Prop Bets for the Big Game

Big Game LII Image

Welcome to Super “Big Game” (what copyright?) week! The Periscope Data team put our sharpest minds together to come up with the best gambling guide on the net. We’ll walk you through a few of the prop bet predictions and how we got a little fancy with the SQL and Python code along the way. If you want to play with the data to make your own assessment click here to interact with the dashboard. (Our Head of Sales Engineering expects 20% of the winnings, but he’ll settle for a few shares.)

To kick things off let’s take a look at the nation’s favorite prop bet, the cherished Gatorade color - which is essentially like going to the casino to place money on black or red in roulette and forgetting that green is a thing,. The first interesting point to make about this analysis is that ‘Clear’ and ‘None’ are acceptable values which can represent both clear Gatorade and the oh-so-boring choice of water. For the purposes of the bet ‘Clear’ is the proper name of the wager so we combined those counts with the ‘None’ values. For anyone that hasn’t created a CASE statement to change the value of an existing set of data in SQL, this is a good example. Here is the query…

 case liquid_color when 'None' then 'Clear' else liquid_color end

-- I adjusted this to blend the 'Clear' and 'None' results ('None' is water but wins on 'Clear') 
  , count(*)
group by 1
order by 2 desc

Now that we have our dimensions in a good place we can plot the chart and place our bets!

There seems to be a CLEAR choice here, but feel free to make your own prediction.

For the data nerds out there you may have noticed that although our query did not break out the results into individual series, we were still able to color code the bars to align with the liquid colors. For that we used the same editor in Periscope and just wrote some Python on top of the output to make it look exactly how we wanted it. This is a simple use of Python but it’s still awesome to be able to use it in a single workflow with SQL. Imagine the possibilities!

# SQL output is imported as a pandas dataframe variable called "df"
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns

palette = ['#DFE1F3','#FE6B02','#DDE023','#7F67BF','#66BFD5']
plt.rcParams.update({'font.size':16, 'figure.figsize': [14,10]})
ax = sns.barplot(x='count', y='liquid_color', data=df, palette=palette, edgecolor='.3')
plt.xlabel('Number of Super Bowl Wins')
plt.ylabel('Gatorade Color')

# Use Periscope to visualize a dataframe or an image by passing data to periscope.output()

Let’s take a look at a few other popular bets! Another prop bet is the over/under on the length of the National Anthem. Pink is singing this year and her singing style is throwing a lot of people off with the over/under 2 minute wager. To help, Periscope Data mapped the anthem's length in seconds for past games:

And you can also filter based on genre by clicking the “FILTERS” tab in the upper left hand corner of the dashboard

We all agreed at the office that Pink is more of a pop singer than any other category that has sang in the big game, so let’s analyze only pop singers to see if we can identify a clear trend.

Once we click the “APPLY” button the chart will refresh to focus on that data:

And now it looks pretty clear that the safer bet is to take the under on this one with 9/14 of the other pop singers staying below the 2 minute mark. Pink is a little bit of a wild card as usual, so make sure you take the time to analyze the data on your own too.

The final bet we will focus on in this post is the big one- the over/under for the game. This data set was readily available on a ton of websites, but none of them offered up the data in raw form so we did a little scraping in Python and made our own CSV. This data goes back 51 years so the chart resembles an R-style scatterplot that you may need a magnifying glass to understand.


By the way, wouldn’t it be awesome if Periscope also offered R support? Place your bets! I’m taking the under on Q2. Just like the National Anthem chart a filter comes in handy here to identify a more focused trend in the data. In this case the “AFC” filter makes the most sense since New England has been in plenty of SBs.

That’s interesting: the total data set showed both the Total Score and the over/under trending up and to the right, and when we look at New England specific SBs the over/under line is still trending up and to the right — but the rolling average of the total score is trending down. With that trend and the Eagles defense on a hot streak we’re taking the under in SB 52!

For everyone that wants to play with the dashboard and experience all of the charts for themselves, we have to make one final note on behalf of our Philadelphia employees: Yes, there are Patriot and Tom Brady specific stats on our dash, and no that is not a bias on our part but rather good form as data gurus. We only analyzed the data sets that had enough data to identify trends in the Big Game. So, whether you are gearing up for a New England cookout party or a Cheesesteak and Pretzel feast we wish you good luck and good fortune! And don’t forget our cut of the winnings!    

Want to discuss this article? Join the Periscope Data Community!

Ryan Segar