How I built Beneish Pass App using Streamlit🎗️

Overview👀

Objectives🎯

Use Case👥

About the Beneish Model🔍

 

Creator - Professor Messod Beneish Purpose - to uncover financial manipulation by identifying the red flag areas Beneish M-score Equation

M - Score	Inference
< - 2.22	This company doesn't manipulate.
> -2.22	This company manipulates.

Beneish M-score - Wikipedia

The Beneish model is a statistical model that uses financial ratios calculated with accounting data of a specific company in order to check if it is likely (high probability) that the reported earnings of the company have been manipulated.

https://en.wikipedia.org/wiki/Beneish_M-score

Beneish M-Score

Beneish M score is the mathematical model which was created by Professor Messod Beneish and it is used for the purpose of finding out that whether the company has done any sort of manipulation with its earning with the help of the different financial ratios and the eight mentioned different variables.

https://www.wallstreetmojo.com/beneish-m-score/

Limitations❗

TOOLS USED📲

Data Fetching📝

from yahooquery import Ticker

# Fetching the Tickers Module

symbols = ['FB', 'AAPL', 'BRK.B', 'TSLA', 'MCD', 'VZ', 'BA', 'NKE', '^GSPC', 'NQ=F', 'ALB', 'AOS', 'APPS', 'AQB', 'ASPN', 'ATHM', 'AZRE', 'BCYC', 'BGNE', 'CAT', 'CC', 'CLAR', 'CLCT', 'CMBM', 'CMT', 'CRDF', 'CYD', 'DE', 'DKNG', 'EMN', 'FBIO', 'FBRX', 'FCX', 'FLXS', 'FMC', 'FMCI', 'GME','GRVY', 'HAIN', 'HBM', 'HIBB', 'IEX', 'IOR', 'GOOGL', 'MAXR', 'MPX', 'MRTX', 'NSTG', 'NVCR', 'NVO', 'OESX', 'PENN', 'PLL', 'PRTK', 'RDY', 'REGI', 'REKR', 'SBE', 'SQM', 'TCON', 'TWTR', 'TGB', 'TRIL', 'UEC', 'VCEL', 'VOXX', 'WIT', 'WKHS', 'XNCR']

# Create Ticker instance, passing symbols as first argument
# Optional asynchronous argument allows for asynchronous requests

tickers = Ticker(symbols, asynchronous=True)
dat = tickers.get_modules("summaryProfile quoteType")
symb = pd.DataFrame.from_dict(dat).T

# flatten dicts within each column, creating new dataframes

dataframes = [pd.json_normalize([x for x in symb[module] if isinstance(
    x, dict)]) for module in ['summaryProfile', 'quoteType']]

# concat dataframes from previous step

symb = pd.concat(dataframes, axis=1)
symb = symb[['shortName', 'symbol']].dropna()
symb = symb.sort_values('symbol')
symb.set_index('shortName', inplace=True, drop=True)
symb = symb.reset_index()  # reset index
symb.index = symb.index + 1  # add 1 to each index
symb.columns = ['Companies', 'Symbol']
data = symb.copy()
symb['Companies'] = symb['Companies'].str.replace("'", "''")

• The symbols for each company are put in a list, and then the company names are retrieved with yahooquery's Ticker module and put in a pandas dataframe.

• The single quotes in the company names make it hard to store them in the database, so the single quotes are replaced with double quotes.

Data Analysis 📊

import yfinance as yf

comp = yf.Ticker(symb.at[ch, 'Symbol'])
incomeStatement = comp.financials
balanceSheet = comp.balancesheet
cashFlow = comp.cashflow

# Cleaning the data

    # Income Statement
    incomeStatement = incomeStatement[incomeStatement.columns[0:2]]
    incomeStatement.columns = ['2022', '2021']
    incomeStatement = incomeStatement.fillna(0).astype(float)

    # Balance Sheet
    balanceSheet = balanceSheet[balanceSheet.columns[0:2]]
    balanceSheet.columns = ['2022', '2021']
    balanceSheet = balanceSheet.fillna(0).astype(float)

