Challenge #409: Catch or Drop? Predicting Business Outcomes with Randomness

[AYXAcademy]

Hi Maveryx,

A solution to last week’s challenge can be found here. 

The Super Bowl is just around the corner, and this year’s championship will be a blast. When the football season is over, there are always stats to calculate! To get into the spirit of post-season football, we analyzed the NFL teams and the players’ catch-and-drop abilities following the same model as the video game Madden NFL 23. The game uses a dice-roll system and player matchup data to determine whether a throw will be caught or dropped by the receiver.

Your task is to create an Alteryx workflow that models randomness and chance to predict a binary business outcome. You will simulate a scenario where randomness plays a crucial role in decision-making, similar to how it affects outcomes in sports or real-life business scenarios.

The dataset provided contains key attributes for analysis, including team and player stats (Madden 22). Your work will be focused on the All Players sheet.

Challenge Tasks:

• Create a dice rolling mechanism with three six-sided dice (3d6) and find the modifiers that will influence the outcome based on the dice roll. These modifiers should reflect the impact of random chance on your chosen business scenario. (This is an advanced task. If you are a beginner Designer user, you can use the start workflow Challenge_409_start_file_beginner.)
• Find the highest-rated team using their players’ stats.
• Pick two teams to go against each other and identify the best players for each position for both teams.
  • Team 1 will be offense: quarterback (QB) and wide receiver (WR) positions.
  • Team 2 will be defense: cornerback (CB) position.
• Apply the modifier to the matchup between Team 1 and Team 2 to determine if there is a “catch” or a “drop.”

This challenge can be solved in two levels. For those eager to enhance their data creation abilities, you have the option to craft the dice and dice rolling mechanism from the ground up. Alternatively, a ready-made dice rolling mechanism is available for use if you are a beginner Designer user. Should you choose to build your own tool, it must include three six-sided dice (3d6), along with all possible combinations of outcomes they can produce.

Learn more about 3d6 dice notation here: https://en.wikipedia.org/wiki/Dice_notation.

Get inspiration on how to solve this Football Fanalytics challenge by reading the NFL Analytics with Alteryx and Madden Blog. Read how to leverage Madden NFL football as inspiration to create better model training data.

Source: https://www.reddit.com/r/Madden/comments/ouzjvd/madden_22_player_ratings_spreadsheet_fully/?rdt=5217...

Good luck!

[ed_hayter]

Given Washington's NFL performance this year and my deep-rooted hatred of the Cowboys I wanted to see a result in Alteryx I didn't get to see in real life a Washington win.

The flow could do with more tuning on the modifiers and the attributes used in each scenario (I was quite generous towards Terry over Diggs) but the cruz of it is that we randomly select a throw distance and then use that distance to pick the RR and accuracy ratings for WR and QB respectively - if the throw is inaccurate it will be incomplete. If the throw is accurate we then check the coverage the CB is playing compared to route running to determine how open the receiver is. His openness determines the attributes used to determine if a catch is made or not.

Spoiler

This sample run had Mclaurin catching a short throw from Fitzpatrick against Diggs playing zone coverage where Mclaurin is wide open.

[balajilolla2]

Spoiler

[Dev73]

My solution 

Spoiler

[RWvanLeeuwen]

Not a sports fan! But tried to solve this anyways :-)

I am guessing that a Drop means that it was a touchdown and a Catch means the Opponent has intercepted the football?

Here's my take:

0:

Spoiler

(1 run of this workflow is 1 match) -- I figured that I needed a random dice to filter the WR multiplier. In parallel I found the best Team, and a random opponent. Then I applied the WR to the Overall scores and compared the scores per match.

1:

Spoiler

Then I converted the prior workflow to a batch macro using a control parameter. output the results to a YXDB containing ~11k simulated results

2:

Spoiler

Then I built a scaffold of all possible matches (the best as offense, versus the best as defense) between the best team vs all others. I created a scaffold per "MatchID" just to keep track of the rows and how think the confrontation would end. If the overall avg is higher for team 1 then they win and it is a "drop" in my book. The output from the previous step (random drop rate) was then combined with the expected drop rate.

3:

Spoiler

A correlation helped to finalise the expected drop (0 or 1) with the simulated results. This resulted in a perfect correlation. (I could have dichotomised the random drop rates, but I felt a visualisation would be more appropriate)

[Kenda]

Spoiler

Instructions were a bit confusing, but here is my take

[mmontgomery]

Took it a bit far but more details in spoiler

Spoiler

Goal was to compare a sim of 30 coin flips with 30 results based on my logic resulting in a confusion matrix

[AkimasaKajitani]

My solution.

Spoiler

To be the honest, it is too dificult to understand the rule. Originally, the comparing should be done with the appropriate offensive and defensive parameters for each position. But at this time, I use the Overall field.

 

[olga_strubbe]

This was a fun and timely challenge.  I liked the 'Random % Sample' tool as I don't get to use it nearly as much.  I tried to incorporate the random % sample as much as I could for picking the teams and QB, WR and CB players along with the catch or drop outcome.  It was quite entertaining!

[RolandSchubert]

Spoiler

Spoiler

Spoiler