pip install streamlit transformers torch requests pandas plotly

# data.py — Fetch Bitcoin market data from CoinGecko API
import requests
import pandas as pd
from datetime import datetime

def get_bitcoin_price():
    """Fetch current Bitcoin price and 24h metrics"""
    url = "https://api.coingecko.com/api/v3/coins/bitcoin"
    params = {
        "localization": "false",
        "tickers": "false",
        "community_data": "false",
        "developer_data": "false"
    }
    response = requests.get(url, params=params, timeout=10)
    data = response.json()
    market = data["market_data"]

    return {
        "price": market["current_price"]["usd"],
        "change_24h": market["price_change_percentage_24h"],
        "change_7d": market["price_change_percentage_7d"],
        "volume": market["total_volume"]["usd"],
        "market_cap": market["market_cap"]["usd"],
        "ath": market["ath"]["usd"],
        "ath_change": market["ath_change_percentage"]["usd"],
        "last_updated": data["last_updated"]
    }

def get_price_history(days=7):
    """Fetch Bitcoin price history for chart"""
    url = f"https://api.coingecko.com/api/v3/coins/bitcoin/market_chart"
    params = {"vs_currency": "usd", "days": days, "interval": "hourly"}
    response = requests.get(url, params=params, timeout=10)
    data = response.json()

    prices = data["prices"]
    df = pd.DataFrame(prices, columns=["timestamp", "price"])
    df["datetime"] = pd.to_datetime(df["timestamp"], unit="ms")
    return df

# sentiment.py — AI sentiment analysis using FinBERT
from transformers import pipeline
import streamlit as st

# Load model once — @st.cache_resource prevents reloading every refresh
@st.cache_resource
def load_sentiment_model():
    """Load FinBERT — financial sentiment analysis model"""
    return pipeline(
        "text-classification",
        model="ProsusAI/finbert",
        return_all_scores=True
    )

def analyze_sentiment(texts: list[str]) -> dict:
    """
    Analyze a list of news headlines.
    Returns: sentiment label, score (0-100), and per-headline results.
    """
    model = load_sentiment_model()
    results = []

    for text in texts:
        scores = model(text[:512])[0]  # FinBERT max 512 tokens
        score_dict = {s["label"]: s["score"] for s in scores}
        label = max(score_dict, key=score_dict.get)
        confidence = score_dict[label]
        results.append({
            "text": text[:80] + ("..." if len(text) > 80 else ""),
            "label": label,
            "confidence": round(confidence * 100, 1)
        })

    # Calculate overall sentiment score 0-100
    pos_count = sum(1 for r in results if r["label"] == "positive")
    neg_count = sum(1 for r in results if r["label"] == "negative")
    total = len(results)
    score = round((pos_count / total) * 100) if total > 0 else 50

    overall = "Bullish" if score > 60 else "Bearish" if score < 40 else "Neutral"

    return {
        "score": score,
        "label": overall,
        "positive": pos_count,
        "negative": neg_count,
        "neutral": total - pos_count - neg_count,
        "headlines": results
    }

# app.py — AI Bitcoin Sentiment Monitor | AIByTec
import streamlit as st
import plotly.graph_objects as go
import plotly.express as px
from data import get_bitcoin_price, get_price_history
from sentiment import analyze_sentiment
from datetime import datetime

# ── PAGE CONFIG ──────────────────────────────────
st.set_page_config(
    page_title="AI Bitcoin Monitor | AIByTec",
    page_icon="🤖",
    layout="wide",
    initial_sidebar_state="collapsed"
)

# ── SAMPLE HEADLINES (replace with real news API) ─
HEADLINES = [
    "Bitcoin falls below $63,000 amid extreme investor fear",
    "US tariff proposals trigger crypto market selloff",
    "Bitcoin ETF sees $203 million in outflows this week",
    "CZ predicts Bitcoin could reach $200,000 this cycle",
    "Fear and Greed Index hits record low of 5",
    "Agentic AI systems now control 20% of crypto trading",
    "Institutional investors see opportunity in Bitcoin dip",
    "Prediction markets give 90% odds of sub-$60K Bitcoin",
]

