for one of the cheaper QBs in an auction (i.e. spending less of your budget than the rest of your league) is employing “Late Round QB on Steroids.”
If you stop and think about it, a snake draft and an auction aren’t really all that different except for the order in which players get picked. In both cases you have a certain amount of capital (draft picks or dollars) to spend to create your team. In a snake draft, you burn your high picks on the players you think will score more points than all the remaining players at their position. The same should hold true for auction – you burn the bigger chunk of your dollars on the players you think will score more points than all the remaining players at their position. Equally as important as the player you DO pick is the player you DON’T pick, also known as “opportunity cost”. Each pick you make in a snake draft carries the opportunity cost of NOT drafting every other player who goes after him since most of those guys will end up on your competitors’ teams. Likewise with an auction, every player you spend money on also carries the opportunity cost of NOT being able to spend that money on all other players. Which, again drives home the point – you invest the biggest chunks of your capital (draft picks or auction dollars) on the players you think will score the biggest margin of points over all of the other available players at their position.
So, let’s tie this opportunity cost concept back to the original example where we looked at the Points-per-Dollar analysis of the top players. What we need to look at here is not just the points-per-dollar of the players we’re buying but also the points-per-dollar of all of the other players we could be buying with that money. So what we really need to look at is relative production and relative cost as opposed to absolutes.
Using the same data from FantasyPros.com we can look at the projected points for each position. If we let the top scoring player at each position be 1, we can measure each successive player as a percent of the top player. We’ll call it Relative Points. For example, Calvin Johnson is projected to be the #1 WR in 2013 with 225 points. His Relative Points is 1 (225/225 = 1). A.J. Green is projected to be the WR2 with 202 points. His Relative Points is .90 (202/225 = .90).
We can look at the average auction values from Fantasy Pros and do the same relative calculation. Calvin Johnson gets the top dollar value of $49 and we’ll let that be a Relative Price of 1. ($49/$49 = 1). Then we can then measure every other WR’s price relative to Calvin. A.J. Green has the 2nd highest price of $44 for a Calvin-Relative Price of .90 ($44/$49 = 0.90). We iterate this across the rest of the WRs and then across the rest of all of the other fantasy positions.
Note: Because rankings (and auction values) can sometimes differ a bit from projections – due to the concepts of upside, downside, and consistency – I took liberty of reassigning the auction values in descending order to match the projections. For example, Andre Johnson has a point projection of 176 which puts him at WR8, but he’s the WR7 in terms of average auction value. I’ve assigned him the WR8 price to represent the dollar values that go with the positions. The adjustments are small (only a dollar or two here and there) but I wanted to give full disclosure in case you noticed that any of the numbers don’t match up exactly with what’s in the Salary Cap App.
Expanding on this notion of relativity, if we divide each player’s Relative Points by their Relative Price we get what I like to think of as Leverage – a term borrowed from investing. In the portfolio management sense (which is essentially what creating a fantasy football team is) Leverage involves borrowing money to gain extra market exposure, thus amplifying gains or losses. The same concept applies here. For our purposes Leverage tells you how far you’re stretching your precious auction dollars by selecting each player. You’re financing the purchase of more expensive players (RB and WR) by “borrowing” from players with greater Leverage (QB and TE). The more points you can “finance” really cheaply (via say QB12), the more you can invest in high cost players. So, a higher Leverage number is better. Remember, we’re really trying to find out what the opportunity cost is of selecting each player. So on a position-by-position how much Leverage are we gaining by taking the cheaper player?
Here’s what the top 5 players at each position look like:
|Player||Projected Points||Proj Auction Value||Points/$||Relative Pts||Relative $||Leverage|
|Aaron Rodgers (GB, QB)||350.62||$44||7.97||1.00||1.00||1.00|
|Drew Brees (NO, QB)||339.87||$43||7.90||0.97||0.98||0.99|
|Cam Newton (CAR, QB)||320.35||$39||8.21||0.91||0.89||1.03|
|Peyton Manning (DEN, QB)||319.17||$35||9.12||0.91||0.80||1.14|
|Tom Brady (NE, QB)||302.26||$33||9.16||0.86||0.75||1.15|
|Adrian Peterson (MIN, RB)||274.35||$63||4.35||1.00||1.00||1.00|
|Arian Foster (HOU, RB)||259.69||$61||4.26||0.95||0.97||0.98|
|Doug Martin (TB, RB)||239.81||$57||4.21||0.87||0.90||0.97|
|Jamaal Charles (KC, RB)||238.47||$55||4.34||0.87||0.87||1.00|
|Marshawn Lynch (SEA, RB)||229.12||$53||4.32||0.84||0.84||0.99|
|Calvin Johnson (DET, WR)||225.11||$49||4.59||1.00||1.00||1.00|
|A.J. Green (CIN, WR)||201.80||$44||4.59||0.90||0.90||1.00|
|Dez Bryant (DAL, WR)||199.15||$42||4.74||0.88||0.86||1.03|
|Brandon Marshall (CHI, WR)||195.68||$41||4.77||0.87||0.84||1.04|
|Julio Jones (ATL, WR)||192.75||$36||5.35||0.86||0.73||1.17|
|Jimmy Graham (NO, TE)||170.66||$34||5.02||1.00||1.00||1.00|
|Rob Gronkowski (NE, TE)||160.73||$26||6.18||0.94||0.76||1.23|
|Tony Gonzalez (ATL, TE)||132.07||$12||11.01||0.77||0.35||2.19|
|Jason Witten (DAL, TE)||127.13||$10||12.71||0.74||0.29||2.53|
|Vernon Davis (SF, TE)||119.98||$7||17.14||0.70||0.21||3.41|
What this is telling us is that for Julio Jones you’re expecting to get 86% percent of Megatron’s production while paying 73% of Megatron’s price. Or, put another way, Julio’s 5.35 points per dollar is 1.17x Megatron’s 4.59 points per dollar. That 1.17x factor is Julio’s Leverage. The dollars you spend on Julio give you 1.17x the points as the dollars you spend on Megatron. You get the point…
What you start to notice is how the Leverage increases pretty rapidly with TE relative to the other three positions. You stretch your dollars much further by “financing” those Vernon Davis points cheaply (3.41 Leverage) than you do trying to save some dough by passing on AP for Marshawn Lynch (whose Leverage is basically just 1). QB exhibits a Leverage increase also, but not quite as rapid as TE. More on that in minute.
