Você está na página 1de 39

Dan Stauffer and Connor Grabowski

Ms. McKelvie
Statistical Reasoning in Sports
1 August 2015
Our Fantasy Football Fantasy
Nowadays, fantasy sports have become hugely popular. The Fantasy Sports Trade
Association (FSTA) estimates that nearly 57 million people play in some kind of fantasy sports
league. Of that, about 73% play in at least one fantasy football league. With the average annual
amount spent on fantasy leagues at $465 per player, this is clearly an enormous market. In fact,
theres more money exchanged by fantasy players than actual NFL teams! An article on
Forbes.com reports that in 2013, 32 million Americans spent about 11 billion dollars on fantasy
football leagues, while in that same year, the National Football Leagues total revenue sat just
below 10 billion dollars. Certainly, the fantasy sports market is already enormous, and is
continuing to grow. With billions of dollars changing hands annually, theres more pressure now
than ever to draft the best players. To figure out which players should be taken in the first round,
which players could be potentially good value picks later in the draft, and which players
shouldnt even be drafted entirely, well compare their performances to the league average, and
determine how much better or worse they are than an average player.
Weve decided to compare football players at three positions - quarterback, wide receiver,
and place kicker. Players at these positions are integral to every fantasy team, with QBs and
wide receivers often being boom-or-bust options and kickers ideally providing a steady point
total week-to-week. In our analysis of quarterbacks, we chose to compare the players ranked 6-
15 by the ESPN.com fantasy football gurus. We decided to look at these quarterbacks because
common fantasy football knowledge dictates that the top few quarterbacks are often drafted too
early, not providing as much value as a top running back or wide receiver; it is often smart to
wait until the fourth or fifth round to take a more average quarterback. Of course, with many
middling options seemingly providing very similar value, it would benefit us to discern which
quarterback provides the most value.
The statistics we chose to rate quarterbacks are passing yards, passing touchdowns,
interceptions, completion percentage, and completions. Passing yards and passing touchdowns
give a good indication of the quarterbacks ability to throw the ball with the other three stats
showing how the player reacts to the defense in front of him. We are measuring the quarterbacks
performance in the 2014-2015 season. The same goes for the wide receivers and kickers. We are
analyzing performance among the athletes peers in the 2014-2015 NFL season. For the kickers,
we are choosing to examine the number of field goals made, number of extra points made,
longest field goal, field goal percentage, and number of fifty yard or greater field goals made.
These five stats show how both effective and efficient the kicker is. The number of extra points
made is an indication of how skilled the players offense is. If a kicker has a very good offense
behind him, he will have more opportunities to kick and score for fantasy teams. Ideally, the
selected kicker will have a balance of team production and individual production. And finally for
wide receivers, we chose to look at receiving yards, touchdowns, receptions, yards per catch and
fumbles. These stats show us how the player performs catching the ball, holding onto the ball
once he gets it, running routes to maximize yardage. We obtained all statistics from Pro-football-
reference.com.
For kickers and wide receivers, we are picking out the top 10 ranked players by
ESPN.com. We often select wide receivers as early as the first or second round so we would like
to figure out which players are worth the early pick. With kickers, most are taken in the last
round or two. Therefore, we would like to determine which kickers are perhaps worth securing a
little earlier. The exact list of players chosen are displayed in the charts below. Also in the charts
are the statistics for each player in the five categories were using to rate them. First, well look
at quarterbacks.

Quarterback
Quarterback is widely considered the most important position on a football team, and
even sometimes in all of sports, but that doesnt necessarily mean you need one of the best
quarterbacks on your fantasy team. With most leagues allowing teams to field two or even three
running backs and another two or three receivers, it often makes sense to draft players at one of
those positions earlier and address your quarterback situation later on. By the time you get
around to drafting a signal caller, it can be the fourth or fifth round, and the top five QBs can be
off the board. So to decide which of the next ten players to draft, well have to take a look at
their statistics and compare their performances to the mean performances of the leagues
quarterbacks. Well factor in the statistics of players who played in twelve regular season games
in 2014, that is 75% of their teams scheduled contests. Here are the ten quarterbacks well be
considering drafting and their accompanying statistics from the 2014 regular season:

Number Player Completion Completions TDs Passing INTs


Percentage YDs

1 Roethlisberger 67.1 408 32 4952 9

2 Ryan 66.1 415 28 4694 14

3 Newton 58.5 262 18 3127 12

4 Romo 69.9 304 34 3705 9

5 Brady 64.1 373 33 4109 9

6 Tannehill 66.4 392 27 4045 12

7 E. Manning 63.1 379 30 4410 14

8 Rivers 66.5 379 31 4286 18

9 Stafford 60.3 363 22 4257 12

10 Kaepernick 60.5 289 19 3369 10

Completion Percentage
First, well take a look at the completion percentage of the quarterbacks. In 2014, there
were 26 quarterbacks who played in twelve or more of their team's games; thus, well evaluate
our potential draft picks against these 26 eligible players. Here is the five number summary,
mean, and standard deviation for the data set of the completion percentage of eligible
quarterbacks:

Minimum Q1 Median Q3 Maximum Mean SD

55.3 60.5 64.15 66.1 69.9 63.446 3.554

And here is a dotplot of the data set; each dot represents the completion percentage of one
of the eligible quarterbacks. Our potential draft picks are identified with arrows and numbers,
which correspond to the numbers in the first column of our first chart.
We can look at the dotplot above and, by observing the shape, outliers, center, and spread
of the graph, we can obtain a better understanding of the data set. Speaking to shape, the plot is
relatively normal, symmetric and unimodal. In terms of outliers, there is a potential outlier near
55% and two potential outliers up near 70%. The center of the the plot is very congested with the
median being 64.15% and the mean being 63.446%. Finally, with spread, the range is about 15%
with an IQR of 5.6%.
So next, we will determine the z-scores for the 10 selected quarterbacks. To do this we
use the formula (performance - mean) / standard deviation. The z-scores tell us the number of
standard deviations of one performance away from the mean of all performances. The z-scores
can then be transferred to p-values, more commonly known as percentiles. The z-scores for
quarterback completion percentage are as follows: Ben Roethlisberger (1) had a z-score of 1.028,
Matt Ryan (2) had a score of 0.747, Cam Newton (3) scored a -1.392, Tony Romo (4) achieved a
1.816, Tom Brady (5) got a 0.184, Ryan Tannehill (6) received a 0.831, Eli Manning (7) got a
-0.097, Philip Rivers (8) scored a 0.859, Matthew Stafford (9) got a -0.885 with Colin
Kaepernick scoring -0.829.
However, z-scores are not the only valuable way of ranking players. P-values tell us
what percentage of the league a player is performing better than. Assuming that a distribution is
normal, we can look in the back of our textbooks at the standard normal probabilities chart.
Looking up the z-scores for a player, we can determine what percentage of the league they are
likely performing better than. When we did so for quarterbacks, looking at completion
percentage, Ben Roethlisberger (1) came out with an 84.85% mark, meaning he performed at a
higher level than 84.85% of quarterbacks in 2014 when it came to completion percentage. Matt
Ryan (2) came in at 77.34%. Cam Newton (3) earned a paltry p-value of 8.23%. Tony Romo (4)
got 96.56%. Tom Brady (5) earned a 57.14% mark. Ryan Tannehill (6) achieved a 79.67%
mark. Eli Manning (7) came in at 49.6%. Philip Rivers (8) got a 80.51% p-value. Matt Stafford
(9) had a value of 18.67%. Finally, Colin Kaepernick (10) came in at 20.33%. With those values
calculated, observed, and recorded, we can move onto completions.

