Methodology & Sources

How an analysis is produced, verified, and what its levels mean.

Every analysis TradeRead AI publishes on a financial instrument carries a disclosure line pointing here. This page is that reference: who produces the analysis, from which data, what each published level means, over what horizon it is judged, and how the outcome is checked afterwards.

1. Who produces these analyses

Analyses are produced by Rodigital UG (haftungsbeschränkt), Herdecke, Germany — see the Impressum for the full registered details.

They are automated and non-personalised. No analysis is prepared for a particular person, none takes anyone's circumstances, holdings, or objectives into account, and none is investment advice. Rodigital UG is not supervised by BaFin or any other financial authority and provides no investment advice, no order execution, and no portfolio management.

2. How a read is produced

An analysis is generated from a chart image of one instrument at one stated timeframe. An AI model reads the price structure directly from that image — swing highs and lows, ranges, moving-average relationships — and is given, as text alongside it:

the higher-timeframe trend, computed from real closed candles (daily for an intraday read, weekly for a daily read) rather than inferred from the picture;
recent news headlines for the instrument, where available;
for crypto, derivatives context — funding rate, open interest, and liquidation clusters;
its own previous published reads of the same instrument, with the verified outcome where one exists.

The model then outputs a structured scenario, or states that there is no setup. Levels are read from the chart and from the supplied data; they are not price predictions and no part of the process forecasts a price.

3. Data sources

US equities and ETFs: Alpaca, consolidated SIP feed — quotes, trades, and daily/intraday bars.
Crypto: Binance spot market data (klines and tickers), with Alpaca as a fallback when a symbol or the feed is unavailable.
Derivatives context: Binance and Bybit — funding, open interest, liquidation levels.
News headlines: Tavily.
Crypto Fear & Greed Index: alternative.me.

Outcome verification always uses exchange price data, never a model's own recollection of what happened.

4. What each published level means

Confirmation — the price condition that would activate the scenario. It requires a candle close beyond the level, not a wick through it. Until that close happens, the scenario is not active.
Invalidation — the level that negates the scenario. It is fixed when the analysis is published and is never widened afterwards.
Target 1 / Target 2 — structural objectives derived from levels visible on the chart, not price forecasts.
Risk/Reward — the distance from confirmation to Target 1 divided by the distance from confirmation to invalidation. Target 1 must sit at least 1.2× as far as the invalidation. Below that the setup does not exist and the honest output is NO SETUP.
Confidence — the analysis's own stated conviction, LOW, MEDIUM or HIGH. A call against the higher-timeframe trend never carries HIGH.

NO SETUP is a valid and frequent output: it means no position is described at all. CONTINUATION means an earlier published setup is still running and nothing new is being published — it is not a second call on the same move.

5. Time horizon

Each analysis is judged over a fixed window tied to its timeframe — a trigger window in which the confirmation must be reached, and then a resolution window in which a target or the invalidation must arrive:

1-hour reads: 5 days to trigger, then 10 days to resolve.
4-hour reads: 7 days to trigger, then 14 days to resolve.
Daily reads: 14 days to trigger, then 28 days to resolve.

For stocks and ETFs these windows are extended by half, because those markets trade roughly five days in seven and a daily swing needs longer to play out. An analysis that never reaches its confirmation inside the trigger window expires and is recorded as such.

6. Fact and opinion are not the same thing

Measured facts: every price, close, high and low; the levels read from the chart; the higher-timeframe trend; funding, open interest and liquidation data; and every verified outcome. All of these come from exchange data or from arithmetic on it.

Opinion and estimate: the direction chosen, which scenario is preferred, the stated confidence, the reading of what the structure implies, and the wording of the thesis. These are the analysis's judgement and may be wrong.

7. How outcomes are verified

Verification is automated and rule-based. No result is entered by hand and none is removed:

A setup counts only after a candle close beyond the confirmation level.
Target and invalidation are evaluated from the next candle onward. When both are touched inside the same candle, it counts as the stop — the unfavourable reading, because the sequence inside a candle is unknown.
Setups that never trigger expire and are excluded from the hit rate, but stay visible.
Analyses published with both directions (NEUTRAL range maps) always trigger one side by construction, so they are tallied separately and kept out of the headline rate.
NO SETUP calls are stands-aside, not calls, and are not counted at all.
When the verifier's reading of a published call is corrected, the old and new readings are both recorded with the reason in the corrections log. The analysis text itself is never edited.

The complete record — every published read in order, with its levels, its outcome, and the open and expired ones included — is on the track record.

8. Conflicts of interest

Analyses are produced automatically and no position is taken as a result of publishing one. The signals shown elsewhere on this site come from a paper-trading instance — no real money stands behind that log.
Rodigital UG and its management may separately hold positions in instruments that also appear in a published analysis, including positions opened before publication. Such positions are not taken on the basis of an analysis and do not influence what is published.
No payment is accepted from any issuer, exchange, broker or third party for covering an instrument, for the direction of an analysis, or for its publication. No remuneration of any person involved is tied to the content of a particular analysis.
Rodigital UG is not a market maker, does not execute orders, holds no client money, and has no corporate-finance or advisory relationship with any issuer mentioned.
Instruments are selected by a fixed published schedule or by the user's own request — never by payment.

9. Risk warning

Nothing published here is investment advice or a personal recommendation. Technical analysis describes what a chart has done, not what a market will do. Trading and investing carry the risk of losing money, and leveraged instruments can cause losses exceeding the amount invested. Past performance, verified or simulated, does not indicate future results. Any decision taken on the basis of an analysis is the reader's own. Full terms are in the Disclaimer and Terms.