Sierra Chart for Futures Prop Trading: Why Serious Order-Flow Traders Keep Choosing It

Sierra Chart is a Windows desktop platform for futures charting, order-flow analysis, and execution. Its reputation among tape readers rests on two engineering decisions you never see on screen: custom studies compile to native C++ machine code, and intraday data lands on disk as a flat file of fixed 40-byte records, one per trade. Almost everything traders praise about the platform, and almost everything they curse, follows from those two facts.

What is Sierra Chart, and why does it look like that?

Sierra Chart is a 64-bit Windows application, and the utilitarian interface marks where the engineering effort did not go. Menus nest deeply, dialogs are dense, and configuration lives in chartbooks and per-chart settings rather than a guided setup flow. No onboarding wizard hands you a working footprint chart on your first afternoon.

The platform boundary is firm. Sierra Chart's documentation states that only 64-bit operating systems are supported, and there is no native GNU/Linux or macOS build. Traders do run it under Wine or CrossOver, and Sierra Chart documents that setup to the point of calling one network IO performance fix mandatory for Linux and Wine users, but a documented workaround is not a supported platform. The company says it has long intended to add other operating systems, Linux first, without publishing a timeline. Mobile access follows from the same fact: this is a Windows desktop program, so phone or tablet use in practice means remote-desktopping into a Windows machine.

Budget a week before you are productive. It goes on finding where settings live, how chartbooks and symbol settings interact, and how a chart's data record size changes what studies can compute. Traders who need output on day one bounce off it.

How does Sierra Chart store tick data?

In a flat binary file with no database engine in front of it. Sierra Chart's intraday files use the .scid format: a 56-byte header followed by fixed 40-byte records. Each record carries open, high, low, and close as 4-byte floats, plus number of trades, total volume, bid volume, and ask volume as 4-byte unsigned integers. The timestamp is an 8-byte value counting microseconds from the SCDateTime epoch of December 30, 1899, always in UTC.

The convention that matters for order flow hides inside those OHLC fields. When Sierra Chart stores individual trades, each tick is its own record. For a single-trade record carrying bid and ask context, Open is set to zero (the SINGLE_TRADE_WITH_BID_ASK constant), High carries the ask price, Low carries the bid price, and Close carries the trade price. That packing is how trade-at-bid versus trade-at-ask classification survives to disk, and it is exactly what a footprint study reads back later.

Fixed width has a consequence worth doing the arithmetic on. Ten million ticks occupy 10,000,000 x 40 = 400,000,000 bytes, plus the 56-byte header, for 400,000,056 bytes: 400 MB in decimal units, or 381.47 MiB. Now suppose a symbol records a million ticks a day, a round illustrative figure rather than a measurement of any specific contract, across 250 trading days. That is 250,000,000 records. 250,000,000 x 40 = 10,000,000,000 bytes, exactly 10 GB decimal, or 9.31 GiB.

A year of true tick history for one instrument is therefore a ten gigabyte flat file with no index, no schema, and no query planner. Record N begins at byte 56 + 40N, so seeking is arithmetic and scanning is a linear read. Sierra Chart opens these files in shared mode when it reads or writes, buffers streaming writes internally and flushes to the file every five seconds rather than on every trade, and the Days to Load setting governs how much of that history a chart pulls when it opens.

Sierra Chart's reputation is not a UI opinion. It is a consequence of a 40-byte record and a C++ compiler.

Why are custom studies written in C++ instead of a scripting language?

Because Sierra Chart chose compiled native code over an interpreter and says so directly. ACSIL, the Advanced Custom Study Interface and Language, is native C++, and Sierra Chart's ACSIL documentation argues that the native code ACSIL uses "is simply faster" than C# and .NET, which it calls "slower and more bulky" and Microsoft proprietary. The page cites "actual real-life tests" without publishing any, so treat the claim structurally rather than numerically: compiled code, no interpreter, no managed runtime, and no per-tick object allocation between your loop and a fixed-width record.

The build path is less forbidding than "write C++" sounds. Three options exist. Remote Build, the recommended one, compiles your source on Sierra Chart's servers with the MinGW GCC C++ compiler, so you need no local toolchain, though custom #include files of your own are ignored in that mode. Local Build (Release) compiles with Visual C++ on your machine. Local Build (Debug) emits a DLL plus a PDB you can step through in Visual Studio. Toolchain details move over time, so check the current page before setting up.

