Full Documentation: Trading Bot

1. System Architecture

The Trading Bot is not just a script, but a complex ecosystem that joins technical analysis, artificial intelligence, and real-time execution. It was designed to operate 24/7 without human intervention, utilizing an "Integrity Shield" to avoid operations during data failure moments.

Data Layer (Bybit API) Direct connection with the exchange to fetch prices (Candles), Order Book, and Real Balance.

Analytical Layer (Quantitative Engine) Mathematical processing of indicators like EMA200, RSI, ADX, and ATR to define the market regime.

Decision Layer (AI Tribunal) Google Gemini 1.5 Flash acts as the final JUDGE, validating if technical analysis makes sense within the macro context.

2. The Python Engine (Backend)

The heart of the project is in backend/services/bybit_service.py. It manages all communication with the exchange. Below is the real integration class code, with protected credentials.

Python / Bybit ServiceCORE

import os
from pybit.unified_trading import HTTP

class BybitEngine:
    def __init__(self):
        # Credentials loaded from environment variables for security
        self.api_key = os.getenv("BYBIT_API_KEY") # SUBST_YOUR_KEY
        self.api_secret = os.getenv("BYBIT_API_SECRET") # SUBST_YOUR_SECRET

        # Unified Trading Account (UTA) Session Initialization
        self.session = HTTP(
            testnet=False,
            api_key=self.api_key,
            api_secret=self.api_secret,
        )

    def get_candles(self, symbol="BTCUSDT", interval="1h", limit=100):
        """Fetches historical candles for technical analysis"""
        res = self.session.get_kline(
            category="spot",
            symbol=symbol,
            interval="60", # 1 hour
            limit=limit
        )
        return res['result']['list'] # Returns OHLCV candle list

For Beginners:

The code above creates a "phone line" to talk to the Bybit exchange. The get_candles method asks the exchange for the price history (candles). Without this history, the bot wouldn't know if the price went up or down.

3. Indicator Intelligence

In indicators.py, we transform raw prices into logical decisions. The bot uses the EMA200 (Exponential Moving Average of 200 periods) to know if the long-term trend is bullish or bearish.

Python / Quantitative EngineMATH

def calculate_all(candles_1h):
    df = pd.DataFrame(candles_1h)

    # 200-period average: The water divider
    df['ema200'] = df['close'].ewm(span=200, adjust=False).mean()

    # RSI: Indicates if the market is "expensive" or "cheap"
    # RSI > 70 (Overbought) | RSI < 30 (Oversold)
    delta = df['close'].diff()
    gain = (delta.where(delta > 0, 0)).rolling(window=14).mean()
    loss = (-delta.where(delta < 0, 0)).rolling(window=14).mean()
    df['rsi'] = 100 - (100 / (1 + (gain / loss)))

    # Market Regime:
    last = df.iloc[-1]
    if last['close'] > last['ema200']:
        regime = "UPTREND"
    else:
        regime = "DOWNTREND"

⚠️ Safety Rule: If the bot detects the price is too far from the EMA200 or the volume is low (RVOL < 0.1), it enters safety mode and does not open orders.

4. The AI Tribunal (Gemini 1.5)

The main differentiator of this project is using Artificial Intelligence to filter false signals. The bot sends all technical data to Gemini and asks: "Is it worth entering?".

Python / Gemini ServiceAI

class GeminiAnalyst:
    def analyze_trend(self, indicators):
        prompt = f"""
        You are the SUPREME JUDGE of the Nômade Tribunal.
        Analyze the data and reply if we should BUY or WAIT.
        DATA: {indicators}
        Reply in JSON: {{"signal": "BUY", "score": 85, "reasoning": "..."}}
        """
        response = self.model.generate_content(prompt)
        return response.text

5. Front-end Dashboard

Bot control is done via a high-tech web interface using trading.html and market.html. It connects to Firebase to show real-time data without needing to refresh the page.

HTML5 / JS / FirebaseUI/UX

<!-- Real-time monitoring logic example -->
<script type="module">
    import { db } from './firebase-config.js';
    import { ref, onValue } from "https://www.gstatic.com/firebasejs/10.12.0/firebase-database.js";

    // Listens to the current Bitcoin price in the database
    onValue(ref(db, 'market/BTCUSDT/current'), (snapshot) => {
        const data = snapshot.val();
        document.getElementById('btc-price').innerText = `$ ${data.price}`;
    });
</script>