    # Cash Flow
    cashFlow = cashFlow[cashFlow.columns[0:2]]
    cashFlow.columns = ['2022', '2021']
    cashFlow.dropna()

Data Evaluation 🧮

def DSRI(df):
    return (df.at["Accounts Receivables", "2022"] / df.at["Revenue", "2022"]) / (df.at["Accounts Receivables", "2021"] / df.at["Revenue", "2021"])


def GMI(df):
    return ((df.at["Revenue", "2021"] - df.at["Cost of Goods Sold", "2021"])/df.at["Revenue", "2021"]) / ((df.at["Revenue", "2022"] - df.at["Cost of Goods Sold", "2022"])/df.at["Revenue", "2022"])


def AQI(df):
    AQI_t1 = (1 - (df.at["Current Assets", "2022"] +
              df.at["Property, Plant & Equipment", "2022"]+df.at["Securities", "2022"])) / df.at["Total Assets", "2022"]
    AQI_t2 = (1 - (df.at["Current Assets", "2021"] +
              df.at["Property, Plant & Equipment", "2021"]+df.at["Securities", "2021"])) / df.at["Total Assets", "2021"]
    return AQI_t1 / AQI_t2


def SGI(df):
    return (df.at["Revenue", "2022"] / df.at["Revenue", "2021"])


def DEPI(df):
    DEPI_t1 = (df.at["Depreciation", "2021"] / (df.at["Depreciation",
               "2021"] + df.at["Property, Plant & Equipment", "2021"]))
    DEPI_t2 = (df.at["Depreciation", "2022"] / (df.at["Depreciation",
               "2022"] + df.at["Property, Plant & Equipment", "2022"]))
    return DEPI_t1 / DEPI_t2


def SGAI(df):
    return (df.at["Selling, General & Admin.Expense", "2022"] / df.at["Revenue", "2022"]) / (df.at["Selling, General & Admin.Expense", "2021"] / df.at["Revenue", "2021"])


def LVGI(df):
    return ((df.at["Current Liabilities", "2022"] + df.at["Total Long-term Debt", "2022"]) / df.at["Total Assets", "2022"]) / ((df.at["Current Liabilities", "2021"] + df.at["Total Long-term Debt", "2021"]) / df.at["Total Assets", "2021"])


def TATA(df):
    return (df.at["Net Income from Continuing Operations", "2022"] - df.at["Cash Flow from Operations", "2022"]) / df.at["Total Assets", "2022"]


def BeneishMScore(dsri, gmi, aqi, sgi, depi, sgai, lvgi, tata):
    return -4.84+0.92*dsri+0.528*gmi+0.404*aqi+0.892*sgi+0.115*depi-0.172*sgai+4.679*tata-0.327*lvgi

# Beneish M Score
    m_score = BeneishMScore(DSRI(data),
                            GMI(data),
                            AQI(data),
                            SGI(data),
                            DEPI(data),
                            SGAI(data),
                            LVGI(data),
                            TATA(data))
    if(m_score < -2.22):
        res = '##### Company is not likely to manipulate their earnings'
        st.write(f"##### M- Score = {round(m_score,2)}")
        st.write(f"{res}")
        # print(res)
    else:
        res = " ##### Company is not likely to manipulate their earnings"
        st.write(f"##### M- Score = {round(m_score,2)}")
        st.write(f"{res}"

Visualization 🎴

import plotly.express as px

# The Line Chart using Plotly
    fig = px.line(
        temp_ratios,  # Data Frame
        x="Index",  # Columns from the data frame
        y="Ratios",
        title="Financial Ratio Indexes",
    )
    fig.update_traces(line_color="blue")

 with st.container():
      st.plotly_chart(fig)

Data Warehousing 💽

Requirement - Snowflake Account

Connect Streamlit to Snowflake - Streamlit Docs

This guide explains how to securely access a Snowflake database from Streamlit Cloud. It uses the snowflake-connector-python library and Streamlit's secrets management.

https://docs.streamlit.io/knowledge-base/tutorials/databases/snowflake

import snowflake.connector as sf

# SnowFlake  Initialize connection.
    def init_connection():
        return sf.connect(**st.secrets["snowflake"])

    conn = init_connection()

    cur = conn.cursor()

try:
    cur.execute(
        f"INSERT INTO FAR.PUBLIC.HISTORY(COMPANY,M_SCORE) VALUES('{symb.at[ch, 'Companies']}',{round(m_score,2)})")
     cur.execute('''DELETE FROM FAR.PUBLIC.HISTORY WHERE (COMPANY)  in 
        (SELECT COMPANY FROM FAR.PUBLIC.HISTORY GROUP BY COMPANY HAVING COUNT(COMPANY)> 1)
                    ''')
     cur.execute(
            'SELECT * FROM FAR.PUBLIC.HISTORY')
    finally:
        cur.close()
    conn.close()

• Once the user makes an entry for the first time, the Company name and its m-score are saved.