The cost is skill, not setup. You manage arrays, indices, and persistent variables yourself, and an off-by-one in a lookback loop silently produces a plausible wrong line instead of an error. If your mental model of custom indicators comes from C# platforms, the contrast in our NinjaTrader vs Quantower breakdown is the useful one: both hand you a managed language and a friendlier editor, and both put a runtime between your code and the data.

How do market data and order routing connect?

Separately, on independent connections with independent credentials. Understanding that split before you buy anything saves more money than any other single fact in this article.

Denali is Sierra Chart's own broker-neutral data service. Sierra Chart's Denali documentation lists CME, CBOT, NYMEX, COMEX, EUREX, CFE, NASDAQ TotalView including NYSE and AMEX, the US equities consolidated tape, and CBOE indexes. It states that up to 1400 levels of market depth per side are provided, and that three simultaneous connections to the feed are supported per device. An Integrated service package is required.

Rithmic sits on the other rail for many funded traders. Sierra Chart's Rithmic documentation requires Service Package 10 or higher and exposes three separate credential sets: trading username, market data username, and historical data username. Mixing rails is explicitly documented: Rithmic for order routing, Denali for market data, inside one running instance. For background on the rail itself, our explainer on Rithmic as a futures data feed covers what that connection actually does.

Picture one Sierra Chart process on a Windows box with two arrows leaving it. The left arrow is the order-routing connection (Rithmic, or Sierra Chart's own Teton service) authenticated with the trading username, carrying orders out and fills back in. The right arrow is the market-data connection (Denali, on its own separate credentials) carrying ticks and depth in. A third, optional credential set pulls historical data. Everything arriving on the right arrow is written locally into the .scid file described above. The point of that picture: your prop firm usually dictates the left arrow, and you often still choose the right one.

Two caveats. First, the Rithmic page states that Sierra Chart "does not provide any technical support for issues with Rithmic market data (whether historical or real-time)" and says the same about Rithmic trading functionality. That is one vendor publishing its own support boundary, not a judgment on Rithmic's quality, and support policies change. Second, 1400 depth levels is a ceiling, not a description of a live book. On ES, 1400 levels at 0.25 points per tick spans 350 points on one side; at the standard $50 per point multiplier that is $17,500 of price ladder per side, or 2,800 levels and 700 points across both sides, $35,000 of range. No order book is ever populated that deep intraday. The entitlement removes depth as a binding constraint rather than delivering 1400 useful rows.

Which service package do you actually need?

The package number is an entitlement gate, not a discount tier, and that trips up more new users than anything else about Sierra Chart. Base packages cannot connect to any external data or trading service at all: they support Sierra Chart provided data and trading services only, including its simulated trading service. Interactive Brokers, CQG, Rithmic, and Denali all require an Integrated package.

Those are list prices at the time of writing, taken from Sierra Chart's packages page, which itself warns that "all pricing is subject to small price changes over time without notice." Confirm current figures there before budgeting, and treat the feature-to-package mapping as current rather than permanent. The Volume Profile interactive drawing tool makes the point: it is documented as available in all packages from version 2778 onward, so a tool that was once a tier argument no longer is. There is no perpetual license either, and the company justifies subscription-only pricing by pointing to its market data services, exchange order routing, support, and the continuous updates needed to keep working with external services.

Two pieces of arithmetic make the decision concrete. The order-flow toolset most traders arrive for is the gap between Integrated Standard and Integrated Advanced: $46 minus $36 is $10 a month, or $120 a year at list. Market by Order takes you to Package 12: $56 minus $36 is $20 a month over Package 10, or $240 a year. That is the entire price of admission to what Sierra Chart is famous for, sitting on top of the base platform.

Prepaying moves the number meaningfully. Discounts run 15% for three months, 25% for six, and 35% for twelve. Package 11 at $46 a month prepaid twelve months works out to $46 x 0.65 = $29.90 a month, or $358.80 for the year, against $552.00 paying month to month. The saving is $193.20, which cross-checks directly: $552.00 x 0.35 = $193.20. Six months lands at $34.50 a month, three months at $39.10.

Exchange fees are separate and are set by exchanges, not by Sierra Chart. The Denali page states that exchange fees are billed in full each month no matter when they are activated, and that authorizing CME real-time data requires Sierra Chart to connect to a live funded futures trading account. Published fee schedules shift, so confirm current figures with Sierra Chart and the relevant exchange before building a monthly cost estimate.

Why do footprint traders end up here?

