Create account
TickFeed — Field guide 01Gold & forex market data · REST + MCP

Build your
AI trading
bot.

Give your AI agent live gold and forex market data. Connect TickFeed, build your strategy, and start testing.

Build my AI trading bot See how it works

$25 free credit for new accounts until 10 Nov. No card needed.

Example conversationClaude + TickFeed MCP
You

Is gold trading right now?

Agent

Calling get_price("XAUUSD") on TickFeed…

TickFeed
Real response · captured 3 Nov 2026, 21:38 UTC 
{
  "symbol": "XAUUSD",
  "bid": 3927.98,
  "ask": 3928.15,
  "spread": 0.17,
  "spread_points": 17,
  "point": 0.01,
  "tick_time": "2026-11-03T21:38:25.869000Z",
  "age_seconds": 0.184,
  "freshness": "fresh",
  "market_open": true,
  "stale_after_seconds": 120,
  "source": "live"
}
Agent

Yes. XAUUSD is open, and the last quote was 0.2 s old, so it is fresh. Bid 3,927.98, ask 3,928.15, spread 17 points.

A snapshot, not a live price: "source": "live" describes the feed at the moment it was captured. Your requests return the current quote.

02 — Bot builder

Your bot.
Your rules.

Pick a market, a strategy and where you build. You get a precise starter prompt that wires in TickFeed data, the signal logic, risk limits and a historical test, with paper trading by default.

01Choose a market
02Choose a strategy
03Choose your environment

Templates written by TickFeed, filled in instantly. Nothing you choose is sent to an AI.

03 — Workflow

From market data
to trading decisions.

Every cycle runs the same four stages. Most end in “no trade”, and that is the bot working as designed: it waits until its rules and risk limits agree.

  1. 01TickFeed market data

    The latest bid/ask with its freshness label, and completed candles. Read-only.

  2. 02AI agent / strategy

    Your rules, written in the prompt or your code, evaluate the completed bars.

  3. 03Signal & risk checks

    A signal must pass the freshness, spread, open-position and daily-loss checks.

  4. 04Paper trade / no trade

    The decision and its reason are logged. Nothing is sent to a market.

Decision recordIllustrative
Market
XAUUSD
Timeframe
H1, completed bars
Price
Fresh, market open
Signal
No trade
Reason
Strategy conditions not met
Risk
Within configured limits
Action
Logged. No paper trade opened.
Illustrative example of one cycle’s log entry, not live data and not a trading recommendation.

04 — Integration

Real market data.
Real possibilities.

One request returns the latest XAUUSD bid and ask with the facts an agent needs to decide whether to act: is the market open, and is this price fresh?

# pip install tickfeed
from tickfeed import TickFeed

tf = TickFeed()  # reads TICKFEED_API_KEY
quote = tf.price("XAUUSD")

if quote["market_open"] and quote["freshness"] == "fresh":
    bars = tf.candles("XAUUSD", "M15", limit=50)["candles"]
    closed = [b for b in bars if b["complete"]]  # skip the forming bar
    # ...your signal logic on closed bars...
else:
    print("No fresh price; doing nothing:", quote["freshness"])
Response · HTTP 200
{
  "symbol": "XAUUSD",
  "bid": 3927.98,
  "ask": 3928.15,
  "spread": 0.17,
  "spread_points": 17,
  "point": 0.01,
  "tick_time": "2026-11-03T21:38:25.869000Z",
  "age_seconds": 0.184,
  "freshness": "fresh",
  "market_open": true,
  "stale_after_seconds": 120,
  "source": "live"
}

Sample data: a real response captured 3 Nov 2026, 21:38 UTC. Not a live price.

  1. amarket_open and freshness

    The agent acts only on a "fresh" price while the market is open. "market_closed" and "stale" mean wait.

  2. bbid, ask and spread_points

    Simulated buys fill at the ask and sells at the bid, and a signal is skipped when the spread is unusually wide.

  3. ctick_time and age_seconds

    Every decision is logged with the exact quote time in UTC, so paper results can be checked against history later.

05 — Setup

From idea to
first signal.

  1. 01

    Create your TickFeed account.

    Sign up with your email and confirm it; your free credit is added then. Create an API key in the dashboard (it is shown once), and turn on two-factor sign-in within 14 days to keep API access.

    Create account
  2. 02

    Connect TickFeed to your environment.

    In Claude, add TickFeed as a custom connector with the URL https://api.tickfeed.pro/mcp and sign in; no key in the chat. In Python, pip install tickfeed and set TICKFEED_API_KEY.

    MCP setup Developer guide
  3. 03

    Build, backtest and paper-test.

    Run the strategy on stored candles first (gaps are reported, never filled), then paper-trade it on live prices. A first signal is a starting point, not proof that a strategy is profitable.

    Candles and history API reference

06 — Launch offer

$25 free credit.

Start building with real market data.

  • For new accounts created or confirmed by 10 Nov 2026, 23:59 UTC. Added when you confirm your email.
  • $25 (or £25 / €25): enough for 12,500 price requests, or about 1.25 million candles.
  • No card and no subscription. Only successful requests use credit.
Get started free

07 — Questions

Before you build.

Does TickFeed place trades?

No. TickFeed supplies market data: live bid/ask prices, candles, ticks and gap reports. It does not connect to trading accounts or execute orders. The starter prompts keep your bot in paper trading by default.

Which AI tools can I use?

Claude connects to TickFeed's MCP server as a custom connector, and Claude Code with one command. Any MCP client can use the same server, and Python code can call the REST API or the tickfeed package.

Can I backtest my strategy?

Yes. Candles from M1 to D1 are stored for every market, with up to 5,000 bars per request and pagination for longer ranges. Gaps are reported, never filled, so a backtest never trades on invented bars.

Will my bot make money?

Nobody can promise that, and neither do we. A signal or a good backtest is not proof that a strategy is profitable. Test with historical data, then paper-trade before risking anything.

How much does it cost?

New accounts get $25 (or £25 / €25) of free credit when they confirm their email until 10 Nov. After that, requests draw from a prepaid balance: a price is one unit and candles are one unit per 100 bars.

Market data only. TickFeed does not execute trades or give investment advice. Trading carries risk, and past performance, including backtests, does not guarantee future results.