7/28/09
This article investigates the cost of a mistake in a fantasy football
draft. It is a continuation of my article from last week, Evaluating
PreDraft Trades. We start by clarifying the baseline assumptions
and our approach to the problem. We then provide illustrative results
for a baseline case and some relevant sensitivities. Finally, we
take a step back and provide perspective that we hope you find useful
in your draft preparations.
Background
Last week we built a Monte Carlo simulation model that allows us
to compute the expected value of draft picks at each position in
a 12team PPR league under a certain framework of assumptions. We
used this last week to investigate the impact of uncertainty on
the relative value of draft positions under one set of assumptions.
We used the idea of Optimal drafting results, which involves drafting
with perfect knowledge of the future. And we leveraged the ideas
of Optimistic, Realistic and Pessimistic drafting results, layering
in increasing likelihood of underperformance of draft picks.
Key conclusions were:
 Existing tools that value draft positions are a good start
 While a good start, they tend to assume perfect knowledge of the
future
 We incorporated the uncertainty of the future and saw that it
can impact the value of picks
This week we will use the same
model to investigate the value of a mistake in the draft. Throughout
the article in each simulation all picks except one (i.e. 47 out
of 48) will choose optimally. Then, for a single pick in the draft
we will layer in increasing likelihood of underperformance.
We make a slight change this week to the language we use for the
drafting assumptions. What we called “Realistic” last
week we will now refer to “Moderately Pessimistic”.
And what we called “Pessimistic” last week we will
call “Very Pessimistic” this week. This is meant to
clarify that we do not view the moderately pessimistic assumptions
as being calibrated to reality – but rather just being somewhere
between very pessimistic and optimistic.
We define our tests with the goal of leading us to a useful conclusion
that you might be able to use in your upcoming fantasy football
draft.
Baseline Assumptions We use the
same baseline assumptions as last week for the most part:
 12team PPR league, with 15 rounds
 Focus on rounds 1, 2, 3, and 4
 Assume the following distribution of picks:
Draft Pick 
RB 
QB 
WR 
TE 
1 
100.0% 
0.0% 
0.0% 
0.0% 
2 
100.0% 
0.0% 
0.0% 
0.0% 
3 
100.0% 
0.0% 
0.0% 
0.0% 
4 
100.0% 
0.0% 
0.0% 
0.0% 
5 
100.0% 
0.0% 
0.0% 
0.0% 
612 
60.0% 
20.5% 
20.0% 
0.0% 
Round 2 
30.0% 
20.0% 
50.0% 
0.0% 
Round 3 
30.0% 
20.0% 
40.0% 
10.0% 
Round 4 
25.0% 
20.0% 
45.0% 
10.0% 

In other words we assume the first 5 picks are always RB and
no TE is picked before round 3. We assume round 2 will on average
have 4 RB’s, 2 QB’s, and 6 WR’s picked but that
will vary by scenario of the simulation, etc.
Our conclusions are not unique to 12team PPR leagues, and would
not change a whole lot for many changes to these baseline assumptions.
We stick with these as a starting point for simplicity.
We define the value of a player based on their total points scored
in excess of the last drafted player for that position, assuming
48 RB’s, 24 QB’s, 60 WR’s, and 12 TE’s
will be drafted. In other words, we focus on their value in excess
of a waiver wire pick. Then we run 1000 simulations of drafts,
letting the choice of each position vary for each draft –
and we look at the average value of each pick over all the simulations.
Baseline Results
Using our usual methodology, we normalize the value of the #1
pick under the baseline assumptions to equal 100. Then we look
at the sum of the value of the top 4 draft picks under the baseline
assumptions, where all teams draft “optimally”. Recall
that under this set of assumptions Tom Brady would go undrafted
in 2008. The chart below shows the baseline results for the average
of the sum of the value of the first four rounds of draft picks
by team:
Team 
Baseline 
1 
237 
2 
227 
3 
227 
4 
220 
5 
216 
6 
220 
7 
220 
8 
221 
9 
221 
10 
221 
11 
220 
12 
219 