Completions
A quarterback with a low completion percentage can still be very valuable in a fantasy
setting. If that player just attempts significantly more passes than his peers, he could have a
comparable number of completions, but a lower completion percentage. So were going to take
a look at which quarterbacks on our list of potential draft picks completed the most passes, and
well analyze how impressive their performances were.

Again, well compare our ten possible draft picks to the 26 total eligible quarterbacks
from 2014. Here is the five number summary, mean, and standard deviation of the data set for
total completions by eligible QBs in the 2014 season:

Minimum Q1 Median Q3 Maximum Mean SD

197 285 342.5 379 456 329.962 65.439

And here is a histogram showing the distribution of completion totals of the eligible
quarterbacks; our potential draft picks are labeled with arrows and numbers.
Again, well look at the shape, outliers, center, and spread. The shape is relatively
symmetric. It is unimodal and approximately normal. In terms of outliers, one potential outlier is
out in the 437-477 bin. The center is once again a congested clump with the mean being about
330 yards and the median at 342.5 yards. The range of the data set is about 250 yards with the
IQR at 94 yards.
Now, we can take each quarterbacks performance, subtract the mean, and divide that
entire thing by the standard deviation to find z-scores. Ben Roethlisberger (1) had a z-score of
1.383, Matt Ryan (2) had a z-score of 1.507, Cam Newton (3) had a z-score of -1.204, Tony
Romo (4) yielded a -0.460, Tom Brady (5) gave 0.763 as a z-score, Ryan Tannehill (6) gave
1.099, Eli Manning (7) yielded 0.869, Philip Rivers (8) had a z-score of 0.869, Matthew Stafford
(9) had a score of 0.585, and finally Colin Kaepernick (10) had a score of -0.726. Lets look at
p-values.
Using the same method as we did with completion percentage, we found p-values for
each player in terms of completions. Ben Roethlisberger (1) had a 91.62% p-value. Matt Ryan
(2) came in at 93.45%. Cam Newton (3) handed in an unimpressive 11.51%. Tony Romo (4)
achieved a 32.38% mark. Tom Brady (5) came in at 77.64%. Ryan Tannehill (6) earned a mark
of 86.43%. Eli Manning (7) came out with a mark of 80.78%, as did Philip Rivers (8). Matt
Stafford (9) put up a solid 72.24%. Finally, Colin Kaepernick (10) came in at an underwhelming
23.27%. Lets move onto touchdown passes.

Touchdowns
Just as completion percentage is not a full picture of a quarterbacks ability, the number
of completions is also not a perfect snapshot of how a quarterback performs game in and game
out. Touchdowns are a very valuable stat to track in fantasy football as each touchdown thrown
by a quarterback provides a significant amount of points. So just as before, we are going to look
at how our possible future players stack up to the competition in terms of touchdowns thrown.
Like before, we will compare our 10 quarterbacks to the 26 eligible quarterbacks from
the 2014/2015 season. Presented below is the 5 number summary for touchdowns as well as the
average and standard deviation:

Minimum Q1 Median Q3 Maximum Mean SD

11 18 24.5 32 40 24.692 8.708

In the following dotplot, each dot shows one quarterbacks touchdown total; our potential
picks are pointed out with arrows and their numbers.

Once again, we can examine this dotplot by its shape, outliers, center and spread. The
shape is relatively symmetric and unimodal, but one could make a case for bimodal. The shape is
relatively normal with no major outliers other than perhaps the quarterbacks who threw for either
40 or under 15 touchdowns. The center is a little more spread out than previous graphs with the
mean being 24.692 touchdowns and the median being 24.5 touchdowns. Finally, the spread of
the plot is fairly vast. The range is 39 touchdowns with an IQR of 14 touchdowns.
Next, we take the quarterbacks touchdown performance, subtract the mean and divide
the difference by the standard deviation. Ben Roethlisberger (1) had a z-score of 0.839, Matt
Ryan (2) had a z-score of 0.380, Cam Newton (3) had a z-score of -0.768, Tony Romo (4)
achieved a 1.069, Tom Brady (5) fielded a score of 0.954, Ryan Tannehill (6) gave 0.265, Eli
Manning (7) yielded 0.610, Philip Rivers (8) had a z-score of 0.724, Matthew Stafford (9) had a
score of -0.309, and finally Colin Kaepernick (10) had a score of -0.654. Now to analyze some p-
values.
In this case, Ben Roethlisberger (1) had p-value of 79.95%. Matt Ryan (2) was at 64.8%.
Cam Newton (3) had another unimpressive showing at 22.06%. Tony Romo (4) earned a mark
of 85.77%. Tom Brady (5) came out with an 82.89% score himself. Ryan Tannehill (6) earned a
60.64% score. Eli Manning (7) got a 72.91% p-value. Philip Rivers (8) put up 76.42% as his
score. Matthew Stafford (9) came in at only 37.83%. Colin Kaepernick (10), finally, earned a
mark of 25.78%. Now, well look at total passing yards.

Passing Yards
Like touchdowns, passing yards are a key stat to monitor with quarterbacks. Many
leagues are scored such that each time a quarterback reaches a certain yardage threshold, he
receives a certain amount of points. For that reason, it is important to select a quarterback who
not only throws for touchdowns but also throws for high yardage. Thus, we will take a look at
which of our potential quarterbacks threw for the most yards.
Below is the five number summary as well as the mean and standard deviation for the
passing yards data set:

Minimum Q1 Median Q3 Maximum Mean SD

2483 3265 3758.5 4381 4952 3775.384 743.390


And here is a histogram showing the general distribution of passing yard totals of the
eligible quarterbacks; our potential draft picks are labeled.

Like before, we can examine this histogram for shape, outliers, center and spread. The
shape is for the most part normal and symmetrical. There is a slight right skew, but the set is
mostly unimodal. There seems to be no gaping outliers. The center revolves around 4000 yards
with the mean being about 3775 yards and the median being 3758.5 yards. The spread is a range
of about 2500 yards, and the IQR is 1116 yards.
And now once again, we can take the performance, subtract the mean and divide by
standard deviation to obtain z-scores for the quarterbacks. Ben Roethlisberger (1) led the pack
with a z-score of 1.582, Matt Ryan (2) earned a 1.235, Cam Newton (3) got a -0.872, Tony Romo
(4) scored a -0.095, Tom Brady (5) obtained a 0.448, Ryan Tannehill (6) got a 0.362, Eli
Manning (7) achieved a 0.853, Philip Rivers (8) got a 0.686 with Matthew Stafford (9) scoring a
0.647 and Colin Kaepernick scoring a -0.546. All in all, Roethlisberger was the leader with
Kaepernick and Newton rounding out the bottom.
In terms of p-values, here were the results (in order of their assigned number):
Roethlisberger (1), 94.29%; Ryan (2), 89.25%; Newton (3), 19.22%; Romo (4), 46.02%; Brady
(5), 67.36%; Tannehill (6), 64.06%; E. Manning (7), 80.23%; Rivers (8), 75.49%; Stafford (9),
74.22%; Kaepernick (10), 29.46%. Finally, were onto interceptions.
Interceptions
And last but certainly not least, we will be taking a look at interceptions. Interceptions are
important to limit because they usually act as a point deduction. So here we will look at what
quarterbacks throw the fewest interceptions. Below is the five number summary as well as the
mean and standard deviation:

Minimum Q1 Median Q3 Maximum Mean SD

5 9 12 15 18 12.192 3.699

And here is a dotplot showing the number of interceptions each quarterback threw; again,
our possible draft choices are labeled with arrows and numbers.