# ── MAIN APP ─────────────────────────────────────
st.title("🤖 AI Bitcoin Sentiment Monitor")
st.caption("Powered by FinBERT + CoinGecko API | Built with AIByTec")
st.divider()

# Load data with spinner
with st.spinner("Fetching live Bitcoin data..."):
    btc = get_bitcoin_price()
    history = get_price_history(days=7)

# ── PRICE METRICS ROW ─────────────────────────────
col1, col2, col3, col4 = st.columns(4)
with col1:
    st.metric("BTC Price", f"${btc['price']:,.2f}",
              f"{btc['change_24h']:.2f}%")
with col2:
    st.metric("7D Change", f"{btc['change_7d']:.2f}%")
with col3:
    st.metric("From ATH", f"{btc['ath_change']:.1f}%")
with col4:
    vol_b = btc['volume'] / 1e9
    st.metric("24h Volume", f"${vol_b:.2f}B")

st.divider()

# ── PRICE CHART ───────────────────────────────────
fig = go.Figure()
fig.add_trace(go.Scatter(
    x=history["datetime"], y=history["price"],
    mode="lines", name="BTC/USD",
    line=dict(color="#f7931a", width=2),
    fill="tozeroy",
    fillcolor="rgba(247,147,26,0.05)"
))
fig.update_layout(
    title="Bitcoin Price — 7 Day History",
    paper_bgcolor="rgba(0,0,0,0)",
    plot_bgcolor="rgba(0,0,0,0)",
    font=dict(color="#d4d0e8"),
    xaxis=dict(gridcolor="rgba(255,255,255,0.05)"),
    yaxis=dict(gridcolor="rgba(255,255,255,0.05)", tickprefix="$"),
    height=350
)
st.plotly_chart(fig, use_container_width=True)

# ── AI SENTIMENT ANALYSIS ─────────────────────────
st.subheader("🧠 AI Sentiment Analysis")
with st.spinner("Running FinBERT sentiment analysis..."):
    sentiment = analyze_sentiment(HEADLINES)

col_s1, col_s2, col_s3 = st.columns([1, 2, 1])
with col_s1:
    color = "🟢" if sentiment["label"] == "Bullish" else "🔴"
    st.metric("AI Sentiment",
              f"{color} {sentiment['label']}",
              f"{sentiment['score']}/100")

with col_s2:
    df_sent = px.bar(
        x=["Positive", "Neutral", "Negative"],
        y=[sentiment["positive"], sentiment["neutral"], sentiment["negative"]],
        color=["Positive", "Neutral", "Negative"],
        color_discrete_map={"Positive":"#00e5a0","Neutral":"#7b3fe4","Negative":"#ff4d6d"},
        title="Sentiment Distribution",
        height=200
    )
    df_sent.update_layout(paper_bgcolor="rgba(0,0,0,0)",
                          plot_bgcolor="rgba(0,0,0,0)",
                          showlegend=False,
                          font=dict(color="#d4d0e8"))
    st.plotly_chart(df_sent, use_container_width=True)

# ── HEADLINE TABLE ────────────────────────────────
st.subheader("📰 News Headlines — AI Sentiment Scores")
for h in sentiment["headlines"]:
    icon = "🟢" if h["label"] == "positive" else "🔴" if h["label"] == "negative" else "⚪"
    st.write(f"{icon} **{h['label'].upper()}** ({h['confidence']}%) — {h['text']}")

st.caption(f"Last updated: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')} | Model: ProsusAI/FinBERT")

streamlit>=1.32.0
transformers>=4.38.0
torch>=2.2.0
requests>=2.31.0
pandas>=2.1.0
plotly>=5.18.0