Fantrax Home NFL Home

NFL Football Column: Fantasy Football

More NFL columns
June 14, 2012
Bookmark and Share

Fantasy Football for Smart People

Jonathan Bales
Senior Writer, Fantrax
For the past few years, I've performed stat-based analysis on a variety of aspects of the National Football League, providing content to the New York Times and Dallas Cowboys. At the start of 2012, I decided to turn my attention toward Fantasy football. The result was Fantasy Football for Smart People: How to Dominate Your Draft.

You can purchase the book at the previous link or on Amazon.

Below, I'm giving Fantrax users a glimpse into the book. Users of Fantrax aren't your ordinary ESPN or Yahoo Fantasy sports owners. You folks require indepth analysis, and that's exactly what I've provided in my book. Enjoy the sample, and I will be back throughout the summer to post more unique Fantasy football coverage. 

Tiering Up: How to Create Basic Projections and Tiered Rankings

I’ve talked a bit about player stat projections and their importance in our rankings, but why make projections at all?  Why not simply rank players at their respective positions based on last year’s final rankings and a few subjective factors?

The most obvious answer is stat projections simply help with our rankings.  By projecting a player’s yards, touchdowns, and so on, it becomes far easier to determine how he ranks among his peers. 

This is particularly true with players whose stats don’t necessarily match up with their true value to their football team.  Back before Michael Vick went to jail, for example, he was always a player whose fantasy value didn’t match up with his perceived value.  With the exception of a few stretches of incredible play and even with his rushing prowess, Vick was never really a dominating figure in fantasy football.  Despite that, owners continued to overdraft him because we all knew he was valuable to the Atlanta Falcons.

Another reason it is smart to perform stat projections is the ease with which you can then rank and re-rank players based on league scoring.  Michael Turner is a solid running back in traditional fantasy formats, but in leagues with point-per-reception scoring, he’s basically a third option at the position.  Projecting stats and later developing rankings from those projections simplifies the process. 

The most important reason we need to make projections, though, ties back in with the idea of VORP.  Remember, VORP, or Value Over Replacement Player, suggests we should identify the biggest gap in points between current draft considerations and replacement players at their respective positions who could be drafted later. 

The example between wide receivers and tight ends that I used in the tight end section of my analysis of season-to-season consistency is an example of an employment of VORP.  In effect, VORP is the temporary bypassing of maximum points for greater overall points down the road.  Remember, since fantasy football requires the selection of players from multiple positions, any worthwhile draft strategy must possess an overarching vision; draft strategies like “Best Player Available” are too shortsighted, unnecessarily limiting the projected points you can acquire down the road in favor of more now.

Without projections (or some sort of rating system), VORP draft strategy is impossible.  A value system is necessary to decipher the “worth” of a player.  We can rank players all day to help us compare players within particular positions, but a comparison of players that play different positions is worthless without a rating system. 

How to Add Positional Consistency Into Projections

One quick and easy method to implement positional consistency into player ratings is to multiply projected points for a position by the correlational strength of consistency.  That is:

C(P), where C is correlational strength and P is projected points

As I stated earlier, these correlations are 0.62 for tight ends, 0.60 for quarterbacks, 0.48 for running backs, and 0.42 for wide receivers. Thus, if we project a quarterback to score 300 points and a wide receiver to score 200, those value shift to 0.60(300)=180 and 0.42(200)=84, respectively.  Note that those numbers aren’t projected points, but rather weighted values that make comparisons of various players easier.

The primary problem with this method, in my view, is it values the consistent positions too greatly, widening the “scarcity” gap at these spots.  For example, if we assume the 300-point quarterback and 200-point receiver are the top players at their respective positions and that our second-ranked players were projected to score 285 (QB) and 190 (WR), the above formula would change those projections to 171 and 79. 

Whereas the second-ranked players were projected to score five percent less than their top-ranked counterparts in the original projections, the new consistency-infused projections have the second quarterback still at five percent behind the top signal-caller, but the second receiver 6.0 percent behind the top pass-catcher.  In effect, multiplying position correlation strength by projected points increases the “scarcity” of the most consistent positions, improperly inflating their worth.

To compensate for this effect on scarcity, we can use a new formula that incorporates average points for each position.  To obtain better projections, we can multiply the difference in projected points and average points by the aforementioned correlational strength of each position, then add that number to the average pointsThat is:

C(P-A) + A, where C is correlational strength, P is projected points, and A is average points for fantasy starters at the position

Let’s assume we project a tight end to score 200 points and a wide receiver to score 220 points, with the average at the positions being 150 and 180, respectively.  We could factor positional consistency into those projections by multiplying the difference between the projection and the average by 0.62 and 0.42, respectively.  Our new projections would be:

Tight End: 0.62(200-150) + 150 = 181

Wide Receiver: 0.42(220-180) + 180 = 197

Since the positional scoring mean is incorporated into the formula, we can effectively control the effect of inflated scarcity that plagued the initial formula.