Once more, we can analyze this plot by speaking about its shape, outliers, center and
spread. It is for the most part relatively symmetric, normal and unimodal. There doesnt seem to
be any outliers other than maybe the quarterbacks who threw for 5 and 6 interceptions. The
center hovers around the 8-12 mark with the mean being 12.192 and the median being 12
interceptions. Finally, the spread boils down to the range and IQR. The range in this case is 13
interceptions with the IQR being 6 interceptions.
One final time, we are taking the performance, subtracting the mean, and dividing by the
standard deviation. But, if a quarterback throws a lot of interceptions, his z-score will be
positive. If he throws few interceptions, his z-score will be negative. So in the case of
interceptions, it is actually better to have a lower z-score, as it signifies less interceptions thrown
and a better ability to maintain possession. The z-scores ended up as such: Ben Roethlisberger
(1) had a -0.863, Matt Ryan (2) had a score of 0.489, Cam Newton (3) had a score of -0.052,
Tony Romo (4) scored -0.863, Tom Brady (5) also earned a -0.663, Ryan Tannehill (6) achieved
a -0.052, Eli Manning (7) scored a 0.489, Philip Rivers (8) earned a 1.570, with Matthew
Stafford (9) getting a -0.052 and Colin Kaepernick (10) scoring a -0.593. Again, we are looking
for low z-scores so Roethlisberger, Romo and Brady were the top scorers in this category. One
last time for QBs, well look at p-values.
Looking in the chart for p-values based on the z-scores we just found will yield untrue
results. Instead, because we actually want lower z-scores, we need to take the reciprocal of the
p-value we find in the books chart. Doing so, we get the following values: Roethlisberger (1),
80.23%; Ryan (2), 31.21%; Newton (3), 51.99%; Romo (4), 80.52%; Brady (5), 80.51%;
Tannehill (6), 51.99%; E. Manning (7), 31.21%; Rivers (8), 5.82%; Stafford (9), 51.99%;
Kaepernick (10), 72.24%. With all of that figured out, we can compare the quarterbacks on their
performance in each of the categories weve looked at.
With quarterbacks given z-scores for each of the statistics were looking at, we can now
add the z-scores for completion percentage, completions, touchdowns, and passing yards, and
subtract the z-score for interceptions. This will give us a total z-score which we can use to easily
compare our potential draft picks; the higher the z-score, the better the player projects to be this
season based on his statistics from last year. Here is a chart that lists the z-scores for each of our
potential draft picks for each of the statistics weve looked at; included in the chart is a total z-
score for each player:

Quarterback Z-Scores

Number Player Completion Completions TDs Passing INTs Total


Percentage YDs

1 Roethlisberger 1.028 1.383 0.839 1.582 -0.863 5.695


2 Ryan 0.747 1.507 0.380 1.235 0.489 3.380

3 Newton -1.392 -1.204 -0.768 -0.872 -0.052 -4.184

4 Romo 1.816 -0.460 1.069 -0.095 -0.863 3.193

5 Brady 0.184 0.763 0.954 0.448 -0.863 3.212

6 Tannehill 0.831 1.099 0.265 0.362 -0.052 2.609

7 E. Manning -0.097 0.869 0.610 0.853 0.489 1.746

8 Rivers 0.859 0.869 0.724 0.686 1.570 1.568

9 Stafford -0.885 0.585 -0.309 0.647 -0.052 0.090

10 Kaepernick -0.829 -0.726 -0.654 -0.546 -0.593 -2.162

Simply looking at the total z-scores, we can find which player it makes sense to draft.
The player with the highest number figures to be the best player. By this logic, it makes sense to
draft the Pittsburgh Steelers Ben Roethlisberger before any of the other players listed. Here is
the order in which these players rank based on their total z-scores, from best to worst:
Roethlisberger, Ryan, Brady, Romo, Tannehill, E. Manning, Rivers, Stafford, Kaepernick,
Newton. Heres a recap of these players performances based on their p-values. Keep in mind
that these percentages represent the percent of players our potential draftees performed better
than. Note that, for interceptions, weve taken the reciprocal of the p-value produced by the z-
score above because we want to know how good these players are at not throwing interceptions,
as opposed to how good they are at throwing interceptions. In the following chart, weve also
provided an average p-value in the right column to capture the average difference in the players
performance and the rest of the leagues performance.

Quarterback P-Values

Number Player Completion Completions TDs Passing INTs Average


Percentage YDs

1 Roethlisberger 84.85% 91.62% 79.95% 94.29% 80.23% 86.188%

2 Ryan 77.34% 93.45% 64.80% 89.25% 31.21% 71.21%

3 Newton 8.23% 11.51% 22.06% 19.22% 51.99% 22.62%


4 Romo 96.56% 32.28% 85.77% 46.02% 80.52% 68.23%

5 Brady 57.14% 77.64% 82.89% 67.36% 80.51% 73.108%

6 Tannehill 79.67% 86.43% 60.64% 64.06% 51.99% 68.558%

7 E. Manning 49.60% 80.78% 72.91% 80.23% 31.21% 62.946%

8 Rivers 80.51% 80.78% 76.42% 75.49% 5.82% 63.804%

9 Stafford 18.67% 72.24% 37.83% 74.22% 51.99% 50.99%

10 Kaepernick 20.33% 23.27% 25.78% 29.46% 72.24% 34.216%

Looking at the total p-values, we can see that Ben Roethlisberger is still our top
quarterback. And since p-values are dependent on z-scores, the pecking order is about the same
as with total z-scores. There is, however, a little shuffling. Tom Brady takes over the second
spot, pushing Matt Ryan to third. Ryan Tannehill takes fourth place, pushing Tony Romo to five
on our list. Philip Rivers leap frogs Eli Manning for sixth spot, pushing Eli to seventh. Matt
Stafford, Colin Kaepernick, and Cam Newton remain eight-nine-ten. In order from best to worst,
here are our quarterback ranked based on p-values: Roethlisberger, Brady, Ryan, Tannehill,
Romo, Rivers, Manning, Stafford, Kaepernick, Newton. But of course quarterbacks dont make
an entire fantasy team. Next, well take a look at which wide receivers we should draft.
Wide Receiver
Wide receivers and running backs are often the core of most fantasy football teams. In
most cases, the first two or three rounds are filled with top flight running backs and wideouts.
But running backs are more predictable--the first string running backs (the ones who get the most
carries) usually score the most points. This rule doesnt apply, though, for receivers, as there are
often three or even four of them on the field at the same time. So which wide receivers are better
than the others? If you have a pick in the first or second round, which receiver should you take
to get the most value? Well find z-scores for the top 10 receivers as rated by ESPNs fantasy
football gurus; well factor in receptions, yards, touchdown catches, yards per catch, and
fumbles. Here is a chart showing the ten receivers were analyzing and their accompanying
statistics:

Number Player Receptions YDs YPC TDs Fumbles


1 Brown 129 1698 13.2 13 1

2 Thomas 111 1619 14.6 11 0

3 Bryant 88 1320 15 16 0

4 Beckham Jr. 91 1305 14.3 12 0

5 Nelson 98 1519 15.5 13 0

6 Jones 104 1593 15.3 6 2

7 Johnson 71 1077 15.2 8 0

8 Green 69 1041 15.1 6 2

9 Jeffery 85 1133 13.3 10 0

10 Cobb 91 1287 14.1 12 1

Just as we did by limiting our quarterback sample to players who played 12 games, well
have to somehow restrict what receivers well compare our potential draft picks to. In this case,
the restriction, instead of being the number of games each player participated in, will be the
number of receptions each players had in the 2014 season. Well factor in players who caught at
least 50 passes that year and are listed as primarily wide receivers; this restriction leaves exactly
57 eligible receivers. Once we have total z-scores for each, well be able to compare our ten
potential draft picks to the rest of the league. First, well take a look at receptions.

