How CME Market Data Fees Actually Work: Non-Display, Redistribution, and the Extra Line Items on Your Bill

Every trader who has compared data plans across two futures platforms has run into the same puzzle: the exact same CME futures data, the same E-mini S&P contract, the same Treasury futures book, costs a different amount depending on which vendor or platform you buy it through. Most explanations for this stop at "vendors mark things up." That's true as far as it goes, but it skips the actual licensing mechanics that make a real CME market data bill look the way it does, and it skips something that matters more if you're running any kind of automation, including a trade copier: CME licenses data differently depending on whether a human is looking at it or a machine is reading it, and that split shows up as a real, separate line item that a simple charting subscription doesn't necessarily cover.

What's the Difference Between Display Use and Non-Display Use?

CME Group licenses its real-time and delayed futures data under something called an Information License Agreement, or ILA. Buried inside that framework is a distinction that has nothing to do with which exchange product you're watching and everything to do with what happens to the data once it leaves the wire: is a human looking at it, or is software acting on it. CME's own market data licensing overview describes the ILA as the umbrella agreement under which individual use cases get their own schedule, and the first fork in that road is Display Use versus Non-Display Use.

Display Use is the case everyone already understands intuitively: a person opens a chart or a DOM, quotes update on screen, they read the number and make a decision. That's the world covered by the familiar professional and non-professional per-exchange fee tiers, which our Level 1 vs Level 2 market data breakdown already covers in detail. We're not repeating those tiers here.

Non-Display Use is the other fork, and it's the one that catches automation builders off guard. CME's non-display licensing FAQ defines it, in effect, as any non-viewable use of the information: feeding it into a system, process, program, machine, or calculation rather than putting it on a screen for a person to read. That single distinction is why a trade copier, a risk engine, a backtester, or a custom sizing script can trigger a different licensing conversation than the chart sitting next to it, even when both are pulling from the same underlying feed.

What Actually Counts as Non-Display Use?

CME's guidance lists specific categories rather than leaving it abstract, and it's worth reading closely because several of them describe things funded traders already do without thinking of them as licensing events. The published examples include profit and loss calculation, portfolio valuation and management, order processing, risk management, trade internalization, research and analysis, and, most relevant here, use within automated, semi-automated, programmatic, or electronic trading systems.

Notice what isn't the deciding factor: whether a human ever sees an output. A trader still watches an account balance update, still gets a fill notification, still glances at a dashboard. The question CME's framework asks isn't whether a person sees anything, it's whether the raw market data itself, the quotes and prints coming off the exchange, is being read and acted on by code before, or instead of, being displayed to a person as a live quote. That's a subtler line than it sounds.

A trader watching an account balance update is not the same thing, under CME's licensing framework, as a trader watching a live quote tick.

Concretely, this covers things like a script that pulls tick data to size a position algorithmically, a risk manager that flattens positions automatically off live price moves, or a copier that reads a source account's activity and the underlying instrument's pricing to replicate trades onto follower accounts. Whether any specific implementation of any of those actually requires a separate non-display license isn't something this article, or any article, can determine in the abstract. It depends on exactly what the licensing agreement in place for that particular data feed says, what data is actually being consumed programmatically versus only displayed, and how the vendor supplying the feed has structured its own sublicense. That's a conversation to have directly with your data vendor or CME Group, not an assumption to make either way.

How Does CME's Non-Display Tier Structure Work?

Once something qualifies as non-display use, CME doesn't price it per device the way display fees work. It prices by application count. Per CME's published structure, there's a Basic tier for a single application consuming the feed, a Premium tier once you're running two to three applications, and an Enterprise tier at four or more applications, which also removes the requirement to report each application individually. That's a genuinely different axis from the per-device display fee most traders are used to thinking about, and it stacks on top of it rather than replacing it.

