Saturday, February 02, 2008

Filling in your Super Bowl XLII Challenge Sheet

My sister, one of the commissionioners of RKB Fantasy Football, has a fun Challenge sheet that her fantasy football league plays for every Super Bowl. You get points for each correct question, some of which are random and some of which might require knowledge of the teams. She claims it is all random, but I went to NFL.com Superbowl history to compile the statistics I could to try to improve my odds. Here is the list of questions with some analysis:

These first three questions are fifty/fifty random questions. I suggest just picking them:

Coin Toss Heads/Tails (5 pts)
Coin Toss Winner (5 pts)
First Play (not kickoff as seen on TV) Left to Right or Right to Left (5 pts)
First team to take a timeout
(5 pts) - Not recorded in the stats, probably 50/50
First team to score
(5 pts) - This one depends on how good the teams are. Over the 41 Superbowls, NFC leads AFC 21 to 20. First to score wins the game 27 out of 41 times. This year I think the Patriots are a safe bet to win, since the oddsmakers have given New York 12 pts.

First play to score Def/Pass TD/Rush TD/FG (10 pts) - This is recorded in the stats, and is displayed in the chart below. Overwhelmingly field goals (FG) are the first to score. But that still only occurs 50% of the time over the history of the Superbowl, still that's better than a 1 out of 4 guess.

First FG distance 0-19/20-29/30-39/40+/noFG (5 pts) - The first field goal distance is spread more evenly over the options. Eliminating no field goals and less than 19 yds improves the chances, but it is a tossup among the last three options.


These next few questions are data not compiled in the stats, and seem to be answers that could be fifty/fifty either way. A glance further down at some of the stats from fumbles, interceptions and penalties suggest that choosing neither is a low probablity response and will cost you the points, choose a team here.
First team to Fumble (Patriots, Giants or neither) (5 pts)
First team to throw an interception (Patriots, Giants or neither) (5 pts)
First team with a penalty (Patriots, Giants or neither) (5 pts)
First Penalty is on (Offense, Defense or neither) (5 pts)
First Team to Challenge a Play (Patriots, Giants or neither) (5 pts)

Point Scoring seems to be a function of how good the teams are and requires some knowlegde of football, which usually means I will not get it right. It seems that ties happen as often as one team having more points. These are tricky question and hard to get from the data. 11 out of 41 Superbowls the 1st quarter was tied, the next three quarters were more likely to be won by one team or the other.

Team to score the most points by quarter
1st Quarter (Patriots, Giants, Tie) (5 pts)
2nd Quarter (Patriots, Giants, Tie) (5 pts)
3rd Quarter (Patriots, Giants, Tie) (5 pts)
4th Quarter (Patriots, Giants, Tie) (5 pts)

The following questions all have collected data associated with them.

Total fumbles (5 pts) - Almost half the time it is 2 or 3, as demonstrated in the chart below.

Total interceptions (5 pts) - The same is true with interceptions, in half the 41 Superbowl games there were 2 or 3 interceptions.

Total TD (5 pts) - It seems from the chart below that we should pick either 4 or 5 or 6 & up. We do have another bit of information and that is that touchdowns are correlated with the total score of the game.

If we examine a chart of the score of the game as a function of touchdonws over the 41 Superbowls and use the over/under prediciton of about 54 points for the total number of points for the game, it seems that 6 touchdowns and up is the best pick for this category.

Total FG (5 pts) - Field goals are less correlated with the score of the game than touchdowns, they represent the error in the chart above. It also seems that over the history of the Superbowl that 2 or 3 field goals occur half of the time.

Total Sacks (5 pts) - More than 6 sacks is the most likely outcome according to the chart below, but it still only happens less than half of the time. Still the comparison we are making is against a random choice, so any information helps.

Total penalties (5 pts) - More than 11 penalties occurs in almost half of the Superbowl games played. I wonder if this statistic would benefit from more data from other games in the season.

Total punts (5 pts) - As demonstrated by the chart below, half the time in the 41 Superbowl games played there are 8,9,10 or 11 punts.

Total points (at the end of the game) 0-35, 36-45, 46-55, 56+ (10 pts) - For this one I suggest using the over/under to utilize the knowledge of the bookies and the betters. I have seen 54 1/2, but I have also seen AccuScore predict a score of 52 from 10,000 simulations of the game. I think 46-55 might be an OK bet here.

Winning Team (5 pts) This one is self explanatory. Earlier I mentioned the Patriots will probably win based on the spread. Pick them this year, everybody else is.

MVP position QB/K/WRTE/RB/DEF (10 pts) - As shown in the chart below, Quarterbacks are overwhelmingly most likly to be the MVP, at least half of the time. With Tom Brady playing, this seems to be a safe bet.

Tie Breaker
The Total Amount of Points You Predict You Will Score
(Out of a possible 145 points) - A glance up through the questions above shows that 70 points are a 50/50 random shot so we should take half of those (35 points) and add them to our sure shots (75 points) as determined by our statistical compilation and analysis. I think my winning score this year will be 110 points.

Was this data helpful or interesting to you? Do you agree with my methods and conclusions? Please leave a comment below. I have the data compiled in a spreadsheet. If you would like it leave a message in the comments.

Enjoy your Super Bowl

2 comments:

Richard Koehler said...

I wonder if the commenters on this blog realize that I can delete obnoxious comments with bad language in them? I love comments that further the conversation. Bad language and general obnoxiousness will be removed.

For this post why don't we talk about football or statistics instead. That was a crazy Superbowl wasn't it? I might characterize it as having a low probability.

Richard Koehler said...

I am still going to delete comments with only bad language. How about offering an opinion.