app.py

import streamlit as st
st.set_page_config(layout="wide")
import numpy as np
import pandas as pd
import gspread
import pymongo

@st.cache_resource
def init_conn():
        scope = ['https://spreadsheets.google.com/feeds', 'https://www.googleapis.com/auth/drive']

        credentials = {
          "type": "service_account",
          "project_id": "model-sheets-connect",
          "private_key_id": "0e0bc2fdef04e771172fe5807392b9d6639d945e",
          "private_key": "-----BEGIN PRIVATE KEY-----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\nF/j9eUXfXjAlb042MphoF3ZC\n-----END PRIVATE KEY-----\n",
          "client_email": "[email protected]",
          "client_id": "100369174533302798535",
          "auth_uri": "https://accounts.google.com/o/oauth2/auth",
          "token_uri": "https://oauth2.googleapis.com/token",
          "auth_provider_x509_cert_url": "https://www.googleapis.com/oauth2/v1/certs",
          "client_x509_cert_url": "https://www.googleapis.com/robot/v1/metadata/x509/gspread-connection%40model-sheets-connect.iam.gserviceaccount.com"
        }
        uri = "mongodb+srv://multichem:[email protected]/?retryWrites=true&w=majority&appName=TestCluster"
        client = pymongo.MongoClient(uri, retryWrites=True, serverSelectionTimeoutMS=100000)
        db = client["testing_db"]

        collection = db["DK_MLB_seed_frame"] 
        cursor = collection.find()
    
        raw_display = pd.DataFrame(list(cursor))
        raw_display = raw_display[['SP1', 'SP2', 'C', '1B', '2B', '3B', 'SS', 'OF1', 'OF2', 'OF3', 'salary', 'proj', 'Team', 'Team_count', 'Secondary', 'Secondary_count']]
        DK_seed = raw_display.to_numpy()

        collection = db["FD_MLB_seed_frame"] 
        cursor = collection.find()
    
        raw_display = pd.DataFrame(list(cursor))
        raw_display = raw_display[['P', 'C_1B', '2B', '3B', 'SS', 'OF1', 'OF2', 'OF3', 'UTIL', 'salary', 'proj', 'Team', 'Team_count', 'Secondary', 'Secondary_count']]
        FD_seed = raw_display.to_numpy()
        
        MLB_Data = 'https://docs.google.com/spreadsheets/d/1f42Ergav8K1VsOLOK9MUn7DM_MLMvv4GR2Fy7EfnZTc/edit#gid=340831852'

        gc_con = gspread.service_account_from_dict(credentials, scope)

        client.close()
      
        return gc_con, client, db, DK_seed, FD_seed, MLB_Data

gcservice_account, client, db, DK_seed, FD_seed, MLB_Data = init_conn()

percentages_format = {'Exposure': '{:.2%}'}
dk_columns = ['SP1', 'SP2', 'C', '1B', '2B', '3B', 'SS', 'OF1', 'OF2', 'OF3', 'salary', 'proj', 'Team', 'Team_count', 'Secondary', 'Secondary_count']
fd_columns = ['P', 'C_1B', '2B', '3B', 'SS', 'OF1', 'OF2', 'OF3', 'UTIL', 'salary', 'proj', 'Team', 'Team_count', 'Secondary', 'Secondary_count']

@st.cache_data(ttl = 60)
def init_baselines():
    sh = gcservice_account.open_by_url(MLB_Data)
    
    worksheet = sh.worksheet('Main_ROO')
    load_display = pd.DataFrame(worksheet.get_all_records())
    load_display.replace('', np.nan, inplace=True)
    load_display['STDev'] = load_display['Median'] / 3
    
    dk_raw = load_display.dropna(subset=['Median'])
    
    worksheet = sh.worksheet('Main_FD_ROO')
    load_display = pd.DataFrame(worksheet.get_all_records())
    load_display.replace('', np.nan, inplace=True)
    load_display['STDev'] = load_display['Median'] / 3
    
    fd_raw = load_display.dropna(subset=['Median'])

    return dk_raw, fd_raw

@st.cache_data
def convert_df(array):
    array = pd.DataFrame(array, columns=column_names)
    return array.to_csv().encode('utf-8')

@st.cache_data
def calculate_DK_value_frequencies(np_array):
    unique, counts = np.unique(np_array[:, :9], return_counts=True)
    frequencies = counts / len(np_array)  # Normalize by the number of rows 
    combined_array = np.column_stack((unique, frequencies))  
    return combined_array 

@st.cache_data
def calculate_FD_value_frequencies(np_array):
    unique, counts = np.unique(np_array[:, :8], return_counts=True)
    frequencies = counts / len(np_array)  # Normalize by the number of rows 
    combined_array = np.column_stack((unique, frequencies))  
    return combined_array

@st.cache_data
def sim_contest(Sim_size, seed_frame, maps_dict):
    SimVar = 1
    Sim_Winners = []
    fp_array = seed_frame
    