Receptions
Receptions are a key part of determining any wide receivers fantasy value. If nothing
else, a high number of receptions indicates that a receiver is thrown to a good amount; maybe,
hes his quarterbacks favorite target. Some leagues even award receivers one point per
reception; obviously, receptions are even more important in this setting. So with all that being
said, heres a five number summary for the data set of receptions:

Minimum Q1 Median Q3 Maximum Mean SD

50 60 68 83 129 72.298 16.915


Here is a histogram showing the general distribution of receptions of the qualified wide
receivers; our possible picks are identified with arrows and their numbers.

Well again look at the spread, outliers, center, and shape of the graph to get more
information about the data set. This graph looks like right-skewed and unimodal. Looking at the
graph, there dont appear to be any outliers. However, some of the values are actually outliers.
By adding 1.5 times the IQR to the Q3, we see that accepted values fall under 117.5. Antonio
Browns mark of 129, therefore, is in fact an outlier. In terms of spread, the range of the set is
79, while the IQR is 23.
To discern which receivers are better in terms of receptions and how much better they
are, well take their performance, subtract the mean number of receptions, and divide that total
by the standard deviation. Doing this led us to find that Antonio Brown (1) was the most prolific
wideout in terms of receptions, coming in with a z-score of 3.352. Next is Demaryius Thomas
(2) with a 2.288 z-score. Continuing on, Dez Bryant (3) scored a 0.928, and Odell Beckham (4)
scored a 1.106. Next, Jordy Nelson (5) scored a 1.519 followed by Julio Jones (6) with a z-score
of 1.874. Calvin Johnson (7) scored a below-average -0.077, and A.J. Green (8) was dead last in
the player pool with a score of -0.177. Lastly, Alshon Jeffery (9) scored a 0.751 with Randall
Cobb (10) earning the same score as Beckham with a 1.106.
Now onto the p-values for receptions. Once again, these are determined by assuming our
data sets to be normal and using the chart in the textbook in conjunction with the previously
calculated z-scores. For receptions, Antonio Brown (1) found himself in the 99.96%, Thomas (2)
was in the 98.68%, Bryant (3) had a p-value of 82.38%, Beckham Jr. (4) ended up with a value
of 86.65%, Nelson (5) was in the 93.57%, Jones (6) scored in the 97.06%, Johnson (7) was in the
49.6% as A.J. Green (8) scoring in the 42.07% with Jeffery (9) scoring in the 77.34% and Cobb
(10) in the 86.86%. Next, well look at receiving yards.

Yards
For obvious reasons, the number of yards a receiver accumulates is very important in
terms of fantasy value. A large chunk of a receivers points likely comes from their receiving
yard totals. For the set of total receiving yards in the 2014 season, observing only the eligible
receivers, here is the five number summary:

Minimum Q1 Median Q3 Maximum Mean SD

465 752 921 1077 1698 951.246 295.239

Following is a histogram showing the distribution of receiving yard totals for the entire
group of eligible wide receivers; as always, our possible picks are singled out with arrows and
their numbers.

To examine this histogram we will break it down into shape, outliers, center and spread. The
shape is right skewed but also some traces of symmetry and normality. It is unimodal with a
potential outlier where receiver #1 (Antonio Brown) falls in the histogram. The center hovers
around 1000 with the mean being about 951 yards and the median being 921 yards. Finally, the
spread is extremely vast with a range of 1233 yards and an IQR 325 yards, meaning the middle
50% of receivers were within 325 yards of each other.
Once again we obtained the z-scores by the formerly mentioned process. Antonio Brown
(1) scored a whopping 2.529 with Demaryius Thomas (2) just behind with a 2.262. Dez Bryant
(3) scored a 1.239 with Odell Beckham (5) earning a 1.198. Nelson scored a 1.923 with Jones (6)
getting a 2.174. Rounding out the pack are Johnson (7) with a 0.426, Green (8) with a 0.304,
Jeffery (9) with a 0.616 and Cobb (10) with a 1.137. Just like before, the top performers were
Brown and Thomas.
In terms of p-values, here are the yardage scores with the players in order of their chart
placement: Brown (1) 99.43%, Thomas (2) 98.81%, Bryant (3) 89.44%, Beckham Jr. (4) 89.44%,
Nelson (5) 97.26%, Jones (6) 98.50%, Johnson (7) 66.64%, Green (8) 61.79%, Jeffery (9)
77.34% and Cobb (10) 87.29%. Next, well look at yards per catch.

Yards Per Catch


Yards per catch is a unique stat to monitor because it tells the statistician what the player
does with the ball once he catches it. Is he quickly tackled? Does he extend the play? We want
players that catch a lot of passes but also have a friendly yards per catch ratio. We will check out
the z-scores to discover who excels in this category. Below is the five number summary along
with the mean and standard deviation:

Minimum Q1 Median Q3 Maximum Mean SD

9 11.5 13 14.6 20.9 13.065 2.219

Up next is a histogram showing the distribution of yards per catch values for the eligible
receivers; singled out with arrows and their numbers are our potential selections.
Looking at the shape, outliers, center, and spread, well find more information. This graph is
tough to discern; it looks to be right skewed, but could also turn out to be roughly symmetric
with some bin adjusting. Looking at it with these bins, though, and seeing that there are 22
values to the left of the peak and only 15 to the right, I feel comfortable calling this graph right
skewed. Judging by the graph, there looks to be an outlier in the far right bin. Multiplying the
IQR by 1.5 and adding that to Q3, we see that accepted values must fall under 19.25. Therefore,
the max value in this set, 20.9, is an outlier. On this graph, there is a clear center peak at the bin
between 12.6 and 14.4. The mean (13.065) and median (13) both fall into this bin. In terms of
spread, the range here is 11.9, and the IQR is 3.1.
In terms of yards per catch z-scores, here is the data. Antonio Brown (1) scored a 0.061,
Demaryius Thomas (2) scored a 0.692, Dez Bryant (3) had a 0.872, Odell Beckham (4) scored a
0.577. Next, Jordy Nelson (5) was at 1.097, and Julio Jones (6) was at a 1.007. Continuing on,
Calvin Johnson (7) had a 0.962 with AJ Green scoring a 0.917. Finally, Alshon Jeffery scored a
0.106 with Randall Cobb scoring a 0.466. One would find it interesting that all 10 of our
selections were above average in this category. However, this stat is not the end all, be all of
stats. A player could still have great yardage by catching passes shorter to the line of scrimmage.
Yards per catch is very important, but once again, it does not paint the full picture of a wide
receivers performance or ability. Lets take a quick look at p-values to see how the selections
stack up against the rest of the league.
The p-values for yards per catch in order of the players chart placement are as follows:
Brown (1) 52.39%, Thomas (2) 75.49%, Bryant (3) 80.78%, Beckham Jr (4) 71.23%, Nelson (5)
86.43%, Jones (6) 84.13%, Johnson (7) 83.15%, Green (8) 82.12%, Jeffery (9) 54.38%, and
Cobb (10) 68.08%. Next, well take a look at receiving touchdowns.

