Choosing software & data
Learning objectives
- Name the three jobs in a futures trader's stack (execution, charting, data) and which tool does each one on this desk.
- Explain why a footprint needs tick data with the side of every trade, and why 1-minute bars cannot rebuild it.
- Check whether your platform's delta comes from a recorded aggressor side or from a guess, and say why small deltas can flip with the method.
- State what the evidence says about building volume profiles from 1-minute bars, with its grade.
It is 10:23 ET (21:23 ICT) on Monday 21 September 2026, during RTH (phiên chính, 09:30–16:00 ET). In one minute NQ climbs from 30,497.25 to 30,502.75 and 1,576 contracts change hands. The footprint built from our recorded data says sellers were the more aggressive side: delta −112. Rebuild the same minute with a common shortcut that guesses each trade’s side from the price ticks, and delta reads +102.
Same trades, opposite answers. Before you pay for software or trust an order-flow number, know which of the two your screen shows, and why it matters more for small numbers than big ones.
Key terms
Trading stack (bộ công cụ giao dịch): the tools that send your orders (execution), draw your charts (charting) and deliver the market’s prints (data) Definition. One program can do all three, or you can split them.
Tick data (dữ liệu tick): every trade with its time, price, size and, on a good feed, the aggressor side (bên chủ động): whether the buyer or the seller crossed the spread to make the trade Definition. CME’s own feed carries an aggressor-side field on each trade.
Tick rule (quy tắc tick): a shortcut that guesses the side when it is missing: a trade above the last different price counts as a buy, below it as a sell Definition (Lee and Ready, 1991).
Terms you need: Delta is the volume traded at the ask (aggressive buyers) minus the volume traded at the bid (aggressive sellers), in one bar or at one price (M3.04). An OHLC bar keeps only four prices—Open, High, Low and Close—plus total volume, while a footprint opens a bar to show bid and ask volume at every price inside it (M0.09). A bid/ask tape is the scrolling list of individual trades coloured by side, and the CME non-professional declaration is the form stating you trade for yourself rather than as a financial professional so you qualify for lower exchange data fees.
Chart first: one minute, labelled two ways
One minute of NQ, two ways to label buyers and sellers
DefinitionHow it works
Three jobs, three tools on this desk
On this desk the jobs are split on purpose: ATAS X (with the academy’s OrderflowSuite studies) is the reading screen on its own tick feed, and a funded Topstep account is traded in TopstepX, the firm’s platform (a house choice, House rule). Vexity adds options context; its index data is live only from 09:30 to 16:15 ET and frozen outside that window (PLAT1.02).
You do not need this exact stack. You need to know, for every number on your screen, which tool produced it from what data. Pick one charting platform and learn it deeply.
What order flow needs: every trade, with its side
A 1-minute bar keeps four prices and one volume number. Which prices traded how much, and who was aggressive at each, is gone Definition. Footprints, delta and the tape all start from tick data with the side of each trade.
So the first question about any data source is: does it deliver every trade with its side, in real time? Free charting sites, delayed quotes and bar downloads do not. They are fine for studying history, and useless for reading order flow live. Depth of book (resting orders) is a separate product again; MKT1.07 covers it.
Recorded side vs guessed side
When the feed carries the aggressor side, your platform adds it up. When it does not, software guesses, usually with the tick rule. The guess fails in a predictable place: when many trades print at one price, the tick rule keeps labelling them with the direction of the last price change, even if sellers took over there.
That is what happened at 10:23. Price drifted up, so the tick rule called most trades at unchanged prices buys, while the recorded side shows sellers hitting the bid on the lower rows. The methods disagree most when buying and selling are nearly balanced: exactly the minutes where a small delta tempts you to tell a story.
Myth / hypothesis: “An up bar with negative delta is hidden selling that is about to push price down.” This is the absorption story, graded Lore on NQ: in a public practitioner study, bars with absorption reversed no more often than bars without it LoreOne trader's study (M3.02). The 10:23 bar adds a second problem: with a different labelling method, the “divergence” disappears.
What the data says
In practice: if you read footprints or delta, pay for a feed that records the side, and treat a delta of a few dozen contracts on an NQ minute as within reach of a method change.
Worked example: one rule, two platforms
Suppose you test an illustrative, untested long rule on MNQ on two platforms: “buy the close of an up bar whose delta is positive; stop one tick below the bar’s low.” At 10:23 on 21 Sep 2026 the bar closes up at 30,502.75.