Because the footprint study reads back the same bid and ask packing the file format already preserves, at 1-tick granularity, over long histories. Numbers Bars, Sierra Chart's footprint study, works only on intraday charts and not on historical daily charts, and expects the chart's data records set to 1 Tick; with anything coarser the documentation warns the displayed data will be less accurate. Numbers Bars have no dependency on market depth data at all, only the best bid and ask price, and they require an Advanced package.

Follow one trade through the pipeline and the constraint becomes obvious. An exchange trade arrives carrying bid and ask context. Stored at 1 Tick, it becomes a single 40-byte .scid record with the ask price in the High field and the bid price in the Low field. Days to Load pulls a span of those records into the chart, the Numbers Bars study reads each one, and it can therefore split volume into bid-side and ask-side at every price level. Now change exactly one node in that chain: pre-aggregate the same trades into one-minute records before they are written. The High and Low fields now describe the minute's range instead of one trade's bid and ask. The classification is gone permanently, because it was never written to disk, and the footprint study has nothing left to split.

If footprint and depth-of-market reading are new to you, our order flow and DOM explainer covers what these charts are telling you before you pay to render them.

Can you use Sierra Chart on a funded account?

Frequently that decision is not yours. Work it as four steps, in order.

Step one: does your prop firm's current approved platform list include Sierra Chart? If not, nothing else here matters, and no amount of native-code study performance changes the answer. Approved lists change, so check yours rather than trusting a forum post from last year.

Step two: which rail does the firm require? Because trading and market data are separate connections, a firm can mandate the order-routing side and leave the data side to you. That is why many funded traders run Sierra Chart as a front end on a firm-mandated routing connection with a data feed of their own choosing.

Step three: are you on an evaluation account? Sierra Chart's Rithmic documentation states that trading evaluator accounts, naming Apex, Top Step Trader and several other firms, "do not meet the requirement as a live and funded trading account according to CME policy," and directs those users to its separately documented Easy Solution path for non-professional exchange fees. Your fee treatment therefore differs from a funded trader's. Confirm the current requirement with your firm and your data provider.

Step four: does your chosen feed carry bid volume and ask volume? If not, your footprints degrade regardless of which package you bought.

Steps one and two are decided by someone other than you, which is the plainest honest statement in this piece. One execution detail is worth knowing: the Rithmic page lists server-side OCO orders as supported and server-side bracket orders as not supported as of its writing. Server-side OCO means the stop and target cancel each other away from your machine, so the pairing survives a client disconnect. Protective orders that the platform manages locally instead depend on your machine and connection staying up, which on a funded account can be the difference between a dropped connection being an inconvenience and being a rule breach.

One task it wins, one task it loses

Task A, where Sierra Chart is genuinely hard to beat: scan a full year of 1-tick history, the roughly 10 GB, 250,000,000-record file computed earlier, with a custom study that classifies every trade at bid or at ask and accumulates a per-level imbalance. That is one linear pass over fixed-width records in compiled native code, with the bid and ask context already sitting in each record's High and Low fields. No query planner, no interpreter, no per-tick allocation. The work is close to the metal by construction, not by tuning.

Task B, where it loses badly: you had an idea this morning, you want three indicators and two conditions wired together, and you want an equity curve before lunch. In ACSIL that means writing C++, declaring your inputs, compiling to a DLL, and debugging your own array indexing. A drag-and-drop strategy builder gets a non-programmer to a first result in one sitting. ACSIL does not, and pretending otherwise costs you a week.

The asymmetry is in the number of steps and the skill prerequisite, both of which you can verify yourself. Neither task has a published timing worth trusting, including from Sierra Chart.

Who should not buy Sierra Chart

Four groups, stated plainly. Traders who want a polished interface working well on day one: the UI is utilitarian by design. Traders who need a first-class mobile client: this is a 64-bit Windows desktop application with no native macOS, Linux, or mobile build. Traders who want no-code strategy building: ACSIL is C++, and visual tooling is not the product's center of gravity. And traders whose firm mandates a different platform or a routing rail Sierra Chart is not approved on, which overrides every technical merit above.

The honest tradeoff is that Sierra Chart exchanges approachability for control and performance. It rewards traders who will invest weeks in learning it and punishes traders who need output immediately. There is a quieter waste case too: if you sit on Package 12 and never open a Market by Order or Numbers Bars chart, you are paying $240 a year at list over Package 10 for capability you are not consuming, and Package 10 draws the same basic candles. Buy the tier whose entitlements you will actually use, then re-check that judgment after a quarter of real screen time.

Frequently asked questions