Touchdowns
Scoring touchdowns--thats the goal in football. And in fantasy, scoring touchdowns is
no less important. Some receivers can be their quarterbacks favorite target inside the 20 yard
line. This can increase their fantasy value because theyre likely to score a good amount of
touchdowns. Here is the five number summary, mean, and standard deviation for receiving
touchdowns among our eligible receivers:

Minimum Q1 Median Q3 Maximum Mean SD

0 3 6 8 16 5.895 3.574

And now here is a dotplot showing the number of touchdowns of each of the eligible
receivers; once again, our potential picks are labeled.
As we have many times before, we will look at the shape, outliers, center and spread of
this dotplot. The shape is fairly symmetric, unimodal and normal with a potential right skew.
Outliers may be present where receiver #3 (Dez Bryant) fell and where a few receivers scored 0
or 1 touchdown. The center of the plot is around 6 or 7 touchdowns with the mean being 5.895
and the median being 6. Finally, the spread is quite vast with a range of 16 touchdowns and an
IQR of 5 touchdowns.
In terms of z-scores, most of our potential draft picks did pretty well. Antonio Brown (1)
had a 1.988 z-score. Demaryius Thomas (2) had a 1.428 z-score. Dez Bryant (3) came in at
2.827. Odell Beckham Jr. (4) had a 1.708 z-score. Jordy Nelson (5) came in at 1.988. Julio
Jones (6) had a z-score of 0.029. Calvin Johnson (7) was at 0.589. A.J. Green (8) had a 0.029 z-
score. Alshon Jeffery (9) had a 1.149 z-score. Finally, Randall Cobb (10) came in at 1.708. In
terms of touchdowns, Dez Bryant was the leader out of our potential picks.
The p-values for touchdown receptions for the selected players in order of chart
placement are as follows: Brown (1) 592.67%, Thomas (2) 92.36%, Bryant (3) 99.77%,
Beckham Jr (4) 95.64%, Nelson (5) 97.76%, Jones (6) 51.20%, Johnson (7) 72.24%, Green (8)
51.20%, Jeffery (9) 55.96%, and Cobb (10) 95.64%. Next, well take a look at fumbles.

Fumbles
Fumbles can be the bane of any good football players existence. A case of fumble-itis
can cost a player a roster spot and a job. In fantasy, seeing that a player fumbles more than once
or twice a year is a definite red flag. If were going to spend a first or second round pick on a top
wideout, we need to make sure they dont have fumbling problems. Here is the five number
summary, mean, and standard deviation of the data set for fumbles, observing our 57 eligible
receivers only:

Minimum Q1 Median Q3 Maximum Mean SD

0 0 1 2 7 1.228 1.376

Here is a dotplot showing the number of fumbles of each of the qualified receivers.
Weve identified our potential wide receiver picks with arrows and numbers.

Looking at the shape, outliers, center, and spread of the graph, we can plainly see that the
graph is right skewed, and it appears to be pretty obviously unimodal. There appears to be an
outlier at 7, and doing the math confirms this; accepted values must fall under 5 so the mark at 7
is an outlier. The mark at 5 just narrowly makes it into the accepted data set. In terms of center,
the mean is 1.228, and the median is 1. The large peak on the graph at 1 supports these findings.
The range here is 7, and the IQR is only 2.
Many of our potential draft picks seemed to hold onto the ball in 2014. A few, though,
had fumbling issues. Keep in mind that negative z-scores in this case are a good thing, and
positive z-scores are bad. For instance, Demaryius Thomass score of -0.892 is better than
Antonio Browns score of -0.166 which is better than Julio Joness score of 0.561. Antonio
Brown (1) had a score of -0.166. Demaryius Thomas (2), Dez Bryant (3), Odell Beckham Jr. (4),
and Jordy Nelson (5) all came in at -0.892, tied for the best score with some others as well. Julio
Jones (6) earned the bad mark of 0.561. Calvin Johnson (7) scored -0.892. A.J. Green (8) earned
a z-score of 0.561. Alshon Jeffery (9) came in at -0.892. Finally, Randall Cobb (10) came in at
-0.166.
As for p-values, the players scores are as follows: Brown (1) 56.75%, Thomas (2)
81.33%, Bryant (3) 81.33%, Beckham Jr (4) 81.33%, Nelson (5) 81.33%, Jones (6) 28.77%,
Johnson (7) 81.33%, Green (8) 28.77%, Jeffery (9) 81.33%, and Cobb (10) 56.75%. We can see
from these numbers that Julio Jones (6) and A.J. Green (8) had some issues holding onto the ball
last year; maybe it would be a good idea to look elsewhere for our picks. Now, with z-scores and
p-values found for all the players, we can total them up and ultimately rank our potential
receivers.

Adding the players z-scores for receptions, yards, yards per catch, and touchdowns,
while subtracting their z-scores for fumbles, yields a total z-score. Here is a chart with the z-
scores for each players for each individual statistic. Also included in the right column in the total
z-score of each player:
Wide Receiver Z-Scores

Number Player Receptions YDs YPC TDs Fumbles Total

1 Brown 3.352 2.529 0.061 1.988 -0.166 8.096

2 Thomas 2.288 2.262 0.692 1.428 -0.892 7.562

3 Bryant 0.928 1.249 0.872 2.827 -0.892 6.768

4 Beckham Jr. 1.106 1.198 0.557 1.708 -0.892 5.461

5 Nelson 1.519 1.923 1.097 1.988 -0.892 7.419

6 Jones 1.874 2.174 1.007 0.029 0.561 4.523

7 Johnson -0.077 0.426 0.962 0.589 -0.892 2.792

8 Green -0.195 0.304 0.917 0.029 0.561 0.494

9 Jeffery 0.751 0.616 0.106 1.149 -0.892 3.514

10 Cobb 1.106 1.137 0.466 1.708 -0.166 4.583

Just by looking at the right column, we can find which player out of our potential picks it
makes sense to select. Out of the whole lot, Antonio Brown would be our top pick. Then comes
Demaryius Thomas, who narrowly edges out Jordy Nelson. Fourth is Dez Bryant. Fifth is Odell
Beckham Jr. Sixth is Randall Cobb, who narrowly beats Julio Jones. Alshon Jeffery comes in
eighth, while Calvin Johnson comes in ninth, and A.J. Green comes in a pretty distance tenth,
leading us to question if hes really worth a top round pick; maybe it makes sense to look at a
running back or tight end early in the draft instead of A.J. Green. However, can still gain
information from looking at percentiles and p-values so well look at them next. Here is a chart
with the p-values for each player in each statistic, along with an average p-value given to the
right. Please note that once more, because we want to know how good these players are at not
fumbling the football, as opposed to how good they are at fumbling the football, weve taken the
reciprocal of the p-value produced by the z-score in the fumbles column of the chart above.

Wide Receiver P-Values

Number Player Receptions YDs YPC TDs Fumbles Average


1 Brown 99.96% 99.43% 52.39% 97.67% 56.75% 81.24%

2 Thomas 98.68% 98.81% 75.49% 92.36% 81.33% 89.33%

3 Bryant 82.38% 89.44% 80.78% 99.77% 81.33% 86.74%

4 Beckham Jr. 86.65% 88.49% 71.23% 95.64% 81.33% 84.67%

5 Nelson 93.57% 97.26% 86.43% 97.76% 81.33% 91.27%

6 Jones 97.06% 98.50% 84.13% 51.20% 28.77% 71.93%

7 Johnson 49.60% 66.64% 83.15% 72.24% 81.33% 70.59%

8 Green 42.07% 61.79% 82.12% 51.20% 28.77% 53.19%

9 Jeffery 77.34% 73.24% 54.38% 55.96% 81.33% 68.45%

10 Cobb 86.86% 87.29% 68.08% 95.64% 56.75% 78.92%

Oddly enough, when we transfer all of our z-scores to p-values, some of the results and
rankings change. Now, the top overall performer is Jordy Nelson who narrowly edges Demaryius
Thomas. Next was Dez Bryant followed by Odell Beckham. Fifth was Antonio Brown, evidently
brought down by his poor showing in yards per catch. Sixth was Randall Cobb, and seventh was
Julio Jones who just beat out Calvin Johnson. Rounding out the bottom of the pack is Alshon
Jeffery in 9th and A.J. Green coming in dead last, once again making us wonder if his 2014/2015
are indicative of a player we want to select. We now have one more position to look at and
determine who is worthy of being drafted by our teams.

