|
import streamlit as st |
|
import pandas as pd |
|
import numpy as np |
|
import optuna |
|
import plotly.express as px |
|
|
|
st.title("Portfolio weights calculator") |
|
|
|
help_string = "NOTA: El formato utilizado aquí es llamando cada columna de GOOGLEFINANCE." |
|
|
|
"Stocks - Sheet1.csv" |
|
|
|
check_box = st.checkbox("¿Deseas usar el archivo precargado?") |
|
|
|
if check_box: |
|
uploaded_file = "Stocks - Sheet1.csv" |
|
file_name = uploaded_file |
|
else: |
|
uploaded_file = st.file_uploader("Sube aquí tu archivo de excel", type=[".xls", ".xlsx", ".csv"], help=help_string) |
|
file_name = uploaded_file.name if uploaded_file is not None else None |
|
|
|
if uploaded_file is not None: |
|
|
|
if file_name[-3:] == "csv": |
|
df = pd.read_csv(uploaded_file) |
|
else: |
|
df = pd.read_excel(uploaded_file) |
|
|
|
df = df.drop(0, axis=0) |
|
df = df.drop("Unnamed: 2", axis=1).drop("Unnamed: 4", axis=1).rename({"Unnamed: 0": "Date"}, axis=1) |
|
|
|
df['Date'] = pd.to_datetime(df['Date']).dt.date |
|
|
|
stocks = list(df.columns)[-3:] |
|
stocks_rets = [] |
|
|
|
for i in stocks: |
|
stocks_rets.append(i+"_ret") |
|
df[i] = df[i].astype(float) |
|
df[i+"_ret"] = (df[i] - df[i].shift(1)) / df[i].shift(1) |
|
|
|
st.write(df[["Date"] + stocks_rets]) |
|
|
|
|
|
fig = px.line(df, x=df.Date, y=stocks, labels={'value': 'Value', 'variable': 'Series'}, title='Time Series Plot') |
|
fig.update_layout(xaxis_title='Date', yaxis_title='Value') |
|
|
|
|
|
st.plotly_chart(fig) |
|
|
|
ret_list = df[stocks_rets].mean().to_numpy().reshape(-1, 1) |
|
cov_matrix = df[stocks_rets].cov().to_numpy() |
|
|
|
optim_choice = st.selectbox("Elige la forma de optomizar :", ("max returns", "min variance", "max returns - variance")) |
|
|
|
def portfolio_variance(weights, covariance_matrix): |
|
return np.dot(weights.T, np.dot(covariance_matrix, weights)) |
|
|
|
def portfolio_returns(weights, expected_returns): |
|
return np.dot(weights.T, expected_returns) |
|
|
|
if optim_choice == "max returns": |
|
def objective(trial): |
|
w1 = trial.suggest_uniform('w1', 0, 1) |
|
w2 = trial.suggest_uniform('w2', 0, 1) |
|
w3 = 1 - w1 - w2 |
|
weights = np.array([w1, w2, w3]).reshape(-1, 1) |
|
return np.dot(weights.T, ret_list) |
|
|
|
study = optuna.create_study(direction="maximize") |
|
study.optimize(objective, n_trials=100, show_progress_bar=True) |
|
|
|
elif optim_choice == "min variance": |
|
def objective(trial): |
|
w1 = trial.suggest_uniform('w1', 0, 1) |
|
w2 = trial.suggest_uniform('w2', 0, 1) |
|
w3 = 1 - w1 - w2 |
|
weights = np.array([w1, w2, w3]).reshape(-1, 1) |
|
return np.dot(weights.T, np.dot(cov_matrix, weights)) |
|
|
|
study = optuna.create_study(direction="minimize") |
|
study.optimize(objective, n_trials=100, show_progress_bar=True) |
|
|
|
else: |
|
def objective(trial): |
|
w1 = trial.suggest_uniform('w1', 0, 1) |
|
w2 = trial.suggest_uniform('w2', 0, 1) |
|
w3 = 1 - w1 - w2 |
|
weights = np.array([w1, w2, w3]).reshape(-1, 1) |
|
return np.dot(weights.T, ret_list) - np.dot(weights.T, np.dot(cov_matrix, weights)) |
|
|
|
study = optuna.create_study(direction="maximize") |
|
study.optimize(objective, n_trials=100, show_progress_bar=True) |
|
|
|
w1 = study.best_params['w1'] |
|
w2 = study.best_params['w2'] |
|
w3 = 1- w1 - w2 |
|
|
|
weights = np.array([w1, w2, w3]).reshape(-1, 1) |
|
|
|
yearly_returns = (1 + np.dot(weights.T, ret_list)[0, 0]) ** 252 - 1 |
|
yearly_variance = np.dot(weights.T, np.dot(cov_matrix, weights))[0, 0] * 252 |
|
|
|
st.write(f"Los pesos son: :green[{stocks[0]} -> {w1:,.4f}], :green[{stocks[1]} -> {w2:,.4f}], :green[{stocks[2]} -> {w3:,.4f}]") |
|
st.write(f"El retorno anualizado del portafolio es: :green[{yearly_returns:,.4f}]") |
|
st.write(f"La varianza anualizado del portafolio es: :green[{yearly_variance:,.4f}]") |
|
|