app.py

import streamlit as st
import numpy as np
from numpy import where as np_where
import pandas as pd
import gspread
import plotly.express as px
import scipy.stats as stats
from pymongo import MongoClient
st.set_page_config(layout="wide")

@st.cache_resource
def init_conn():

        uri = st.secrets['mongo_uri']
        client = MongoClient(uri, retryWrites=True, serverSelectionTimeoutMS=100000)
        dfs_db = client["NCAAF_Database"]
        props_db = client["Props_DB"]

        return props_db, dfs_db
    
props_db, dfs_db = init_conn()

game_format = {'Win%': '{:.2%}', 'Vegas': '{:.2%}', 'Win% Diff': '{:.2%}'}
american_format = {'First Inning Lead Percentage': '{:.2%}', 'Fifth Inning Lead Percentage': '{:.2%}'}

@st.cache_resource(ttl=600)
def init_baselines():
    collection = dfs_db["NCAAF_GameModel"] 
    cursor = collection.find()
    raw_display = pd.DataFrame(list(cursor))
    game_model = raw_display[['Team', 'Opp', 'Win%', 'Vegas', 'Win% Diff', 'Win Line', 'Vegas Line', 'Line Diff', 'PD Spread', 'Vegas Spread', 'Spread Diff', 'O/U']]
    game_model = game_model.replace('', np.nan)
    game_model = game_model.sort_values(by='O/U', ascending=False)
    game_model.loc[:, ~game_model.columns.isin(['Team', 'Opp'])] = game_model.loc[:, ~game_model.columns.isin(['Team', 'Opp'])].apply(pd.to_numeric)

    collection = props_db["NCAAF_Props"] 
    cursor = collection.find()

    raw_display = pd.DataFrame(list(cursor))
    market_props = raw_display[['Name', 'Position', 'Projection', 'PropType', 'OddsType', 'over_pay', 'under_pay']]
    market_props['over_prop'] = market_props['Projection']
    market_props['over_line'] = market_props['over_pay'].apply(lambda x: (x - 1) * 100 if x >= 2.0 else -100 / (x - 1))
    market_props['under_prop'] = market_props['Projection']
    market_props['under_line'] = market_props['under_pay'].apply(lambda x: (x - 1) * 100 if x >= 2.0 else -100 / (x - 1))

    return game_model, market_props

def convert_df_to_csv(df):
    return df.to_csv().encode('utf-8')

def calculate_no_vig(row):
    def implied_probability(american_odds):
        if american_odds < 0:
            return (-american_odds) / ((-american_odds) + 100)
        else:
            return 100 / (american_odds + 100)

    over_line = row['over_line']
    under_line = row['under_line']
    over_prop = row['over_prop']
    
    over_prob = implied_probability(over_line)
    under_prob = implied_probability(under_line)
    
    total_prob = over_prob + under_prob
    no_vig_prob = (over_prob / total_prob + 0.5) * over_prop
    
    return no_vig_prob

prop_table_options = ['NCAAF_GAME_PLAYER_PASSING_ATTEMPTS', 'NCAAF_GAME_PLAYER_PASSING_COMPLETIONS', 'NCAAF_GAME_PLAYER_PASSING_INTERCEPTIONS',
                'NCAAF_GAME_PLAYER_PASSING_RUSHING_YARDS', 'NCAAF_GAME_PLAYER_PASSING_TOUCHDOWNS', 'NCAAF_GAME_PLAYER_PASSING_YARDS',
                'NCAAF_GAME_PLAYER_RECEIVING_RECEPTIONS', 'NCAAF_GAME_PLAYER_RECEIVING_TOUCHDOWNS', 'NCAAF_GAME_PLAYER_RECEIVING_YARDS',
                'NCAAF_GAME_PLAYER_RUSHING_ATTEMPTS', 'NCAAF_GAME_PLAYER_RUSHING_RECEIVING_YARDS', 'NCAAF_GAME_PLAYER_RUSHING_TOUCHDOWNS',
                'NCAAF_GAME_PLAYER_RUSHING_YARDS', 'NCAAF_GAME_PLAYER_SCORE_TOUCHDOWN']
prop_format = {'L3 Success': '{:.2%}', 'L6_Success': '{:.2%}', 'L10_success': '{:.2%}', 'Trending Over': '{:.2%}', 'Trending Under': '{:.2%}',
               'Implied Over': '{:.2%}', 'Implied Under': '{:.2%}', 'Over Edge': '{:.2%}', 'Under Edge': '{:.2%}'}