Using Projections to Create Tiers

Ranking your players into tiers is the easiest way to capitalize on VORP draft strategy, as it is a quick and effective way to recognize position scarcity.  Adding tiers to your draft board is relatively straightforward, as you are looking for big jumps in projected points.  Of course, things like season-to-season consistency and risk play a role, but the idea is to separate players based on the likelihood they perform well for you. 

Let’s take an example.  For simplicity’s sake, let’s use the same quarterback projections I created earlier:

Drew Brees – 400 points Risk: 6/10

Aaron Rodgers – 395 points Risk: 1/10

Peyton Manning – 390 points Risk: 9/10

Tom Brady – 385 points Risk: 3/10

Cam Newton – 360 points Risk: 8/10

Matthew Stafford – 355 points Risk: 5/10

Tony Romo – 350 points Risk: 3/10

In this situation, it’s pretty easy to see we should sort quarterbacks 1-4 into one tier, with quarterbacks 5-7 in another.  But what if we were to swap the projected points for Tom Brady and Cam Newton?  Brady would rank fifth with 360 projected points, but he also carries very little risk. 

In such a scenario, we could place Brady into the top tier, even potentially ranking him ahead of Newton based on his limited risk.  There’s no objective way to determine if you should place Brady at No. 4 or No. 5, but the point is things other than projected points should factor into your creation of tiers.  Otherwise, we would mistakenly place Brady in the second tier and potentially miss out on a stud quarterback. 

Tiers and Position Scarcity Can Lead to Counterintuitive Draft Decisions

It’s the sixth round of your draft, and you’ve managed to land Ray Rice, Jimmy Graham, Hakeem Nicks, Fred Jackson, and Percy Harvin in the first five rounds.  You’re sitting pretty, but you really need a top signal-caller to cap off a ridiculous starting lineup.  Your board looks like this:

QB Matt Ryan – 300 projected points (Tier 3)

QB Tony Romo – 295 projected points (Tier 3)

QB Philip Rivers – 290 projected points (Tier 3)

QB Eli Manning – 290 projected points (Tier 3)

RB DeAngelo Williams – 225 projected points (Tier 4)

WR Eric Decker – 180 projected points (Tier 6)

RB Peyton Hillis – 200 projected points (Tier 5)

With Matt Ryan sitting there and capable of rounding out your starting lineup, it seems impossible to pass on the Atlanta Falcon.  It would be a huge mistake, however, as the correct choice is DeAngelo Williams.  Despite already landing two running backs, including Ray Rice, Williams offers the most value. 

Again, let’s allow the math to show why.  Assume you draft Ryan and his 300 projected points, content to gamble on Peyton Hillis in the seventh round.  You know Williams will be off the board by the time your next pick rolls around, but you can’t pass on your top-rated player.  This Best Player Available draft strategy lands you 500 total projected points in Rounds 6-7. 

If you utilize VORP and stick to your tiers, however, you can bypass maximum value to acquire Williams and his 225 projected points.  If the draft stopped after six rounds, you might be in trouble.  Luckily for you, it’s a bit longer. 

With eight quarterbacks drafted by eight different owners prior to your selection, you can be fairly certain one of your Tier 3 quarterbacks, Ryan, Romo, Rivers, or Manning, will be available in the seventh round.  Despite obtaining only 225 projected points in the sixth round, your total for Rounds 6-7, using tier-based VORP draft strategy, would be 515 points even if you landed your last quarterback in Tier 3—15 points more than the Best Player Available draft strategy.

Thus, even though VORP can seem counterintuitive, in the long run, the strategy leads to better overall draft results than BPA.

Understanding Opponent Beliefs with Game Theory

As I mentioned in my intro, game theory is a strategic decision-making process that applies to zero-sum games wherein the beliefs of others affect your decisions.  Game theory has extreme implications to fantasy football draft strategies.

Thus far, I’ve assumed other owners in your fantasy leagues are completely rational.  Obviously this isn’t true, and in each fantasy draft there will be tons of “reaches” when players are selected far higher than consensus rankings.  This doesn’t mean the player chosen is a poor one, but rather that the choice is probably earlier than any other owner would have drafted him.

One of your jobs as a fantasy owner is to determine when other owners will draft specific players, both on a league-wide scale and an individual one.  With regards to the latter, this can be completed by simply talking to other owners, figuring out which positions they value, and so on. 

