Market Data API: Free Data Shows Only Part of the Market
A free market data API shows only part of the market. Interactive Brokers’ free US stock stream comes from Cboe One (Cboe’s four stock exchanges) and IEX only, so the best price can sit on an exchange your strategy never sees.
That is absolutely fine to learn and test with. Before real money goes in, check what your feed covers.
- What is a market data API?
- Level 1, Level 2 or Level 3 market data: which does your strategy need?
- How is a direct feed different from the consolidated feed or a vendor API?
- How clean does market data need to be to trade on?
- Why do futures, options and FX data need extra handling?
- What happens when your market data connection drops?
- What should you ask a manager or platform about their market data?
- Questions traders ask about market data
What is a market data API?
A market data API is a service that sends market prices to your software, either as a live stream or on request. It carries quotes (the best prices to buy and sell), trades, the orders waiting at each price, and history, from an exchange or a data vendor.
Use a stream for live prices and a request API for history, and check what the stream holds: Interactive Brokers’ API documentation says its standard stream “is not tick-by-tick but consists of aggregate snapshots taken several times per second”. A strategy that needs every trade must ask for tick data separately.
The code that turns a vendor’s or broker’s format into your platform’s own is called an adapter, and each one can break when the vendor changes its format. Sending orders is in our trading API integration guide.
Level 1, Level 2 or Level 3 market data: which does your strategy need?
Most strategies need only Level 1 market data: the best bid, best offer and last trade. Strategies that read the order book need Level 2, the orders waiting behind the best prices. Level 3, every single order, matters to market makers, the firms quoting both sides all day.
| Level | What you see | Who needs it |
|---|---|---|
| Level 1 | Best bid, best offer, last trade and volume | Trend, swing and most daily strategies |
| Level 2 | Several price levels on each side, with their sizes | Intraday strategies timing entries on the book |
| Level 3 | Every single order, added, changed or cancelled | Market makers and strategies tracking their place in the queue |
The names are not standard, so read the product sheet first. CME Group publishes its futures book two ways: market by price, the top 10 levels in summary, and market by order, with “individual orders for all price levels”. An exchange’s own depth feed covers that exchange only: Nasdaq TotalView shows Nasdaq’s book only; the orders waiting on NYSE or Cboe need their own feeds.
How is a direct feed different from the consolidated feed or a vendor API?
A direct feed, the consolidated feed and a vendor API differ in how much of the market you see. A direct feed shows one exchange in full detail, fastest. The consolidated feed shows the best prices across US exchanges. A vendor’s market data API repackages either, with history.
- Direct feeds suit firms that compete on speed, and come with exchange fees.
- The consolidated feed in the US, also called the tape, is run by the securities information processors (SIPs). It publishes the official national best bid and offer, a little slower than the direct feeds.
- Vendor APIs are where most private traders and small funds start. Check which exchanges they cover, whether prices are delayed, and where their servers run.
The US tape changed in November 2025, when the SIPs changed the round lot, the standard quote size, according to an SEC order of 15 January 2026. For a stock whose average closing price was above $250 at the last twice-yearly review, a quote of 40 shares now counts toward the national best price; above $1,000 it is 10 shares, above $10,000 one share.
You can still trade any size, but code that assumes the best quote is at least 100 shares needs checking.
Europe has no consolidated tape for shares yet. ESMA selected EuroCTP on 19 December 2025 to run it for shares and ETFs; as of October 2026 its authorisation is still pending.
How clean does market data need to be to trade on?
Market data is clean enough to trade on when it is complete, correctly timestamped and adjusted for splits, dividends and delisted stocks. You should be able to explain every gap, every out-of-order message and every sudden price jump in your history.
- Completeness. Exchange feeds number every message, so a gap in the numbers is a lost message. Most retail APIs hide those numbers, so ask your vendor how it detects and fills a gap.
- Timestamps. The exchange stamps the moment an event happens; your system stamps the moment the message arrives. Keep both, and keep your clock synchronised: firms trading on EU venues must stay within 100 microseconds of UTC (official world time) for high-frequency trading and one millisecond for most other trading, under MiFID II’s clock rules.