Confirm the current tier thresholds, the definition of "application," and the actual fees directly with CME Group or your data vendor before relying on this structure for planning purposes. CME's own guidance also assesses these tiers per licensed use category and typically per exchange, so a real invoice can show several category-specific application counts side by side rather than one flat number. The count that follows is simplified to show the mechanic, not a literal invoice line. CME has changed non-display and delayed-data terms before, including a 2025 consolidation of non-display licenses onto this application-count fee model and a reclassification of previously free end-of-day data as billable delayed data, so treat this as the current published shape rather than a permanent one. Check CME's own Market Data Policy Education Center for whatever is current when you're reading this.

Here's where the counting gets concrete for a funded trader running any kind of automation, and here's the detail that catches people out: a charting platform used only to watch the market on a screen is Display Use and doesn't add to this count at all, that's the separate per-device fee covered above. What counts toward the Non-Display tier is anything reading the feed programmatically without a person watching each tick before it acts. Run an automated trade copier that reads the feed to size positions, that's one Non-Display Application. Add a separate risk monitor that independently reads the same feed to flatten positions on its own logic, and that's a second Non-Display Application. One copier plus one risk monitor equals two applications total, and two applications is enough to cross the line from the single-application Basic tier into the two-to-three-application Premium tier, before a single dollar figure even enters the picture. Add a third tool, a backtester pulling the same feed for research, and you're still in Premium; a fourth pushes you into Enterprise.

Why Does the Same CME Data Cost Different Amounts by Vendor?

This is the part that makes "just compare data prices" bad advice. CME Group doesn't sell most of its real-time data straight to retail end users. It licenses the raw feed out to a roster of independent distributors, each operating under its own agreement, through its Licensed Market Data Distributors program, and each of those distributors negotiates its own redistribution deal and prices its own downstream packaging. A data vendor, a broker, and a charting platform might all be buying the same underlying CME feed at the source, but they're not reselling it under identical terms, and each layers its own infrastructure, support, and margin on top before it reaches you.

That's a structural fact, not a coincidence and not an arbitrary markup someone invented to confuse you. It's the direct, predictable consequence of a licensing model where the exchange sets terms at the wholesale level and lets each redistributor set its own retail packaging. It's also exactly why two platforms built on different underlying distribution paths, say one running on Rithmic's infrastructure and another on dxFeed's, can differ in price and bundling even when both ultimately source the same CME instruments; our dxFeed vs Rithmic comparison and our breakdown of what a Rithmic feed actually is go into how those two specific distribution paths differ operationally. This article is about why the billing itself can differ even before you get to those operational differences. Confirm current packaging and pricing directly with each vendor, since it varies by distributor and can change without notice.

What Are the Actual Layers in a Real Futures Data Bill?

Put the display/non-display distinction and the redistribution structure together and a realistic all-in data bill has up to three separate layers stacked on top of each other, and the number printed on a vendor's pricing page is often only one or two of them.

The figures below are round, illustrative placeholders built only to show the shape of a bill. They are not real current CME, vendor, or platform prices; check actual numbers with your vendor or CME Group before relying on any of them.

Add layers 1 and 2 and you get $6 + $24 = $30 per month, which is typically the number a trader assumes is the whole bill, because it's usually the figure printed on the vendor's pricing page as "market data." Add layer 3, the non-display fee that applies once a copier or sizing engine is also reading that feed, and the real total is $6 + $24 + $50 = $80 per month, more than two and a half times the advertised number. Nobody lied on the pricing page. The pricing page was describing display use, and the automated system quietly moved part of the bill into a different license entirely.

This is also why comparing two platforms by their headline "data package" price alone can point you the wrong way. Suppose Vendor A advertises CME access at $45 per month, and that number covers layers 1 and 2 only. The non-display application fee for running a copier on top is billed separately at $75 per month, for an all-in cost of $45 + $75 = $120 per month. Vendor B advertises CME access at $100 per month, a headline number that's $100 minus $45, or $55 per month higher than Vendor A's on its face, but Vendor B's $100 already bundles the non-display layer into the package, so its all-in cost stays at $100 per month. Once you actually run the automation, Vendor B ends up $120 minus $100, or $20 per month cheaper overall, purely because of which layers each vendor chose to fold into its headline price.