So, now you start to see: The reason I like to think of “Late Round QB on Steroids” in auction leagues is because of what you can do with the extra money you save by selecting the cheaper player. In a snake draft if you pass on Aaron Rodgers you get to draft a second (or third) round RB/WR. If you pass on Peyton Manning, you get to draft a 4th round RB/WR, and so on. You get the value of a player in that round and no more because snake draft picks decline in value in a linear fashion. Auction leagues are different. The way the price curves fall off toward the “later” picks means your waiting rewards you even more than it does in a snake draft. To put in into snake draft terms: if you Pass on Aaron Rodgers and draft Tony Romo in an auction league, you get the equivalent of an extra 3rd round pick for free! You can either spend that 3rd round pick on a 3rd round player or you can use it to bump your 2nd to a 1st, or bump a 4th to a 3rd and a 5th to a 4th, and so on. It’s the logarithmic shape of the price curves in auction leagues is the steroids that juices the Late Round QB strategy. Let me demonstrate.
Here are the Relative Points and Relative Price curves plotted against each other for each of the four positions. The area between the curves represents the points-per-dollar value you gain by waiting for the lower-priced player:
Quarterback (Aaron Rodgers would be QB1)
Running Back (Adrian Peterson = RB1)
Wide Receiver (Megatron = WR1)
Tight End (Jimmy Graham = TE1)
You start to get the sense of how that area between the relative points and relative price curves increases more rapidly for QB and TE relative to RB and WR, but it’s a little difficult to tell exactly how different they are. Now here’s the fun part – I’m going to share with you what I like to call the “Rosetta Stone” of auction drafting. Bringing it all together, we can plot the Leverage for each position (which represents that difference between points and dollars) on the Y axis and let the X axis be each player starting with the #1 scorer and going out to the #60 scorer (QBs and TEs are truncated at 30). Here’s what we see:
Immediately, you can see how far you stretch your dollars when you “wait” on QB and select one of the cheaper options. Yes, you are sacrificing some projected points by moving down from QB1 to QB12, but on a relative basis you’re more than being made up for the drop in production by the drop in price. The dollars you spend on Tony Romo are 2.89 times as powerful as the dollars you would have spent on Aaron Rodgers because you’re getting 79% of ARod’s production for only 27% of ARod’s price.
Meanwhile, the RB and WR markets are much more efficient. The price break you’re getting by taking the 12th RB is very much in line with the drop you’re going to see in production. If you buy Steven Jackson you pay 56% of AP’s price, but you’re only getting 72% of AP’s production. Steven Jackson’s Leverage is only 1.30. You have to spend the dollars in the RB and WR market to get the points. So what you do is, take advantage of the inefficiency of the QB market in order to pay closer to fair value in the WR and RB market. What this means is you probably have a slight disadvantage to your leaguemates with respect to QB (where there is only one weekly starter per team) but you have a bigger advantage at WR and/or RB where you start up to three players per week (assuming flex). And with all of the great writers at RotoViz helping you identify not one, not two, but three stealth stars at QB, you may not even have a disadvantage at the QB position at all!
Note: Obviously the same can be said about waiting on TE (and to an even greater extent than QB). The reason that I focus on QB so much is because they tend to go for much more money than TEs do in general, so what you save on QB on an absolute basis can help you much more than what you would save on TE.
Lest the inefficiency we’re seeing be simply a function of the 10 team league, I ran the numbers with FantasyPros.com 12 team data and the chart looks almost identical:
Is it possible the inefficiency is just a 2013 phenomenon? I say not a chance. I don’t have any formal numbers on this, but from my own personal experience participating in auction leagues for the past 4 years, the the shape of the price curve for each position stays pretty constant. If you look at either average auction value data or expert consensus values they all tend to stay roughly the same for each position too. It’s not as if one year you find top QBs going for no more than 10% and another year they’re going for 25%. There are slight variations, but the curves and relative prices for the positions remain fairly constant.
To me, the inefficiency is exactly the same one that persists in snake drafts where QBs still tend to go too early on the perception of “safety” or the simple “they score the most points” argument. There’s surely a supply/demand effect going on as well. Despite the fact that the 25th quarterback is projected to score almost as much as the 15th, the oversupply drives prices down rapidly. No one wants to pay up for their backup QB. But as long as those price curves stay constant, you can continue to use it to your advantage.
Another great part about this strategy is that all of this assumes perfect forecasting, which we know is far from the truth. The projections we have in the Salary Cap App are the best out there. But consensus opinion (as proxied by ADP) will still drive the price curves. And we KNOW that a large component of ADP is based upon the previous season’s production, even though that has little predictive value for the upcoming season. So, with the help of Similarity Scores and all of the research and actionable information here at RotoViz you’ll be able to improve your odds of landing cheaper players who will outperform their ADP and avoiding expensive players who will underperform. In short, you will absolutely DOMINATE your league.
In the next installment in the series I’ll use the Salary Cap App to take a look at the impact that the price of your QB can have on the look of your team.