- Platform A (recorded side): delta −112. No trade.
- Platform B (tick rule): delta +102. Buy 1 MNQ near 30,503.00, stop at 30,491.50 (one tick below the 30,491.75 low).
Risk on platform B’s trade: 30,503.00 − 30,491.50 = 11.50 points = 46 ticks. At $0.50 a tick on MNQ that is $23.00 per contract, before commissions and slippage (M0.05 has the tick values).
The rule is word for word the same, and one platform trades while the other does not. Over a month you get two trade lists and two “backtests” of one idea. Fix the data definition before you test anything.
The mirror and the failure
Short side. The mirror rule, “sell a down bar with negative delta”, has the same weakness: on a down-drifting minute the tick rule can show negative delta where the recorded side shows buyers lifting the offer.
Where the check fails. Two methods disagreeing does not tell you which is right. Whether our recorded side matches CME’s flag on every trade is not yet measured, and feeds also differ in whether they merge the fills of one aggressive order into a single print (MKT1.07, M3.13).
Do it on the platform
Prop-firm note. On this desk you place orders in TopstepX on the firm’s data and read order flow in ATAS X on your own feed. Two feeds can differ in timing and print grouping, so the firm’s screen is the record for your fills and loss limits; simulated evaluation fills are not live fills (MKT2.03). PLAT2.08 sets up the two-screen workflow.
Common mistakes
Common mistake: Reading meaning into a small delta. On our NQ sample about 1 one-minute bar in 8 changed sign when only the labelling method changed, almost all with small delta. A delta of −40 on a 1,500-contract minute is not evidence until it survives that check.
Common mistake: Buying a delayed or bar-only feed and “reading order flow” off it, or trusting a tool that sells signals without saying what it computes from which data. Every number on your screen should answer “computed from what?”
Drill: audit your own delta
On replay or sim, check five NQ or MNQ 1-minute bars.
- Write down each bar’s delta from the delta study.
- Recompute it from the footprint: ask column total minus bid column total.
- Find out from the platform’s documentation whether the side is recorded from the feed or inferred.
- Mark any bar whose delta is under 50 contracts.
Solution
Steps 1 and 2 should match exactly; if they do not, the study uses a different session, filter or bar setting than the footprint, and you have found a settings problem to fix before anything else.
Step 3 is the real answer to “where does my delta come from”. If it is inferred, treat every small delta as uncertain: on our recording the tick rule flipped the sign on about 1 bar in 8.
The bars from step 4 are the ones to read least into, whatever the method.
What we don’t know yet
- Whether the side our ATAS feed records equals CME’s aggressor flag on every trade: not yet measured.
- Our lab’s own check that 1-minute profiles match tick-built profiles on NQ: not yet measured.
Recap
- A stack has three jobs: execution, charting and data. Know which tool produced every number you look at.
- Footprints and delta need tick data with the side of each trade; 1-minute bars cannot be opened back into trades.
- Delta from a recorded side and delta guessed by the tick rule differ: on our NQ sample about 14% of contracts were labelled differently and about 1 bar in 8 changed sign, mostly bars with small delta.
- Profiles built from 1-minute bars are close to tick-built ones in one practitioner’s study (Plausible, one trader’s study); footprints and delta have no such shortcut.
Further reading
- Lee and Ready (1991), “Inferring Trade Direction from Intraday Data”, Journal of Finance 46(2). Where the tick rule and the quote rule come from.
- Ellis, Michaely and O’Hara (2000), “The Accuracy of Trade Classification Rules: Evidence from Nasdaq”, Journal of Financial and Quantitative Analysis 35(4). How often the classic rules disagree with the true side.
- Order Flow, Measured: All claims. A public, pre-registered practitioner study on NQ; read the measurement rows (ME-01, ME-04, ME-05).
- CME Group education: cmegroup.com/education. Check the current market-data fees there before you subscribe.
- Our lab: the 58-session NQ comparison of recorded side and tick rule in the second evidence box above.
Prerequisites: M0.05 NQ, ES, MNQ, MES contracts & margin; M0.06 Four questions before every trade · Next: M0.08 Basis & fair value, why the ES price on your screen is not the index price; then PLAT2.01 sets up the platform.
Educational content only. Not financial advice. Futures trading involves substantial risk of loss. Not affiliated with Topstep; firm rules change; check the Topstep help center.
Saved in this browser (account sync comes at launch).