MetaTrader 5 API with Python: automate Brazil's B3 exchange and Forex.

MQL5 (MetaTrader's native language) is powerful, but Python has an unbeatable ecosystem: pandas, numpy, scikit-learn, ready-made indicator libraries. The official MetaTrader5 library gives you the best of both worlds — you use Python to think and MT5 to execute. And the part that matters most to you: it is the most accessible way to automate Brazil's B3 (mini index and mini dollar futures) with Python.

01Installation and prerequisites

You need: Windows, your broker's MT5 terminal installed and logged in, and Python 3.11+. Then:

pip install MetaTrader5 pandas

02Connecting to the terminal

With MT5 open and logged into your account (use a demo account!), initialize the connection:

# conectar.py
import MetaTrader5 as mt5

# initializes the connection to the open MT5 terminal
if not mt5.initialize():
    print("Falha ao conectar:", mt5.last_error())
    quit()

# account info (confirm it is a DEMO account)
conta = mt5.account_info()
print(f"Conta: {conta.login} | Saldo: {conta.balance} | Servidor: {conta.server}")

# always close the connection when you are done
# mt5.shutdown()

If you need to log in from code (instead of manually in the terminal), pass credentials to initialize:

mt5.initialize(login=12345678, password="sua_senha", server="NomeDoServidor-Demo")

03Pulling candles (including from B3)

This is the part that plugs into your indicators. Pull candles for any asset your broker offers — including WIN (mini index) and WDO (mini dollar) if it gives you access to B3:

# dados.py
import MetaTrader5 as mt5
import pandas as pd
from datetime import datetime

mt5.initialize()

# symbol: e.g. "WIN$N" (mini index), "WDO$N" (mini dollar) or "EURUSD"
# the exact symbol name varies by broker — check it in the Market Watch
simbolo = "WIN$N"
timeframe = mt5.TIMEFRAME_M5   # M1, M5, M15, H1, etc.

# fetches the last 500 candles
rates = mt5.copy_rates_from_pos(simbolo, timeframe, 0, 500)
df = pd.DataFrame(rates)
df["time"] = pd.to_datetime(df["time"], unit="s")
print(df[["time","open","high","low","close","tick_volume"]].tail())

04Computing a signal with your indicators

This is where Python shines — use the indicator functions you already learned in our guides. The DataFrame that came from MT5 plugs straight in:

# reuse the functions from the indicator guides
from indicadores import calcular_rsi, calcular_macd   # your modules

df["rsi"] = calcular_rsi(df["close"])
macd, sinal, hist = calcular_macd(df["close"])

# example of a combined signal
u = df.iloc[-1]
sinal_compra = (u["rsi"] < 70) and (macd.iloc[-1] > sinal.iloc[-1])

05Sending an order

Sending an order in MT5 is more verbose than on Binance, because you build a detailed "request". Example of a market buy with stop and target:

# executar.py
def comprar(simbolo, lote, stop_pts, alvo_pts):
    info = mt5.symbol_info_tick(simbolo)
    preco = info.ask
    point = mt5.symbol_info(simbolo).point

    request = {
        "action": mt5.TRADE_ACTION_DEAL,
        "symbol": simbolo,
        "volume": lote,
        "type": mt5.ORDER_TYPE_BUY,
        "price": preco,
        "sl": preco - stop_pts * point,   # stop loss
        "tp": preco + alvo_pts * point,   # take profit
        "deviation": 20,                   # allowed slippage
        "magic": 123456,                  # your bot's id
        "type_filling": mt5.ORDER_FILLING_FOK,
    }
    resultado = mt5.order_send(request)
    if resultado.retcode != mt5.TRADE_RETCODE_DONE:
        print("Erro na ordem:", resultado.retcode, resultado.comment)
    return resultado

06Managing positions

For a complete bot, you need to query and close positions:

# list your bot's open positions (by magic)
posicoes = mt5.positions_get(symbol="WIN$N")
for p in posicoes:
    if p.magic == 123456:
        print(f"Posição {p.ticket}: lucro {p.profit}")

# close a position
def fechar(posicao):
    tick = mt5.symbol_info_tick(posicao.symbol)
    request = {
        "action": mt5.TRADE_ACTION_DEAL,
        "symbol": posicao.symbol,
        "volume": posicao.volume,
        "type": mt5.ORDER_TYPE_SELL if posicao.type == 0 else mt5.ORDER_TYPE_BUY,
        "position": posicao.ticket,
        "price": tick.bid if posicao.type == 0 else tick.ask,
        "deviation": 20,
        "type_filling": mt5.ORDER_FILLING_FOK,
    }
    return mt5.order_send(request)

07The bot loop

Putting it all together, the skeleton of an MT5 bot:

import time

mt5.initialize()
try:
    while True:
        df = puxar_candles("WIN$N", mt5.TIMEFRAME_M5)
        sinal = calcular_sinal(df)              # your strategy
        atr = calcular_atr(df).iloc[-1]        # dynamic stop

        if sinal == "COMPRA" and not tem_posicao():
            lote = dimensionar(atr)             # risk management
            comprar("WIN$N", lote, stop_pts=2*atr, alvo_pts=4*atr)

        time.sleep(60)                       # checks every minute
except Exception as e:
    print("Erro:", e)
finally:
    mt5.shutdown()

08Frequently asked questions