tickerall

Official Python client for the TickerAll REST + WebSocket API.

Place trades, stream live market data, and manage broker sessions programmatically — without an MT4/MT5 terminal in the path. No Windows VM, no Wine, no MetaTrader5 terminal to babysit, no thread-safety workarounds.

pip install tickerall

Requires Python 3.9+. Depends only on httpx and websocket-client.

Why

The official MetaTrader5 Python package only runs on Windows, drives a local terminal over a single-threaded IPC channel, and falls over under concurrency. TickerAll hosts the broker connection for you and exposes it as a clean HTTP + WebSocket API, so your bot can run anywhere — Linux, macOS, a container, a Raspberry Pi — and stream ticks instead of polling.

Quickstart

from tickerall import Tickerall

client = Tickerall(api_key="cf_live_...")

# Connect a broker account → get a TickerAll account_id
session = client.sessions.start(
    broker="mt5",
    server="Exness-MT5Trial7",
    account=12345678,
    password="...",
)

# Place a market order
order = client.orders.place(
    session.account_id,
    type="market",
    symbol="BTCUSDm",
    side="BUY",
    volume=0.10,
    stop_loss=58000.0,
    take_profit=72000.0,
)
print(order.ticket, order.status)

client.sessions.end(session.account_id)

The client is a context manager too:

with Tickerall(api_key="cf_live_...") as client:
    ...

Mesh mode (advanced)

Mesh mode routes your broker traffic through a pool topology: you connect your accounts once against a mother endpoint, which warms them on a pool and hands back that pool's edge URL plus a short-lived grant. The SDK then sends all broker-data / trade / stream traffic to the edge (grant-authed) while roster and discovery stay on the mother — refreshing the grant, and following a pool migration, transparently. Opt in with mesh_mode=True; leaving it off is the normal direct connect, unchanged.

client = Tickerall(
    api_key="cf_live_...",           # authenticates you to the mother at discover time
    mesh_mode=True,
    mother_url="https://mesh.tickerall.com",
)

# Connect one or more accounts (they land on ONE pool together). Returns a
# stable account_ref per account — use it for every call and stream channel.
result = client.mesh.connect([
    {"broker": "mt5", "server": "Pepperstone-Demo", "account": 61542511, "password": "..."},
])
ref = result.accounts[0].account_ref

# From here the API is identical to direct mode — it just runs against the edge.
symbols = client.accounts.symbols(ref)
info    = client.accounts.account_info(ref)     # live balance/equity/…
stream  = client.stream.connect()
stream.subscribe_ticks(ref, ["BTCUSDm"])

The grant (1-hour TTL by default) is refreshed for you — proactively before it expires, and reactively on a 401 or a stream drop — so you never handle it. The account_ref is stable across refreshes and pool migrations, so subscriptions and queries keyed on it never need remapping. client.mesh.close() stops the background refresh.

Terminal type (MOBILE / WEB / CLIENT)

terminal_type picks which client the connection presents AS — "MOBILE" (the default), "WEB", or "CLIENT" (a desktop terminal). All expose the full surface (account, quotes, positions, history). The type sets the broker-assigned order origin (ENUM_DEAL_REASON): "MOBILE" → DEAL_REASON_MOBILE, "WEB" → DEAL_REASON_WEB, "CLIENT" → DEAL_REASON_CLIENT — useful where a venue distinguishes desktop-placed orders (e.g. some prop firms).

"WEB" requires the broker's web-terminal URL (web_terminal_url) — web terminals are per-broker-domain, so the URL must be supplied. "MOBILE" and "CLIENT" take neither web field:

session = client.sessions.start(
    broker="mt5",
    server="YourBroker-Server",
    account=12345678,
    password="...",
    terminal_type="WEB",
    web_terminal_url="https://mt5.yourbroker.com",  # required for WEB
    # web_endpoint="wss://host/path",               # optional WS override (rare)
)

For a desktop-origin (DEAL_REASON_CLIENT) connection — no web URL needed:

session = client.sessions.start(
    broker="mt5",
    server="YourBroker-Server",
    account=12345678,
    password="...",
    terminal_type="CLIENT",
)

Streaming — push, not poll

The stream runs on its own background thread. Register callbacks and go; it heartbeats, reconnects with backoff, and re-subscribes automatically.