Place Kicker
Ideally, a place kicker can provide a consistent week-to-week fantasy point output for
your team. There isnt a huge variation, though, between most kickers so theyll usually be
picked in the last few rounds. That one or two point fantasy point differential can mean the
difference between a win or a loss, though, so we need to figure out which kickers are better than
others. Maybe it even makes sense to take a kicker a round earlier than most because one is far
and away a better and more consistent player. Well do this by comparing the top ten kickers as
rated by ESPN.coms fantasy gurus to the rest of the league. Well analyze them using their
2014 statistics, looking at the percentage of kicks they made, field goals made, extra points
made, longest field goal, and number of field goals made over fifty yards. Here is a chart
showing the statistics of the kickers were considering drafting:

Number Player Percentage FGM XPM Long 50+ Made

1 Gostkowski 94.6 35 51 53 1

2 Vinatieri 96.8 30 50 53 7

3 Hauschka 83.8 31 41 58 2

4 Tucker 85.4 29 42 55 4

5 Parkey 88.9 32 54 54 4

6 Bryant 90.6 29 40 54 7

7 Bailey 86.2 29 56 56 5

8 Carpenter 89.5 34 31 58 6

9 Crosby 81.8 27 55 53 4

10 Suisham 90.6 29 45 53 1

Just as we did with quarterback and wide receiver, well need to limit the number of
players were comparing our potential picks to. In this case, well only look at kickers who
attempted 25 or more field goals during the 2014 regular season. Observing this parameter,
were left with 26 eligible kickers, including our potential picks. So well use the statistics of
these 26 players to find z-scores for our potential picks, and then well combine them to find a
total z-score, which we can use to discern which kickers are worth our valuable pick. First, well
look at the percentage of field goals each kicker made.

Field Goal Percentage


Logically, kicker percentage is one of the most vital stats when it comes down to
analyzing kicker performance. Kick percentage basically shows the ratio of makes and misses of
field goals. This is greatly important as we dont want kickers missing kicks and losing points for
the team. Just as before, we will look at z-scores for the kickers on their field goal percentage.
Presented below is the five number summary for the kicker field goal percentage:
Minimum Q1 Median Q3 Maximum Mean SD

74.3% 80.65% 84.2% 88.9% 96.8% 84.67% 5.797%

Below is a boxplot showing the spread and distribution of all the field goal percentages of
eligible kickers. Our potential picks are shown by dots and numbers corresponding to the
numbers given in the initial chart for kickers.

As we did with quarterbacks and wide receivers, we can analyze kicker performance on
the boxplot by looking at shape, center, outliers and spread. The shape is mostly symmetric,
unimodal and normal with a slight right skew. There doesnt seem to be any major outliers, and
the center revolves around 85. The mean is 84.67%, and the median is 84.2%. Finally, the spread
ranges over about 10% with the IQR being 8.15%.
By taking a players field goal percentage, subtracting the mean, and dividing that
quantity by the standard deviation, we can find their z-score for field goal percentage. Stephen
Gostkowski (1) had a z-score of 1.713. Adam Vinatieri (2) had a 2.092 score, the highest score
earned for field goal percentage. Steven Hauschka (3) came in at -0.150. Justin Tucker (4)
earned a score of 0.109. Cody Parkey (5) received a score of 0.730. Matt Bryant (6) got a z-
score of 1.023. Dan Bailey (7) earned a score of 0.264. Dan Carpenter (8) came in at 0.833.
Mason Crosby (9) had the lowest score of the bunch, coming in at -0.495. Shaun Suisham (10)
earned a score of 1.023. But what about those delightful p-values?
In terms of p-values for field goal percentage, our potential picks earned the following:
Gostkowski (1), 95.64%; Vinatieri (2), 98.17%; Hauschka (3), 44.04%; Tucker (4), 54.38%;
Parkey (5), 76.73%; Bryant (6), 84.61%; Bailey (7), 60.26%; Carpenter (8), 79.67%; Crosby (9),
30.85%; Suisham (10), 84.61%. But, of course, field goal percentage is far from the only thing
we can use to determine which kicker is best. Next, well look at the total number of field goals
made.

Field Goals Made


Next, the number of field goals made is very important to a kickers performance because
made field goals are the main source of points. Misses aside, kickers want to drill as many FGs
as possible to earn points for their NFL team and our fantasy squads. Below is the five number
summary along with mean and standard deviation:

Minimum Q1 Median Q3 Maximum Mean SD

20 24.25 28 29.75 35 27.269 3.976

Here is a boxplot showing the spread and distribution of the total number of field goals
made by each of the eligible kickers in 2014. Again, our possible picks are shown by dots and
the players corresponding numbers.
For field goals made, we will examine the boxplot as we have many times before. The
shape is relatively symmetric, unimodal and normal. The center is at about 27% with the mean
being 27.269% and the median being 28%. There does not seem to be any major outliers because
of how small the range is (15). Lastly, the IQR is a meager 5.5%.
Just as weve done for every statistic before this, well get z-scores that represent how
much better or worse a players performance was compared to the average. When we do so for
field goals made, we see that Stephen Gostkowski (1) had a z-score of 1.944; he led the pack in
this category. Then Adam Vinatieri (2) had a 0.687 score. Steven Hauschka (3) earned a 0.938
score. Justin Tucker (4) came in at 0.435. Cody Parkey (5) had a 1.190 z-score. Dan Bryant (6)
earned a 0.435 z-score. Dan Bailey (7) was the worst in this category, receiving a score of
-0.571. Dan Carpenter (8) received a good mark of 1.693. Mason Crosby (9) was slightly below
the league average, with a score of -0.068. Finally, Shaun Suisham (10) came in at 0.435.
Stephen Gostkowski (1) had a p-value of 97.38%. Vinatieri (2) came in at 75.49%.
Hauschka (3) earned a mark of 82.64%. Tucker (4) achieved a score of 67%. Parkey (5) came in
at 88.3%. Bryant (6) achieved a solid p-value of 67%. Bailey (7) put up an underwhelming
28.43%. Carpenter (8) put up 95.45%. Crosby (9) came in at 47.61%. Finally, Suisham (10)
earned a mark of 67%. Onto extra points made!

Extra Points Made


Extra points made, though not a huge point scorer for the average fantasy team, is
important to monitor. Firstly, it shows the team manager how often the kickers team is scoring
touchdowns and giving our kicker an opportunity to perform, and second, it shows if our kicker
is able to stay consistent when kicking very short field goals. Below is the 5 number summary
with mean and standard deviation for extra points made:

Minimum Q1 Median Q3 Maximum Mean SD

21 31 39.5 43.5 56 38 9.891

Following is a boxplot of the data for number of extra points made. Once again, our
possible picks are identified with dots and the players corresponding number.

