An RSI reading of 71 tells you price went up faster than it usually does. That is all. It does not tell you who bought, how badly they wanted it, whether anybody was selling into them, or whether the buying stopped ten seconds ago. It is a summary of what already happened, printed as a number.
Order flow answers a different question: what is happening right now, and who is doing it. That is genuinely different information. It is also, in the hands of most people selling courses about it, wildly oversold.
A moving average is the mean of the last n closes. RSI, MACD, stochastics, Bollinger Bands, ATR: every one is a function whose only inputs are past prices, and sometimes past volume. None contains information that was not already in the chart.
That is not a criticism, it is a definition. An indicator is a compression: a hundred bars in, one number out, so a machine or a tired human can decide without reading the whole history. But compression cannot add information, and a function of the past necessarily lags the present. A 20-period moving average turns after price turns. Always. No setting fixes this, because it is not a bug.
Indicators describe what price did. Order flow describes what participants are doing. Neither one predicts. A serious system uses both as evidence about the current state of the market, not as a signal to be obeyed.
Order flow is the raw transaction data of an exchange, before anyone summarizes it. On CME futures you can see it properly, which is not true everywhere. In spot forex there is no central exchange, so "volume" is your broker's volume and the book is your broker's book. In futures both are real. That is a structural advantage of a centralized market, and the reason serious order flow work happens in futures. There are two raw feeds, and a handful of things people calculate from them.
The DOM is a ladder of prices with the quantity of resting limit orders at each one: bids underneath, offers above. It shows intent: people willing to transact, but only at their price, and free to cancel at any moment. It tells you where size is sitting and how thin the market gets a few ticks away.
Every executed trade, with its price, its size, and, crucially, whether it happened at the bid or at the offer. A trade at the offer means a buyer paid up to get filled now. A trade at the bid means a seller hit what was there. That distinction, between passive orders waiting and aggressive orders taking, is the foundation of everything else.
Delta is aggressive buying minus aggressive selling: contracts traded at the offer minus contracts traded at the bid. Cumulative delta is the running total across the session. If price makes new highs and cumulative delta does not, the move is being driven by something other than aggressive buyers, which is worth noticing.
Heavy aggressive buying hits the offer, thousands of contracts trade, and price does not go up. Somebody is selling into all of it and refilling. That is absorption, one of the few readings that is genuinely hard to fake, because it costs real money to be the one absorbing. Imbalance is the simpler cousin: at a given price, far more traded at the offer than at the bid, or the reverse. It is shown on a footprint chart, which prints both numbers inside each bar. Clusters of it mark prices where one side was clearly more urgent.
Now the corrective, because this is where the money gets lost. Order flow is not a window into institutional intentions. It is a very high-resolution record of the past, and the past is still the past at any resolution.
Higher resolution is not the same thing as better information. It is just more of it.
The overcorrection is just as wrong. Indicators have three properties order flow struggles with: they are objective, cheap to compute, and trivially testable over decades of data. An ATR reading gives a defensible, mechanical answer to "how wide should the stop be today." That is a real question with real money attached, and no amount of tape reading answers it better.
Indicators are good at context and bad at timing. Is this market trending or ranging? Is today's volatility twice normal? Where did the session settle relative to yesterday's range? Those are legitimate uses. Expecting one to tell you when to buy is asking a summary of the past to make a claim about the future.
| Indicators | Order flow | |
|---|---|---|
| Input | Past prices and volume | The live book and every executed trade |
| Answers | What has price been doing? | Who is transacting, and how urgently? |
| Lag | Inherent, because it is a function of the past | Minimal, but it is still the past |
| Best at | Context, regime, volatility, position sizing | Confirmation and rejection at a specific level |
| Data needed | Bars, cheap and universal | Tick and depth data, heavier and exchange-specific |
| Fails when | The regime changes faster than the lookback | Volume is thin, or the book is being gamed |
| Can be backtested? | Easily | Only with stored tick data, and carefully |
That last row is the practical divide. Testing an order flow system honestly needs tick-by-tick historical data and a simulator that respects queue position, which is why many order flow "systems" have never been tested at all. It is also why one that publishes a reproducible multi-year backtest deserves a closer look than one publishing screenshots.
No mystique. The system is awake for one window a day, when the US cash market opens and the day's volume arrives with intent. In that window it assembles a picture from a few things at once:
When it does enter, it places the stop and a target at three times the risk in the market in the same instant as the entry. No overnight positions, ever. The mechanism is written out in how it works.
Here is the calibration that keeps all of this honest: the system does not need to be right most of the time, and it is not built as though it will be. With the target at three times the risk, breakeven sits at a 25% hit rate. Reading order flow does not tell it what is about to happen. It tells it whether conditions justify a trade with a defined loss and a target three times that size. The edge is in the ratio, not in prophecy. Anyone selling order flow as the thing that reveals what the market will do next is selling an ordinary hit rate with better vocabulary.
The full backtest: January 2022 to September 2026, 1,164 trades, the worst drawdown of the period, and the live signal log that archives every new signal with its outcome.
Hypothetical performance. The Algomyth figures cited here come from a backtest over historical data, not from a live account. Simulated results are prepared with hindsight, carry no financial risk, and cannot fully reflect real execution, slippage or liquidity. Past performance, real or simulated, does not guarantee future results.
Two questions cut through most of it. What does it read, and when is it awake? A system whose entire logic is a moving average crossover on a five-minute chart of a contract trading twenty-three hours a day is fitting noise. A system claiming to read institutional order flow that cannot define a signal precisely enough to count it, and has never produced a multi-year test, is telling a story about a chart.
What you want is a defined moment, a defined read, a defined risk, and a report you can regenerate on your own machine. A low bar, and most of the field does not clear it. The rest of the checklist is in how to choose a trading bot, and the wider landscape in algorithmic trading and best trading systems.
No. The system reads it; you do not have to. Understanding roughly what it looks at is worth an afternoon, because it helps you tell a serious system from a marketing deck, but you do not need to become a tape reader. If anything, watching a live DOM while a system trades is a good way to talk yourself into interfering with it.
They answer different questions, so the comparison is malformed. Order flow gives high-resolution evidence about what is happening now at a specific price. Indicators give cheap, testable context about what the market has been doing. Most systems worth having use both, and one that refuses either on principle is sold on ideology rather than results.
Tick-level data, plus a market depth feed for depth-based logic. That means a real futures data subscription rather than delayed or bar-only data, a small monthly cost on top of the platform. Backtesting it yourself also needs stored historical tick data, which takes time to download and real disk space. That is one of the few reasons the machine's specs matter, as covered in the VPS article.
Parts of it. Resting orders can be placed and cancelled to create a false impression of size, which is spoofing and is illegal under US law, and large participants routinely hide size behind iceberg orders. Executed trades on the tape are much harder to fake, because they represent money that actually changed hands. That is why readings built on completed transactions, like delta and absorption, are treated as better evidence than the shape of the book.
Because order flow readings only mean something when there is real flow, and the 9:30 AM ET cash open is where the participation is. Outside it the same patterns appear with a fraction of the volume behind them and stop being evidence of anything. Trading fewer, better-supported moments also keeps costs small relative to each trade, which is the arithmetic in slippage and commissions.
Everything we write comes back to one idea: verify, then decide. The live terminal exists for exactly that.
Unlock Algomyth Live