Methodology
What each reading claims, and where it breaks
These pages explain a method, not a product. Every reading traders take off order flow is a threshold rule applied to two numbers, and the useful description of it is the rule with its threshold written down — a cell at 2.5× the mean volume of its bar with the two sides within 35% of each other, not "absorption detection". Each page states the mechanism, then the specific conditions under which the reading stops being evidence, and only after that how it is implemented here. TradersWeek runs 28 order-flow layers on Binance spot data in a browser; the chart and 123 indicators are free, order flow is Premium at $120 per month.
The readings
| Page | What it answers | Where the reading fails |
|---|---|---|
| Footprint charts and bid × ask | How a cell is built from the trade tape, what delta measures, and the thresholds behind stacked imbalance, absorption, exhaustion and iceberg detection. | Delta is an attribution rather than a measurement. Absorption is indistinguishable in the tape from a queue that emptied and refilled, and an iceberg marker is an inference from repeated hits. |
| Cumulative delta and divergence | What CVD accumulates, why its absolute level carries no meaning, and what a divergence against price actually claims. | It is noise when read directionally, when the venue is not where the aggression is, and when delta is estimated from the candle instead of counted from trades. |
| Automatic Elliott Wave counts | How a labeling engine gets from swings to a count: candidate structures, elimination by rule, and ranking of what survives. | The method is contested and has no established statistical validation in the academic literature. Alternate counts survive the rules routinely, and ranking is not evidence. |
| Why market data is a second bill | Why exchange feeds are priced apart from the platform licence, what four platforms publish for data, and how to total a real monthly cost. | TradersWeek has no data bill because there are no CME, Eurex or US equity feeds here to bill for. On that question it is not the answer. |
| How to read a footprint chart | A context-first workflow for reading executed volume: location first, then the current bar, confirmation, and the condition that invalidates the idea. | A footprint records executed trades at one venue. It does not show the full market, resting liquidity away from the trade, or what will happen after a level is tested. |
| Crypto order-flow data | Which exchange data produces a footprint, CVD and DOM, what is reconstructed, and why a result from one crypto venue is not a market-wide claim. | Binance spot prints describe Binance spot only. They do not include perpetual futures positioning, funding, liquidations, or liquidity resting on another exchange. |
| Why order-flow signals fail | The common ways footprint and CVD readings become false confidence: one-venue data, cell settings, late confirmation and treating an inference as proof. | A failure list can reduce overconfidence, but it cannot turn order flow into a predictive system or replace position sizing and a predefined exit. |
| Order-book heatmap vs volume heatmap | The difference between displayed liquidity in an order book and volume already executed at price, including the questions each chart cannot answer. | A live book can be cancelled before it trades, while executed volume cannot reveal orders that never traded. Neither view establishes intent. |
| Volume profile vs footprint | How volume profile identifies acceptance over a range while a footprint describes the trade-by-trade auction inside a bar. | Profile and footprint can point at the same level without being confirmation of a trade. They answer different questions about the same auction. |
| Order-flow trade invalidation | How to turn an order-flow observation into a conditional plan with a trigger, an invalidation level and a bounded risk decision. | Invalidation controls loss if an idea is wrong; it does not validate the premise or predict the size of a later move. |
Why the limits are on the page
A detector section that lists what a signal catches and stops there is indistinguishable from an advertisement, and it is also the version that costs the reader money. The third column above is the part of each page that cannot be written in our favour: absorption genuinely is indistinguishable in the trade tape from a queue that emptied and refilled, exhaustion genuinely fires without waiting for the next bar, and an Elliott count genuinely rests on a framework with no established statistical validation.
Thresholds are published rather than implied, because a threshold is what makes a claim checkable: 3× for an imbalanced cell, 3 consecutive levels for a stack, 10 hits inside 5 seconds with 80% one-sided volume for an iceberg. All of them are settings, and the defaults are the numbers quoted on the pages. Every figure about TradersWeek comes from a single registry in the codebase, and a test fails the build when a page drifts away from it.
Where this leads
- Comparing platforms rather than methods — the comparison pages carry a source link and a check date on every statement about another product.
- Working out whether a scenario fits — crypto futures, forex scalping and leaving TradingView each state what they do not close.
Thresholds argue better on a live chart than on a page, and the free core opens without an account.
Open terminalLast reviewed 12 October 2026