Once again, we will examine the boxplot in the 4 categories. The shape is mostly
symmetric and normal and appears to have no serious outliers because 1.5x the IQR of about 13
is not close to any of our points. The center occurs at around 37 or 38 with the mean being 38
field goals and the median being 39.5 field goals. The spread is pretty wide with a range of 35
field goals and an IQR of 12.5 field goals.
Finding z-scores for each player, we see that Stephen Gostkowski (1) had a 1.314 score.
Adam Vinatieri (2) came in at 1.213. Steven Hauschka (3) earned a mark of 0.303. Justin
Tucker (4) got a score of 0.404. Cody Parkey (5) came in with an impressive 1.618. Matt
Bryant (6) received an underwhelming 0.202. Dan Bailey (7) led our potential picks in this
category, coming in at 1.820. Dan Carpenter (8) was below average at -0.708. Mason Crosby
(9) had a z-score of 1.517. Finally, Shaun Suisham (10) had a good score of 0.708.
For p-values, here are the results: Gostkowski (1), 90.49%; Vinatieri (2), 88.69%;
Hauschka (3), 61.79%; Tucker (4), 65.54%; Parkey (5), 94.74%; Bryant (6), 57.93%; Bailey (7),
96.56%; Carpenter (8), 23.89%; Crosby (9), 93.57%; Suisham (10), 76.11%. Next, well look at
the longest field goal of each kicker.

Long Field Goal


Though there is no direct scoring total for longest field goal, we want to look at a
kickers longest field goal to give us a general idea of the kickers range. We want to know a
kickers range because it shows us both how confident the team is in the kicker and how the
kicker would perform given a long field goal opportunity. Below is the 5 number summary with
mean and standard deviation.

Minimum Q1 Median Q3 Maximum Mean SD

49 53 53 55 58 53.654 2.208

Here is a boxplot showing the spread and distribution of the longest field goal made by
each eligible kicker during the 2014 season. Of course, we use dots and the numbers of players
to indicate where our potential picks fall.
For the fourteenth time we will examine shape, outliers, center and spread. The shape is
quite interesting with the boxplot having a slightly visible right skew. There are a few possible
outliers out near kickers #3 (Hauschka) and #8 (Carpenter) and at the minimum point of 49
yards; doing the math, 49 is definitely an outlier. The center is very interesting with the median
occurring at Q1s value of 53, showing that many kickers had a maximum kick length of 53
yards. The mean was also just about that mark at 53.654 yards. The spread is ranged over 9 yards
with the middle 50% (IQR) falling between 2 yards.
Calculating z-scores, we see that Stephen Gostkowski (1), who had been well above
average in the previous categories, is actually slightly below average here, coming in at -0.296.
Adam Vinatieri (2) had the same score of -0.296. Steven Hauschka (3) put up an impressive
mark at 1.968; he was tied with Dan Carpenter for the lead in this category. Justin Tucker (4)
had a z-score of 0.610. Cody Parkey (5) had a score of 0.157. Matt Bryant (6) also came in at
0.157. Dan Bailey (7) had a score of 1.063. Dan Carpenter (8) earned a great mark of 1.968,
getting him a share of the lead in this statistic. Mason Crosby (9) received a z-score of 0.610.
Shaun Suisham (10), finally, came in at -0.296.
Now, once again, well look at p-values. Stephen Gostkowski (1) came in at 38.21%.
Adam Vinatieri (2) earned the same mark of 38.21%. Steven Hauschka (3) earned a high mark
of 97.56%. Justin Tucker (4) achieved 72.91%. Cody Parkey (5) came in at 56.36%. Matt
Bryant (6) got a 56.36% mark as well. Dan Bailey (7) came in at a very solid 85.54%. Dan
Carpenter (8) earned 97.56%. Mason Crosby (9) came in at 72.91%. Finally, Shaun Suisham
(10) earned a p-value of 38.21%. Now, well look at the number of field goals made over fifty
yards.

Field Goals Made Over Fifty Yards


Field goals made over fifty yards are scored in many leagues as a certain point bonus on
top of the usual score for making a field goal. We want to find a kicker who consistently makes
field goals as well as bangs them in from long range to get us point bonuses. For that reason, we
are tracking the number of field goals made by our kickers over fifty yards. Below is the 5
number summary with mean and standard deviation:

Minimum Q1 Median Q3 Maximum Mean SD


0 1.25 3 4 7 3.115 1.904

And now, for the final time, we provide a boxplot. This one shows the spread of the
number of field goals over fifty yards by eligible kickers. And, of course, we show our potential
draft choices with a dot and the picks number.

And last but not least, we will analyze the data once more. The shape of the plot here is
roughly normal with a right skew. There do not appear to be any major outliers in the set. The
center occurs at around 3 with the mean being 3.115 field goals and the median being 3 field
goals. Finally, the spread is between 0 and 7 with an IQR of 2.75 field goals.
One more time, here are the z-scores for our potential draft picks: Stephen Gostkowski
(1), -1.111; Adam Vinatieri (2), -0.060; Steven Hauschka (3), -0.586; Justin Tucker (4), 0.465;
Cody Parkey (5), 0.465; Matt Bryant (6), 2.040; Dan Bailey (7), 0.990; Dan Carpenter (8), 1.515;
Mason Crosby (9), 0.465; Shaun Suisham (10), -1.111. Matt Bryant (6) clearly led the group
here, while Gostkowski (1) and Suisham (10) had the lowest marks.
And just for fun, p-values: Gostkowski (1), 13.35%; Vinatieri (2), 47.61%; Hauschka (3),
27.76%; Tucker (4), 68.08%; Parkey (5), 68.08%; Bryant (6), 97.93%; Bailey (7), 83.89%;
Carpenter (8), 93.57%; Crosby (9), 68.08%; Suisham (10), 13.35%. With more statistics
accumulated that any reasonable human being could ever want, well now compare kickers using
total z-scores and average p-values.
Adding the z-scores from field goal percentage, field goals made, extra points made,
longest field goal, and field goals made over fifty yards, we get the players total z-score, which
is given in the right column of the following chart, along with the z-scores from each particular
statistic. We can use all this information to rank players on as many factors as possible.

Place Kicker Z-Scores

Number Player Percentage FGM XPM Long 50+ Total


Made
1 Gostkowski 1.713 1.944 1.314 -0.296 -1.111 3.565

2 Vinatieri 2.092 0.687 1.213 -0.296 -0.060 3.636

3 Hauschka -0.150 0.938 0.303 1.968 -0.586 2.474

4 Tucker 0.109 0.435 0.404 0.610 0.465 2.023

5 Parkey 0.730 1.190 1.618 0.157 0.465 4.159

6 Bryant 1.023 0.435 0.202 0.157 2.040 3.636

7 Bailey 0.264 -0.571 1.820 1.063 0.990 3.566

8 Carpenter 0.833 1.693 -0.708 1.968 1.515 5.302

9 Crosby -0.495 -0.068 1.517 0.610 0.465 2.028

10 Suisham 1.023 0.435 0.708 -0.296 -1.111 0.759

Looking at the total z-scores, we can see that, out of our potential draft picks, we should
select Dan Carpenter of the Buffalo Bills. Hes actually significantly better than the next best
kicker, it might make sense to consider selecting him a round earlier than the other kickers get
picked. After Carpenter, there is a good group of seemingly very competent and consistent
kickers. Ranked from best to worst based on their total z-scores, here are our potential draft
picks at kicker: Dan Carpenter, Cody Parkey, Adam Vinatieri or Matt Bryant (tie), Dan Bailey,
Stephen Gostkowski, Steven Hauschka, Mason Crosby, Justin Tucker, and Shaun Suisham.
However, there is more than just a raw z-score to determine our pick. Some stats and categories
are more valuable than others, and perhaps we feel certain kickers were plagued by injury, tough
schedules, difficult weather or just dumb luck. But, again, we can find more information with p-
values so once more well find them. Following is a chart of p-values for each kicker in each
statistic, and once again, there is an average p-value in the right column.