game_model, market_props = init_baselines()

tab1, tab2 = st.tabs(["Game Model", "Prop Market"])

with tab1:
    if st.button("Reset Data", key='reset1'):
              st.cache_data.clear()
              game_model, market_props = init_baselines()
    line_var1 = st.radio('How would you like to display odds?', options = ['Percentage', 'American'], key='line_var1')
    team_frame = game_model
    if line_var1 == 'Percentage':
        team_frame = team_frame[['Team', 'Opp', 'Win%', 'Vegas', 'Win% Diff', 'PD Spread', 'Vegas Spread', 'Spread Diff']]
        team_frame = team_frame.set_index('Team')
        try:
            st.dataframe(team_frame.style.background_gradient(axis=0).background_gradient(cmap='RdYlGn').background_gradient(cmap='RdYlGn_r', subset=['PD Spread', 'Vegas Spread', 'Spread Diff']).format(game_format, precision=2), use_container_width = True)
        except:
            st.dataframe(team_frame.style.background_gradient(axis=0).background_gradient(cmap='RdYlGn').background_gradient(cmap='RdYlGn_r', subset=['PD Spread', 'Vegas Spread', 'Spread Diff']).format(precision=2), use_container_width = True)
    if line_var1 == 'American':
        team_frame = team_frame[['Team', 'Opp', 'Win Line', 'Vegas Line', 'Line Diff', 'PD Spread', 'Vegas Spread', 'Spread Diff']]
        team_frame = team_frame.set_index('Team')
        st.dataframe(team_frame.style.background_gradient(axis=0).background_gradient(cmap='RdYlGn_r').format(precision=2), height = 1000, use_container_width = True)
    
    st.download_button(
        label="Export Team Model",
        data=convert_df_to_csv(team_frame),
        file_name='NCAAF_team_betting_export.csv',
        mime='text/csv',
        key='team_export',
    )

with tab2:
    if st.button("Reset Data", key='reset4'):
                st.cache_data.clear()
                game_model, market_props = init_baselines()
    market_type = st.selectbox('Select type of prop are you wanting to view', options = prop_table_options, key = 'market_type_key')
    disp_market = market_props.copy()
    disp_market = disp_market[disp_market['PropType'] == market_type]
    disp_market['No_Vig_Prop'] = disp_market.apply(calculate_no_vig, axis=1)
    fanduel_frame = disp_market[disp_market['OddsType'] == 'FANDUEL']
    fanduel_dict = dict(zip(fanduel_frame['Name'], fanduel_frame['No_Vig_Prop']))
    draftkings_frame = disp_market[disp_market['OddsType'] == 'DRAFTKINGS']
    draftkings_dict = dict(zip(draftkings_frame['Name'], draftkings_frame['No_Vig_Prop']))
    mgm_frame = disp_market[disp_market['OddsType'] == 'MGM']
    mgm_dict = dict(zip(mgm_frame['Name'], mgm_frame['No_Vig_Prop']))
    bet365_frame = disp_market[disp_market['OddsType'] == 'BET_365']
    bet365_dict = dict(zip(bet365_frame['Name'], bet365_frame['No_Vig_Prop']))

    disp_market['FANDUEL'] = disp_market['Name'].map(fanduel_dict)
    disp_market['DRAFTKINGS'] = disp_market['Name'].map(draftkings_dict)
    disp_market['MGM'] = disp_market['Name'].map(mgm_dict)
    disp_market['BET365'] = disp_market['Name'].map(bet365_dict)

    disp_market = disp_market[['Name', 'Position','FANDUEL', 'DRAFTKINGS', 'MGM', 'BET365']]
    disp_market = disp_market.drop_duplicates(subset=['Name'], keep='first', ignore_index=True)

    st.dataframe(disp_market.style.background_gradient(axis=1, subset=['FANDUEL', 'DRAFTKINGS', 'MGM', 'BET365'], cmap='RdYlGn').format(prop_format, precision=2), height = 1000, use_container_width = True)
    st.download_button(
        label="Export Market Props",
        data=convert_df_to_csv(disp_market),
        file_name='NCAAF_market_props_export.csv',
        mime='text/csv',
    )