• When a user makes an entry for the same company, duplicate entries are deleted and updated in the database again. This is done to get rid of redundant data and make the best use of space.

Automation🤖

from openpyxl import Workbook, load_workbook

# Excel Automation

    wb = load_workbook('Beneish-M-Score.xlsx')
    ws = wb.active

    while True:   # repeat until the try statement succeeds
        try:
            # or "a+", whatever you need
            myfile = open("Beneish-M-Score.xlsx", "r+")
            break
            # exit the loop
        except IOError:
            print("Failed to open!!")

    ws['C3'].value = symb.at[ch, 'Symbol']
    ws['C4'].value = symb.at[ch, 'Companies']

    excel_data = data.reset_index(drop=True)
    excel_data.index = excel_data.index + 1

    for row in range(8, 21):
        for col in range(3, 5):
            char = chr(65 + col)
            if char == 'D':
                ws[char + str(row)] = excel_data.at[row-7, "2022"]
            if char == 'E':
                ws[char + str(row)] = excel_data.at[row-7, "2021"]

    wb.save('Beneish-M-Score.xlsx')

• On Beforehand, a template is made for calculating the beneish workbook, and the workbook is loaded.

• The process is so easy that all the user has to do is make an entry for a certain company. Using automation, the excel sheet is used to update the data in the appropriate cells.

Streamlit⚡

Streamlit is a Python-based open source application framework. Our data science and machine learning web apps are much faster to develop thanks to it. Scikit-Learn, Keras, PyTorch, SymPy(latex), NumPy, pandas, and Matplotlib are just a few of the popular Python libraries it works with.

import streamlit as st
from streamlit_lottie import st_lottie
import hydralit_components as hc


# Streamlit Page Configuration

st.set_page_config(
    page_title="FA",
    page_icon="chart_with_upwards_trend",
    layout="wide",
    initial_sidebar_state="expanded"
)

hide_st_style = """
<style>
#MainMenu {visibility: hidden;}
header{visibility: hidden;}
footer {visibility: hidden;}
</style>

"""
st.markdown(hide_st_style, unsafe_allow_html=True)

# ---Side-Bar----

with st.sidebar:
    st_lottie(lottie_hello, loop=True, key=None, height=320, width=320)
    st.write(''' It's a DEMO!! ''')

def load_lottiefile(filepath: str):
    with open(filepath, "r") as f:
        return json.load(f)


lottie_analysis = load_lottiefile("lottiefiles/analysis.json")
lottie_hello = load_lottiefile("lottiefiles/hello.json")

# -----Home Page-----

st.title("Analyzing the Quality of Financial Statements using Beneish Model")
with st.container():
    left_col, right_col = st.columns((2, 1))
    with left_col:
        st.dataframe(data)
    with right_col:
        st_lottie(lottie_analysis, height="300",
                  width="500", quality="high", key=None)
# -- Input----

ch = st.number_input('Enter a  number: ')

with hc.HyLoader('Now doing loading', hc.Loaders.standard_loaders, index=[3, 0, 5]):

# The Line Chart using Plotly
    fig = px.line(
        temp_ratios,  # Data Frame
        x="Index",  # Columns from the data frame
        y="Ratios",
        title="Financial Ratio Indexes",
    )
    fig.update_traces(line_color="blue")

    with st.container():
        st.plotly_chart(fig)

# Streamlit Buttons
history = cur.fetch_pandas_all()
if st.button("View History"):
        st.snow()
        st.dataframe(history)

18/March/23 Updations:

TOOLS USED📲

• Complete Code base revamping

• Hydralit Components update

• Incorporating more tickers using StockSymbol API instead of yahooquery
  yfin

  legout • Updated Jul 8, 2023

• Using yfin (beta) for the financial statements

• Clean code writing

• Design Modification of streamlit (use streamlit-option-menu) - attach github corners

• Database to duckdb

• Containerized the app using docker https://hub.docker.com/repository/docker/vidhyavarshany/beneishpass/general