A Python GUI Application predicting the winner of ODI Cricket matches at 80% accuracy taking into account every player’s affinity to ground, recent form and performances against opposition.

Objective -

The very interest towards the cricket matches and the anticipation towards knowing the result beforehand to avoid edge-seating situations triggered us to come up with a Win Predictor model for ODI matches.

Tools used -

• BeautifulSoup – Used for souping webpage (ESPNCricinfo in our case)
• TKinter – Used for front-end GUI

Website Resource to scrap data: ESPN-Cricinfo - Statsguru

Prediction Factors -

• Recent Form – The Statistics of the latest 10 matches of players is used to depict his current form
• Ground Form – The Statistics of the players on that particular ground(venue of the match), indicates how effectively will he be able adapt himself to the specific ground.
• Opposition score – The Statistics of batsmen pitted against the bowling unit of the opposition gives an insight of the ability of players to score runs or take wickets against the opposition

Batsman Score(Recent Form & Ground Score)

SR = (Runs/Balls Faced)*100
runs = (Runs*100)/(Total team score/10)
wkts = (1-((Number of Dismissals)/(Number of matches)))*100
score50 = (Number of 50s/Number of matches)*50
score100 = (Number of 100s/Number of matches)*100
score150 = (Number of 150s/Number of matches)*150
score200 = (Number of 200s/Number of matches)*200

SCORE = SR + runs + wkts + score50 + score100+ score150 + score200

Bowler Score(Recent Form & Ground Score)

Hauls = ((37.5*Number of 3Ws) + (62.5*Number of 5Ws)) / Number of matches 
wkts = (Wickets taken / Number of matches)*100
maidens = Maidens / int(Balls/6)*100*5
overs = int(Balls/6) + float((int(Balls)%6))/10
economy = (1-((Runs given)/(10*overs)))*100

SCORE = hauls + wkts + maidens + economy

Opposition Score

SR = (Runs/BallsFaced)*100
if(Wickets != 0):
    runs_per_wkt = Runs/Wickets
 else:
    runs_per_wkt = Runs
additional = Number of 4s + 1.5*(Number of 6s)
wkts = Wickets*(-5)

SCORE = SR + runs_per_wkt + additional + wkts

The Application is designed to work on Linux Based Distributions and requires python 2 installed

Packages Required -

All the packages mentioned below are meant to be installed apart from the basic pythonic import files

1. TKinter
2. urllib
3. pandas
4. bs4

To run the Application -

Upon proper installation of above specified packages, run the python(2.x) file in terminal as:

python PredictTheOutComeUI.py

Working

Upon successful start of applications, to predict a match's outcome

1. Give the date of the match, playing teams and the Venue Code (indicated by ESPNCricinfo) and Click 'Next'

2. Enter the 11 players you are expecting to play along with thier roles in the team. (Both the batting and bowling statistics is used for the All-Rounder)

3. Click 'Scrap' to get details from the website (This may take some time and requires active internet connection)

Future Work

• Designing Learning Algorithms(Classifier and Regressor models) on top of scrapped data to come-up with robust, effective and better results
• Usage of python's 'multiprocessing' capabilities along with 'asyncio' yields to quicken the process of scrapping.
• Replacing the venue code in GUI with country and ground names.
• Better GUI Design