Does a Trade Copier Need a Non-Display License?

For anyone running automated copying, sizing, or execution logic on top of a data feed, this is the question that actually matters, and the honest answer is: it depends, and only your vendor or CME Group can tell you for certain in your specific case.

Here's why it isn't a yes or no you can answer from a blog post. CME's own list of non-display examples explicitly includes automated, semi-automated, programmatic, and electronic trading systems, so a system that reads market data programmatically to make a trading or sizing decision sits squarely inside the kind of use non-display licensing is written to cover. But whether your particular copier setup triggers a separate license depends on things that vary case by case: what data it's actually reading, raw exchange ticks versus, for instance, fills and account state already sitting inside a broker's own infrastructure, whose agreement that data is licensed under, and how that specific vendor has structured its non-display sublicensing, if it has addressed it at all. Some platforms fold non-display rights into their existing package for every subscriber. Others bill it separately, only once you request algorithmic or API access. Some may not have addressed it clearly in their public terms, which is its own risk to be aware of.

This isn't a compliance verdict for any individual setup, and it isn't a reason to worry about every trade copier on the market. It's a reason to ask the question explicitly rather than assume the answer either way, especially if you or your team built custom automation on top of a raw feed rather than using a platform's own built-in, already-licensed copying feature.

How Should You Compare Data Costs Across Platforms?

Given all that, comparing "data cost" between two prop trading platforms or two copier setups by looking at the number on the pricing page alone is close to meaningless. A shorter, more useful list of questions to put directly to each vendor:

• Does the advertised price include CME's exchange-level display fee only, or does it already include redistribution and infrastructure costs?
• If I run any automated system on this feed, a copier, a sizing script, a backtester, is there a separate non-display or application fee, and is it billed per application?
• How many "applications" does adding a copier count as under this vendor's counting rules if I'm already running other automated tools, like a sizing script or a risk monitor, on the same feed, and does that cross a tier threshold?
• Is the redistribution and infrastructure markup fixed, or does it change as I add accounts, instruments, or automated consumers of the feed?

Ask those four questions of two vendors and you'll often find the "cheaper" one on the pricing page isn't cheaper once you're running the same automation on both, exactly like the Vendor A and Vendor B example above. The headline number was never dishonest, it just wasn't answering the question you actually needed answered.

What This Layered Structure Won't Do For You

Be clear about what this article actually buys you. Knowing that a bill has three potential layers doesn't make any of those layers cheaper. CME sets its own fees, redistributors set theirs, and reading this won't move either number. What it does do is stop you from being blindsided by a bill that's suddenly two or three times the advertised "data package" price the month you turn on a copier or a sizing script, and it lets you compare platforms on the actual all-in number instead of whichever partial number happens to be printed biggest on the pricing page.

It's also worth being honest about who this doesn't apply to. A funded trader running a single prop firm's own platform, watching charts, and using that platform's built-in, already-licensed copy or replication feature, is very likely not the one negotiating any of this directly. The firm or platform has typically already handled licensing for the package it sells, display and non-display alike. This distinction becomes practically relevant mainly when you or a vendor you're using layers custom automation, a self-hosted script, an independent copier, an external risk tool, on top of a raw feed whose license was written with only display use in mind. If that isn't your setup, the honest answer is that this whole question may simply not arise for you, and adding complexity to chase a licensing distinction that doesn't apply to your situation is its own kind of mistake.

And one more honest tradeoff worth sitting with: if your real goal is controlling data cost, the fix is sometimes to reduce the number of separate applications touching the raw feed rather than to add another one. Running a commercial copier alongside a second, home-grown script that independently pulls the same feed for a slightly different purpose can itself push you from Basic into Premium under CME's application-count model, adding cost that has nothing to do with which vendor you chose and everything to do with how many separate tools are running against the same data.

Frequently asked questions