client = Tickerall(api_key="cf_live_...")
session = client.sessions.start(broker="mt5", server="Exness-MT5Trial7",
                                account=12345678, password="...")

stream = client.stream.connect()
stream.on("tick", lambda e: print(e.symbol, e.bid, e.ask, e.timestamp))
stream.on("position", lambda e: print(e.event, e.position.ticket, e.position.profit))
stream.subscribe_ticks(session.account_id, ["BTCUSDm", "ETHUSDm"])
stream.subscribe_positions(session.account_id)

# ... your app runs ...
stream.close()

Keep an in-memory tick cache fresh (zero polling)

A common pattern: let the WebSocket fill a dict so price reads are O(1) with no network call — strictly better than polling a terminal per symbol.

latest: dict[str, "TickEvent"] = {}
stream = client.stream.connect()
stream.on("tick", lambda e: latest.__setitem__(e.symbol, e))
stream.subscribe_ticks(session.account_id, ["BTCUSDm", "ETHUSDm", "XAUUSDm"])

# Anywhere in your app — instant, no IPC, no thread-safety dance:
tick = latest.get("BTCUSDm")

Market data & history

# Historical OHLC candles (coarser timeframes reach further back)
bars = client.candles.get(session.account_id, symbol="BTCUSDm", hours=24, timeframe="M5")
for c in bars:
    print(c.timestamp, c.open, c.high, c.low, c.close)

# Closed-trade history (recent broker window)
trades = client.history.get(session.account_id, symbol="BTCUSDm", limit=100)

# Tradeable symbols and their volume specs
symbols = client.accounts.symbols(session.account_id)
specs = client.accounts.symbol_specs(session.account_id)  # min / max / step per symbol

# Remove an account from your roster (disconnects it + drops it from your list
# and billing; broker account and open positions are untouched). Reversible —
# reconnect the same login with sessions.start to re-add it.
client.accounts.remove(session.account_id)

Positions

detail = client.accounts.get(session.account_id)
for p in detail.positions:
    print(p.ticket, p.symbol, p.side, p.volume, p.profit)

client.positions.modify(session.account_id, ticket=p.ticket, stop_loss=60000.0)
client.positions.close(session.account_id, ticket=p.ticket)          # full close
client.positions.close(session.account_id, ticket=p.ticket, volume=0.05)  # partial

Always-hot sessions & transparent re-arm

For connections that must stay up across restarts, use keep_alive. The credentials live in this process's memory only (never persisted); if the account goes cold (e.g. TickerAll restarted), the next call transparently re-supplies them and retries once.

session = client.sessions.keep_alive(broker="mt5", server="Exness-MT5Trial7",
                                     account=12345678, password="...")
# ... later, after an outage, this just works — the client re-arms under the hood:
client.accounts.get(session.account_id)

# Stop keeping it alive (drops the cached credentials):
client.sessions.stop_keep_alive(session.account_id)

Reliability — idempotency & queue-and-replay

State-changing calls (sessions.start, orders.place, positions.close, positions.modify) carry a stable Idempotency-Key, so a retried call can't double-execute. By default a transient connectivity failure (TickerallServiceUnavailableError, .transient == True) fails fast so you can re-decide with fresh prices:

from tickerall import TickerallServiceUnavailableError

try:
    client.orders.place(account_id, type="market", symbol="BTCUSDm", side="BUY", volume=0.1)
except TickerallServiceUnavailableError:
    ...  # momentary blip — safe to retry

For price-insensitive orders (pending orders, SL/TP edits) you can instead queue-and-replay until connectivity returns:

client.orders.place(
    account_id, type="limit", symbol="BTCUSDm", side="BUY", volume=0.1, price=60000.0,
    queue_if_reconnecting=True, queue_max_s=60.0,
)

Errors

All errors derive from TickerallApiError and carry .status, .code, .request_id, .details, and .transient:

Using it from an async app

REST methods are synchronous and thread-safe, so call them from an event loop via asyncio.to_thread:

detail = await asyncio.to_thread(client.accounts.get, account_id)

The stream is already non-blocking (its own thread) — callbacks fire as events arrive.

License

MIT © Miguel Santos