    # Pre-vectorize functions
    vec_projection_map = np.vectorize(maps_dict['Projection_map'].__getitem__)
    vec_stdev_map = np.vectorize(maps_dict['STDev_map'].__getitem__)
    
    st.write('Simulating contest on frames')
    
    while SimVar <= Sim_size:
        fp_random = fp_array[np.random.choice(fp_array.shape[0], Contest_Size)]
            
        sample_arrays1 = np.c_[
            fp_random, 
            np.sum(np.random.normal(
                loc=vec_projection_map(fp_random[:, :-4]),
                scale=vec_stdev_map(fp_random[:, :-4])),
            axis=1)
        ]

        sample_arrays = sample_arrays1

        final_array = sample_arrays[sample_arrays[:, 10].argsort()[::-1]]
        best_lineup = final_array[final_array[:, -1].argsort(kind='stable')[::-1][:1]]
        Sim_Winners.append(best_lineup)
        SimVar += 1
        
    return Sim_Winners

dk_raw, fd_raw = init_baselines()

tab1, tab2 = st.tabs(['Data Export', 'Contest Sims'])
with tab1:
    col1, col2 = st.columns([1, 7])
    with col1:
        if st.button("Load/Reset Data", key='reset1'):
              st.cache_data.clear()
              for key in st.session_state.keys():
                  del st.session_state[key]
              dk_raw, fd_raw = init_baselines()
              
        slate_var1 = st.radio("Which data are you loading?", ('Main Slate', 'Other Main Slate'))
        site_var1 = st.radio("What site are you working with?", ('Draftkings', 'Fanduel'))
        if site_var1 == 'Draftkings':
            raw_baselines = dk_raw
            column_names = dk_columns
            
            team_var1 = st.radio("Do you want a frame with specific teams?", ('Full Slate', 'Specific Teams'), key='team_var1')
            if team_var1 == 'Specific Teams':
                    team_var2 = st.multiselect('Which teams do you want?', options = dk_raw['Team'].unique())
            elif team_var1 == 'Full Slate':
                    team_var2 = dk_raw.Team.values.tolist()
            
            stack_var1 = st.radio("Do you want a frame with specific stack sizes?", ('Full Slate', 'Specific Stack Sizes'), key='stack_var1')
            if stack_var1 == 'Specific Stack Sizes':
                    stack_var2 = st.multiselect('Which stack sizes do you want?', options = [5, 4, 3, 2, 1, 0])
            elif stack_var1 == 'Full Slate':
                    stack_var2 = [5, 4, 3, 2, 1, 0]
                    
        elif site_var1 == 'Fanduel':
            raw_baselines = fd_raw
            column_names = fd_columns
            
            team_var1 = st.radio("Do you want a frame with specific teams?", ('Full Slate', 'Specific Teams'), key='team_var1')
            if team_var1 == 'Specific Teams':
                    team_var2 = st.multiselect('Which teams do you want?', options = fd_raw['Team'].unique())
            elif team_var1 == 'Full Slate':
                    team_var2 = fd_raw.Team.values.tolist()
            
            stack_var1 = st.radio("Do you want a frame with specific stack sizes?", ('Full Slate', 'Specific Stack Sizes'), key='stack_var1')
            if stack_var1 == 'Specific Stack Sizes':
                    stack_var2 = st.multiselect('Which stack sizes do you want?', options = [4, 3, 2, 1, 0])
            elif stack_var1 == 'Full Slate':
                    stack_var2 = [4, 3, 2, 1, 0]
            
    with col2:
        if site_var1 == 'Draftkings':
            
            st.session_state.working_seed = DK_seed.copy()
            st.session_state.working_seed = st.session_state.working_seed[np.isin(st.session_state.working_seed[:, 12], team_var2)]
            st.session_state.working_seed = st.session_state.working_seed[np.isin(st.session_state.working_seed[:, 13], stack_var2)]
            st.session_state.data_export_display = pd.DataFrame(st.session_state.working_seed[0:1000], columns=column_names)
            
            st.session_state.data_export_freq = pd.DataFrame(calculate_DK_value_frequencies(st.session_state.working_seed), columns=['Player', 'Exposure'])
            st.session_state.data_export_freq = st.session_state.data_export_freq.sort_values(by='Exposure', ascending=False)
            
        elif site_var1 == 'Fanduel':
            
            st.session_state.working_seed = FD_seed.copy()
            st.session_state.working_seed = st.session_state.working_seed[np.isin(st.session_state.working_seed[:, 11], team_var2)]
            st.session_state.working_seed = st.session_state.working_seed[np.isin(st.session_state.working_seed[:, 12], stack_var2)]
            st.session_state.data_export_display = pd.DataFrame(st.session_state.working_seed[0:1000], columns=column_names)
            
            st.session_state.data_export_freq = pd.DataFrame(calculate_FD_value_frequencies(st.session_state.working_seed), columns=['Player', 'Exposure'])
            st.session_state.data_export_freq = st.session_state.data_export_freq.sort_values(by='Exposure', ascending=False)
            