- Adjustments. A 4-for-1 split looks like a 75% crash in unadjusted data. Cheap histories also tend to drop delisted companies, so a backtest (a test of the strategy on past prices) only ever meets the winners.
Our backtesting guide starts with the data, because every decision about a strategy is made on its history.
Why do futures, options and FX data need extra handling?
Futures, options and FX data need extra handling because none of them has one simple price series. Futures contracts expire and are replaced by the next one, options come in thousands of strike prices and expiries, and spot FX has no central exchange.
- Futures. A continuous price series joins contracts together, and the roll date and adjustment method change the history. Ask which roll date and adjustment method your history uses, and trade live on the same one.
- Options. Most strike prices (the price an option lets you buy or sell at) trade rarely, so the last trade is often hours old. Use the bid and offer.
- FX. Currencies traded $9.6 trillion a day in April 2025, spot, forwards and swaps together, according to the BIS Triennial Survey, most of it directly between banks and platforms. Your spot price, for immediate delivery, is your broker’s own price, so a stop-loss order hit at one broker may never be touched at another. Set stops on your own broker’s prices.
What happens when your market data connection drops?
When a market data connection drops, your strategy keeps seeing the last price it received, and that price looks perfectly normal. Unless the code checks the age of every price, it can trade on a market that has already moved. A stale-data check that stops new orders prevents this.
One timeout for every market does not work: a liquid futures contract updates many times a second, while a small stock may not trade for minutes. Set the limit per instrument and time of day, and when it is breached, stop new orders first.
- At 11:48 pm Pacific time on 19 October 2025, a fault in the automation that manages Amazon Web Services’ internet addresses (DNS) made its DynamoDB database service in Northern Virginia unreachable. DynamoDB was back at 2:40 am, but other AWS services needed until the afternoon of 20 October to recover.
- A backup data source in that same cloud region would have failed at the same moment.
A real backup is a second source from another provider, on another network, tested on purpose. What a slow or stale price costs in the end shows up as slippage.
What should you ask a manager or platform about their market data?
Ask a manager or platform four things about their market data API and feeds: the source for each market, how fast they detect a stale or missing price, what happens when the main source fails, and whether the backtest used the same data as live trading.
- On sources, a good answer names the feed and the backup for each market. If you hear “premium data providers”, ask for the names.
- On stale prices, it gives the seconds, the action and who is alerted.
- On failover, it gives the date of the last switch and what changed afterwards. “It has never happened” from a system live for more than two years is a red flag.
- On research data, it says where the history came from. A backtest on one vendor’s data and live trading on another’s are two different strategies.
More checks are in our trading platform due diligence guide and on trading infrastructure.
Questions traders ask about market data
A free market data API is good enough to learn and to test ideas on strategies that trade a few times a day, if you check what it covers. It is often delayed or limited to a few exchanges, which rules it out for order-book or fast intraday strategies.
At a broker, real-time market data costs a private trader from under $2 to under $20 a month per exchange package. As of October 2026, Interactive Brokers charges non-professionals $1.55 a month for CME futures Level 1 and $16.50 for Nasdaq TotalView. Professionals pay $90 for the same TotalView.
A stale feed is a data connection that has stopped updating while still showing its last price. You catch one by checking the age of every price against what is normal for that instrument and hour, and by stopping new orders the moment the limit is passed.
Timestamps differ because the exchange stamps the moment an event happens, while your system stamps the moment the message arrives, after networks, vendors and your own code. Store both and watch the gap: a growing gap is often the first sign of trouble.
The strategies worth featuring can name every data source and its backup
Very few strategy teams can show what their system did the last time a price stopped updating. For every strategy in our selection of reviewed algo and quant strategies, we ask which source feeds each market, how a stale price is caught, and when the backup last took over. We also check that the backtest ran on the same data the strategy trades on today. Vendor failover is one of the five operations criteria in our scoring methodology.
Get in touch →