Place Kicker P-Values

Number Player Percentage FGM XPM Long 50+ Average


Made
1 Gostkowski 95.64% 97.38% 90.49% 38.21% 13.35% 67.014%

2 Vinatieri 98.17% 75.49% 88.69% 38.21% 47.61% 69.634%

3 Hauschka 44.04% 82.64% 61.79% 97.56% 27.76% 62.758%

4 Tucker 54.38% 67.0% 65.54% 72.91% 68.08% 65.582%

5 Parkey 76.73% 88.3% 94.74% 56.36% 68.08% 76.842%

6 Bryant 84.61% 67.0% 57.93% 56.36% 97.93% 72.766%

7 Bailey 60.26% 28.43% 96.56% 85.54% 83.89% 70.936%

8 Carpenter 79.67% 95.45% 23.89% 97.56% 93.57% 78.028%

9 Crosby 30.85% 47.61% 93.57% 72.91% 68.08% 62.604%

10 Suisham 84.61% 67.0% 76.11% 38.21% 13.35% 55.856%

Interestingly, despite these p-values being based on z-scores, this average doesnt provide
the same rankings as the z-scores did. This time, the top two of Dan Carpenter and Cody Parkey
remain on top. Then, Matt Bryant, who had been tied for third before, is an unquestionable third.
Fourth is Dan Bailey, who was fifth. In fifth here is Adam Vinatieri, who was tied for third
before. Gostkowski and Hauschka remain sixth and seventh. Justin Tucker moves up one spot
to eighth, pushing Mason Crosby down one to ninth. Remaining at the bottom is Shaun
Suisham. Still though, if hes on the board, Dan Carpenter is the kicker to select.

Conclusion and Error Analysis


Just to review, the aim of this project was to help two budding fantasy football managers
discern who to include on their teams. We chose to analyze players at three common fantasy
football positions: quarterback, wide receiver and place kicker. We chose the 10 players in each
position based on our fantasy picking strategies. Usually, we take wide receivers early and
quarterbacks a little later. As a result we are looking at the top 10 wide receivers and
quarterbacks ranked 6-15 by espn.com Fantasy Football services. With kickers, most of them are
chosen in the last two rounds so we chose to look at ESPNs Top 10 kickers to see who is worth
an earlier selection. Five different statistics valuable to fantasy scoring were considered at each
position.
With each statistic, we tried to analyze the players performance in the 2014/2015 season
to make a generalization about the players ability to impact our fantasy team this year. To do
this, we found league averages and standard deviation for each of the 5 statistics to then calculate
z-scores for all of the players performances. The z-scores were then converted to p-values using
the standard normal probability table. These p-values can be used to compare players to each
other. So now, 40 pages later, we can provide these final rankings based on the statistics above
and a little bit of our own hunches.
For quarterbacks, we are mostly basing our final rankings off of total z-scores. The
quarterback ranked tenth is Cam Newton, and Colin Kaepernick is ranked ninth. Eighth is
Matthew Stafford. Next, Philip Rivers and Eli Manning are very close in total z-score so we
employed our own gut decision to place Rivers 7th and Manning 6th due to Mannings stronger
receiving core. Fifth in our quarterback ranking is Ryan Tannehill, and Tony Romo is 4th. In
third place is Tom Brady with Matt Ryan finishing second. Finally, the leading quarterback in our
studies and very deserving of a selection onto our team is Ben Roethlisberger. However, all ten
quarterbacks, though differing in stats, have qualities that could pay off for any fantasy team.

Final Quarterback Rankings


Rank Player Total Z-Score Average P-Value

1 Roethlisberger 5.695 86.188%

2 Ryan 3.380 71.21%

3 Brady 3.212 73.108%

4 Romo 3.193 68.23%

5 Tannehill 2.609 68.558%

6 E. Manning 1.746 62.946%

7 Rivers 1.568 63.804%

8 Stafford 0.090 50.990%

9 Kaepernick -2.162 34.216%

10 Newton -4.184 22.620%


Next, we take a look at wide receivers. Recall that our potential picks are the top 10 at
their position, as ranked by ESPN.coms fantasy gurus. Thus, we want a true top tier player
when we select a wide receiver, likely in the first or second round. Using a similar method as
with quarterbacks, we chose our top 10 receivers. We ranked Jordy Nelson above Demaryius
Thomas because Thomas plays with an aging quarterback in Denver, who may have some sort of
drop-off this upcoming year. We put Julio Jones ahead of Randall Cobb because Jones is his
teams best wide receiver, while Cobb plays second fiddle to Jordy Nelson. Nevertheless, here is
how we ranked our wide receivers:

Final Wide Receiver Rankings

Rank Player Total Z-Score Average P-Value

1 Brown 8.096 81.24%

2 Nelson 7.419 91.27%

3 Thomas 7.562 89.33%

4 Bryant 6.768 86.74%

5 Beckham Jr. 5.461 84.67%

6 Jones 4.523 71.93%

7 Cobb 4.583 78.92%

8 Jeffery 3.514 68.45%

9 Johnson 2.792 70.59%

10 Green 0.494 53.19%


And, finally, we can rank place kickers. Since kickers statistics arent a result of the
scheme they play in, and are only slightly influenced by the players around them, we decided to
rank our kickers entirely based on their total z-scores. Here are the kicker rankings:

Final Place Kicker Rankings

Rank Player Total Z-Score Average P-Value

1 Carpenter 5.302 78.028%

2 Parkey 4.159 76.842%

3 Bryant 3.636 72.766%

4 Vinatieri 3.636 69.634%

5 Bailey 3.566 70.936%

6 Gostkowski 3.565 67.014%

7 Hauschka 2.474 62.758%

8 Crosby 2.028 62.604%

9 Tucker 2.023 65.582%

10 Suisham 0.759 55.856%

While this exercise was very factual, informative and beneficial to our future fantasy
teams, there are a few variables we can not neglect or forget about. The first is injury. Players
who were ranked high by ESPN but low by our tests were players who were sidelined and ailed
by lingering injuries last season. Some of these players include: Odell Beckham Jr, A.J. Green,
Calvin Johnson and Cam Newton. Had these players been on the field for the full season, they
might have seen higher or lower rankings. Also, much of a players success boils down to
teammates, and obviously, teams are always changing. Perhaps team have added or subtracted
players who would have been beneficial or detrimental to a players performance in the
upcoming or previous seasons. Due to the high level of competition in the NFL, it is hard to
suggest that a players performance is a fluke and could not reproduce his seasons marks.
However, random chance is always a possibility, and perhaps we caught Antonio Brown during a
good season and Matthew Stafford during a bad season. Also, another possible limitation of one
of the statistics was how fumbles are such an infrequent instance. Thus, players who have
increased or decreased abilities to hold onto the ball may have been affected by random chance
in the fumbles trial. Finally, certain players might have been influenced by harder or easier
schedules this past season and might have performed above or below their true abilities due to
who they were playing that given week.
All in all, the limitations of this project are minor but still need recognition. Either way,
due to this assignment, we both have increased knowledges of fantasy football and statistics in
general. We are officially ready to take our leagues by storm.

Você também pode gostar