        with st.container():
            if 'data_export_display' in st.session_state:
                st.dataframe(st.session_state.data_export_display.style.format(precision=2), height=500, use_container_width=True)
        with st.container():
            if 'data_export_freq' in st.session_state:
                st.dataframe(st.session_state.data_export_freq.style.format(percentages_format, precision=2), height=500, use_container_width=True)
        
        if st.button("Prepare data export", key='data_export'):
            data_export = st.session_state.working_seed.copy()
            st.download_button(
                label="Export optimals set",
                data=convert_df(data_export),
                file_name='MLB_optimals_export.csv',
                mime='text/csv',
            )
            
with tab2:
    col1, col2 = st.columns([1, 7])
    with col1:
        if st.button("Load/Reset Data", key='reset2'):
              st.cache_data.clear()
              for key in st.session_state.keys():
                  del st.session_state[key]
              dk_raw, fd_raw = init_baselines()
        sim_slate_var1 = st.radio("Which data are you loading?", ('Main Slate', 'Other Main Slate'), key='sim_slate_var1')
        sim_site_var1 = st.radio("What site are you working with?", ('Draftkings', 'Fanduel'), key='sim_site_var1')
        if sim_site_var1 == 'Draftkings':
            raw_baselines = dk_raw
        elif sim_site_var1 == 'Fanduel':
            raw_baselines = fd_raw
            
        contest_var1 = st.selectbox("What contest size are you simulating?", ('Small', 'Medium', 'Large', 'Massive'))
        if contest_var1 == 'Small':
            Contest_Size = 1000
        elif contest_var1 == 'Medium':
            Contest_Size = 5000
        elif contest_var1 == 'Large':
            Contest_Size = 10000
        elif contest_var1 == 'Massive':
            Contest_Size = 100000
        strength_var1 = st.selectbox("How sharp is the field in the contest?", ('Not Very', 'Average', 'Very'))
        if strength_var1 == 'Not Very':
            sharp_split = [400000,100000] 
        elif strength_var1 == 'Average':
            sharp_split = [500000,200000]
        elif strength_var1 == 'Very':
            sharp_split = [500000,300000]

    
    with col2:
        maps_dict = {
                'Floor_map':dict(zip(raw_baselines.Player,raw_baselines.Floor)),
                'Projection_map':dict(zip(raw_baselines.Player,raw_baselines.Median)),
                'Ceiling_map':dict(zip(raw_baselines.Player,raw_baselines.Ceiling)),
                'Salary_map':dict(zip(raw_baselines.Player,raw_baselines.Salary)),
                'Pos_map':dict(zip(raw_baselines.Player,raw_baselines.Position)),
                'Own_map':dict(zip(raw_baselines.Player,raw_baselines.Own)),
                'Small_Own_map':dict(zip(raw_baselines.Player,raw_baselines['Small Field Own%'])),
                'Large_Own_map':dict(zip(raw_baselines.Player,raw_baselines['Large Field Own%'])),
                'Team_map':dict(zip(raw_baselines.Player,raw_baselines.Team)),
                'STDev_map':dict(zip(raw_baselines.Player,raw_baselines.STDev)),
                'team_check_map':dict(zip(raw_baselines.Player,raw_baselines.Team))
                }
        Sim_Winners = sim_contest(500, st.session_state.working_seed)
        
        st.table(Sim_Winners.head(10))
                    
        # # Initial setup
        # Sim_Winner_Frame = pd.DataFrame(np.concatenate(Sim_Winners), columns=FinalPortfolio.columns.tolist() + ['Fantasy'])
        # Sim_Winner_Frame['GPP_Proj'] = (Sim_Winner_Frame['Projection'] + Sim_Winner_Frame['Fantasy']) / 2
        # Sim_Winner_Frame['unique_id'] = Sim_Winner_Frame['Projection'].astype(str) + Sim_Winner_Frame['Salary'].astype(str) + Sim_Winner_Frame['Own'].astype(str)
        # Sim_Winner_Frame = Sim_Winner_Frame.assign(win_count=Sim_Winner_Frame['unique_id'].map(Sim_Winner_Frame['unique_id'].value_counts()))
        
        # # Type Casting
        # type_cast_dict = {'Salary': int, 'Projection': np.float16, 'Fantasy': np.float16, 'GPP_Proj': np.float32}
        # Sim_Winner_Frame = Sim_Winner_Frame.astype(type_cast_dict)
        
        # del FinalPortfolio, insert_port, type_cast_dict
        
        # # Sorting
        # st.session_state.Sim_Winner_Frame = Sim_Winner_Frame.sort_values(by=['win_count', 'GPP_Proj'], ascending= [False, False]).copy().drop_duplicates(subset='unique_id').head(100)
        # st.session_state.Sim_Winner_Frame.drop(columns='unique_id', inplace=True)
        
        # # Data Copying
        # st.session_state.Sim_Winner_Export = Sim_Winner_Frame.copy()
        
        # # Data Copying
        # st.session_state.Sim_Winner_Display = Sim_Winner_Frame.copy()