app.py

import gradio as gr
import yfinance as yf
import pandas as pd
import numpy as np
import plotly.express as px

class ottimizzazioneLagrange:
    def __init__(self, tickers_string, period):
        # Pulizia stringa input
        self.tickers = [t.strip().upper() for t in tickers_string.split(',')]
        self.period = period
        self.dati_prezzi = None
        self.matrice_covarianza = None
        self.pesi_ottimali = None
        
    def run_optimization(self):
        # 1. Download dati (senza progress bar per evitare log sporchi)
        dati = yf.download(self.tickers, period=self.period, progress=False)
        if dati.empty: 
            return "Error: No data found / Errore: Dati non trovati.", None
        
        if isinstance(dati.columns, pd.MultiIndex):
            self.dati_prezzi = dati['Close']
        else:
            self.dati_prezzi = pd.DataFrame(dati['Close'])
            self.dati_prezzi.columns = self.tickers
            
        # 2. Calcolo Rendimenti e Matrice Covarianza
        rendimenti = self.dati_prezzi.pct_change().dropna()
        self.matrice_covarianza = rendimenti.cov() * 252
        
        # 3. Ottimizzazione Lagrange (Minima Varianza)
        n_asset = len(self.tickers)
        vettore1 = np.ones(n_asset)   
        cov_inversa = np.linalg.inv(self.matrice_covarianza.values)
        
        num = cov_inversa @ vettore1
        den = vettore1.T @ num
        self.pesi_ottimali = num / den 
        
        # 4. Risultati Testuali
        res = "GLOBAL MINIMUM VARIANCE PORTFOLIO\n" + "-"*35 + "\n"
        for i, ticker in enumerate(self.tickers):
            res += f"Asset: {ticker} | Weight: {self.pesi_ottimali[i]*100:.2f}%\n"
        
        # 5. Creazione Grafico Plotly
        df_plot = pd.DataFrame({
            'Asset': self.tickers,
            'Weight': self.pesi_ottimali * 100
        })
        fig = px.pie(df_plot, values='Weight', names='Asset', 
                     title="Portfolio Allocation (Markowitz Optimization)",
                     hole=0.3)
        
        return res, fig

def inference(tickers, period):
    try:
        model = ottimizzazioneLagrange(tickers, period)
        return model.run_optimization()
    except Exception as e:
        return f"System Error: {str(e)}", None

# UI Gradio
with gr.Blocks(theme=gr.themes.Default()) as demo:
    gr.Markdown("# 📈 Portfolio Optimization via Lagrange Multipliers")
    gr.Markdown("### Author: Dr. Ottaviani Flavio Rubens")
    
    with gr.Row():
        with gr.Column():
            in_tickers = gr.Textbox(label="Tickers (comma separated / separati da virgola)", value="AAPL, MSFT, GOOGL")
            in_period = gr.Dropdown(label="Analysis Period / Periodo", choices=["1y", "2y", "5y", "max"], value="1y")
            btn = gr.Button("Calculate / Calcola", variant="primary")
        
        with gr.Column():
            out_text = gr.Textbox(label="Results / Risultati", lines=8)
            out_plot = gr.Plot(label="Allocation Chart / Grafico Allocazione")

    btn.click(inference, inputs=[in_tickers, in_period], outputs=[out_text, out_plot])

if __name__ == "__main__":
    demo.launch()