Two points stand out here. First the value of the first four
rounds of picks for teams 412 is pretty similar under these baseline
assumptions. Second the value of the top 3 picks has some separation
from the value of the picks of managers 412. This is a direct
result from our assumption that the top 3 picks of the first round
are more reliable than the rest of the first round. This happens
to be consistent with the view of a number of experts who view
MJD, AP, and Forte as being a step above the next set of RB’s.
To the extent your personal view differs from this, this should
be kept in mind.
Approach
We start by focusing on the 6th team in the 12team draft. We
let everyone else in the draft continue to always draft “Optimally”.
We then model the concept of a Team 6 “mistake” in
the fantasy draft by letting team 6’s draft result drift
from “Optimal” down to “Optimistic” down
to “Moderately Pessimistic” down to “Very Pessimistic”
all the way down to “100% Bust”. In reality managerial
draft success will always be between optimal and 100% bust –
and usually be between our Optimistic and Very Pessimistic cases.
In this framework we will try to get some perspective on the cost
of a mistake in the fantasy draft.
Case Study of Team 6 – Let Assumptions
Drop from Optimal to Optimistic
We let exactly one pick of Team 6’s drop from “Optimal”
results to just “Optimistic”. We rerun our simulations
under these assumptions, assuming that suboptimal pick is the
round 1, 2, 3, and 4 pick separately.

Baseline  All teams draft Optimal 
All teams draft Optimal
in all rounds, except for Team 6 drafts results are
only "Optimistic" in rounds... 
Team 
Baseline 
1 
2 
3 
4 
1 
237 
237 
237 
237 
237 
2 
227 
227 
227 
227 
227 
3 
227 
227 
227 
227 
227 
4 
220 
220 
220 
220 
220 
5 
216 
216 
216 
216 
216 
6 
220 
201 
204 
206 
209 
7 
220 
220 
220 
220 
220 
8 
221 
221 
221 
221 
221 
9 
221 
221 
221 
221 
221 
10 
221 
221 
221 
221 
221 
11 
220 
220 
220 
220 
220 
12 
219 
219 
219 
219 
219 

The expected value of team 6’s top 4 draft picks drops
by about 10% if the first round pick drops from being optimal
to only optimistic. The drop grades up to 5% in the later rounds.
Note the drop is most severe if we introduce a nonoptimal draft
decision in round 1 and least severe if we introduce it in round
4.
Case Study of Team 6 – Expand for
Higher Likelihood of Failure

Baseline  All teams draft Optimal 
All teams draft Optimal
in all rounds, except for Team 6 drafts results are
only "Moderately Pessimistic" in rounds... 
Team 
Baseline 
1 
2 
3 
4 
1 
237 
237 
237 
237 
237 
2 
227 
227 
227 
227 
227 
3 
227 
227 
227 
227 
227 
4 
220 
220 
220 
220 
220 
5 
216 
216 
216 
216 
216 
6 
220 
187 
195 
199 
206 
7 
220 
220 
220 
220 
220 
8 
221 
221 
221 
221 
221 
9 
221 
221 
221 
221 
221 
10 
221 
221 
221 
221 
221 
11 
220 
220 
220 
220 
220 
12 
219 
219 
219 
219 
219 

Not surprisingly, if we move from optimal to moderately pessimistic
draft expectations the drops are more severe. The drop is most
severe if we experience it in round 1, and least severe in round
4.

Baseline  All teams draft Optimal 
All teams draft Optimal
in all rounds, except for Team 6 drafts results are
only "Very Pessimistic" in rounds... 
Team 
Baseline 
1 
2 
3 
4 
1 
237 
237 
237 
237 
237 
2 
227 
227 
227 
227 
227 
3 
227 
227 
227 
227 
227 
4 
220 
220 
220 
220 
220 
5 
216 
216 
216 
216 
216 
6 
220 
183 
190 
194 
200 
7 
220 
220 
220 
220 
220 
8 
221 
221 
221 
221 
221 
9 
221 
221 
221 
221 
221 
10 
221 
221 
221 
221 
221 
11 
220 
220 
220 
220 
220 
12 
219 
219 
219 
219 
219 

Obviously, if we move from optimal to very pessimistic draft
expectations the drops are even more severe. The drop is still
most severe if we experience it in round 1, and least severe in
round 4.