Unless you are in an expert league, you can be sure no other owners are using VORP draft strategy.  This means they will be far more likely to fill in their starters prior to selecting any bench players, and you can use this knowledge to your advantage.  When Uncle Bruce has selected a running back, wide receiver, and running back in the first three rounds, you can be sure he isn’t taking another running back in the fourth. 

Game Theory on a League-Wide Scale: Average Draft Position

You can use game theory in a broader way by projecting the draft positions of each player.  One way to do this is a mock draft, although individual mock drafts are susceptible to fluctuations and randomness in a big way.  Change one draft pick in an individual draft and you can drastically alter subsequent picks. 

This is where Average Draft Position (ADP) comes in.  ADP is simply a calculation of where each player is getting drafted in a given year.  ADP is based on thousands of mock drafts, so the “noise” of an individual mock is leveled out.  With ADP, you can get an idea of when Uncle Bruce plans to take the top rookie running back or the quarterback coming off ACL surgery.  Since ADP is based on real mock drafts and the owners selecting in these drafts use the same draft guides as Uncle Bruce, it can give you a solid understanding of the intentions of your league owners. 

Game theory, and ADP in particular, is the reason my dad was unjustified in selecting Matt Forte so high in 2010.  It isn’t that he was necessarily wrong, and it didn’t matter, because the point is ADP allows you to acquire the players you want at the correct spots, allowing VORP to do its thing.

Why Drafting Near the End of a Round is Valuable

As a quick aside, I wanted to point out drafting near the end of rounds (but not as the last pick in a round) can allow for a few advantages.  If you are in a 12-team league and you are allowed to pick your own draft spot, consider that spots 2, 3, 4, 9, 10, and 11 might have more intrinsic value than 1, 5, 6, 7, 8, and 12. 

In snake drafts that use a reverse draft order for subsequent rounds, drafting near the end of a round allows you to accurately predict which players might get selected between your own picks.  If you hold the 11th and 14th overall picks in a 12-team league, for example, you could easily bypass specific players in favor of others if you know the sole owner drafting between you doesn’t need or want the player you intend to draft in the later round.

For instance, assume your pick is approaching in the fifth round and you are considering a running back and a quarterback.  Only one person picks behind you before the round is over, and he already has two running backs, but no quarterback.  In such a situation, it is easy to see why you should select the signal-caller, even if the running back is higher on your board. 

Drafting in the middle of rounds doesn’t afford you this advantage.  Yes, you don’t need to wait nearly two full rounds to make selections as is the case when drafting at the edge of rounds, but it isn’t possible to predict the players to be selected between any of your picks anyway. 

Regardless, it is obvious the first and last spots in any draft are inherently less valuable than other slots, as you have to wait extended periods of time between picks, yet you cannot benefit from game theory in regards to your draft spot.

The Bottom Line

- Projections are not only useful in creating accurate rankings, they are vital.

- We can implement both consistency and scarcity into our rankings with the following formula: ‘C(P-A) + A,’ where C is correlational strength, P is projected points, and A is average points for fantasy starters at the position

- Projections are the basis from which we can create power ratings and positional tiers, but they are not the only factor.

- Risk should also determine the formation of rankings.

- VORP involves taking the “scarcest” player at his respective position, even if it seems counterintuitive.

- Understanding your opponents’ beliefs can lead to significant draft advantages.

- Average Draft Position is the easiest tool to understand public perception and subsequently enhance your implementation of game theory into VORP.

- Drafting near but not at the end of a round is intrinsically valuable because it increases your ability to predict players that will get drafted between your picks.
Recent NFL columns

Post a Comment

Please log in or register for free to have access to commenting on articles.

Jonathan Bales Jun 15, 2012 8:14:01 AM EDT
Thank you! I appreciate it and I look forward to contributing more content.
LeadPipeCinch Jun 15, 2012 2:50:17 AM EDT
Outstanding Jonathan. Welcome to Fantrax and best of luck with your book.

NFL Player News

LaQuan Williams WR - (N/A): Apr 24, 10:59AM: Williams has...

Jeremy Maclin WR - PHI: Apr 24, 10:31AM: Maclin relayed to...

Rob Gronkowski TE - NE: Apr 24, 9:56AM: Gronkowski -- who...

Cecil Shorts III WR - JAX: Apr 24, 9:29AM: Shorts continues...

Josh Hull LB - (N/A): Apr 24, 9:03AM: The Patriots have...

Victor Cruz WR - NYG: Apr 24, 8:59AM: Cruz indicates at...

Rashad Jennings RB - NYG: Apr 24, 8:38AM: The Giants plan...