Baseline  All teams draft Optimal 
All teams draft Optimal
in all rounds, except for Team 6 drafts results are
only "100% Bust" in rounds... 
Team 
Baseline 
1 
2 
3 
4 
1 
237 
237 
237 
237 
237 
2 
227 
227 
227 
227 
227 
3 
227 
227 
227 
227 
227 
4 
220 
220 
220 
220 
220 
5 
216 
216 
216 
216 
216 
6 
220 
138 
159 
176 
185 
7 
220 
220 
220 
220 
220 
8 
221 
221 
221 
221 
221 
9 
221 
221 
221 
221 
221 
10 
221 
221 
221 
221 
221 
11 
220 
220 
220 
220 
220 
12 
219 
219 
219 
219 
219 

If we move from optimal to 100% bust draft expectations (i.e.
the most extreme theoretical case) the drops are obviously even
more severe. The drop is still most severe if we experience it
in round 1 and least severe in round 4.
OK, that’s nice. But What Can We
Actually Do With This?
It’s obvious if one of your picks is more likely to do badly
that it will reduce the expected value of your picks (all else
held equal). The analysis above helps put some numbers to this,
under one reasonable set of baseline assumptions, to give you
some perspective on just how much it “costs” to make
a pick that is more likely to be a mistake.
So what can we learn from this?
 If you are going to go with a
risky pick, it is riskier to do it in round 1 than the later
rounds.
This lesson shines through with every chart above. Namely –
the impact of a drop in the likelihood of draft success is always
more severe in the earliest rounds than the later rounds. The
potential for extreme upside above draft expectations is also
higher in the later rounds (simply given the fact that the expectations
are lower). These points support a draft strategy of minimizing
risk in the first round.
Here’s an example: Let’s say with your round 1 pick
you are thinking of taking someone the experts view as approaching
the end of his career with high injury risk (e.g. Brian Westbrook
or Steven Jackson). If you are weighing one of them against a
younger option that experts view as being safer, but ranked slightly
lower – you might want to consider taking the “worse”
(i.e. lower expectation in the eyes of the experts) but safer
option. For small expected differences, it might make sense to
just pass on that overworked injuryprone back and use your top
pick to choose an option that seems less likely to bust at this
point.
If you are moving in the direction of taking a “risky”
pick in round 1, you should seriously consider what the team is
going to look like if your pick gets injured at the start of the
season and doesn’t play again. What has happened in the
past when your pick has missed a game – has the offense
stepped up and filled the hole with existing resources? Or has
that unstoppable Oakland run defense subsequently shut them down?
I would feel more comfortable taking a risky RB if I had confidence
in the team’s run game with the backup (e.g. San Diego if
L.T. goes down) than I would if I lacked confidence in the offense
(e.g. St. Louis if Steven Jackson goes down).
 Taking on unnecessary risk in
round 1 picks can severely decrease the value of your picks.
This is in some ways a continuation of the above point. As the
likelihood of a single pick to bust or underperform increases,
the value of your first 4 round picks can decrease drastically.
In any given season you might get lucky and not observe this in
your results, but in the long run there is a real cost to taking
on unnecessary risk. The analysis above attempts to quantify this
cost under one set of relevant assumptions.
Round 1 is not the place to take on a lot of risk. It’s
just very difficult for players to generate enough upside above
their expected performance to compensate you for this higher
risk. As your draft proceeds, picks that have a higher than
normal potential to bust or severely underperform can be worth
the risk, because they have a chance to overperform (e.g. Plaxico
Burress this year might be one example of a highrisk player
that has potential to overperform, depending on how the next
month pans out). But common sense combined with the results
above all point in the direction of minimizing risk (as much
as possible) in your round 1 pick. Save that gamble for later.
 Get your fantasy advice from a
range of sources.
FFToday.com is a great resource – use it as one tool in
your basket. Weigh the opinions of a lot of resources, and make
your own adjustments based on your own views. Getting more information
will give you more potential to see warning signs of a potential
bust or severely underperforming pick. That in turn increases
the expected value of the picks you have in your hand. The analysis
above provides additional perspective on how valuable that can
be.
Next Steps
Thanks for the constructive feedback last week. I’ve been
able to respond to everyone who wrote throughout the week, and I
look forward to hearing
your thoughts again this week.
Next week we’ll continue to apply math to draft decisions,
but we’ll take a different approach. We will investigate the
value of consistency at QB with a case study of a specific QB. Bonus
points if you can guess who it is in advance. See you next week
–
