Two very different products share the name

Orderblock only does the third. It has no execution path and makes no forecast.
ClaimWhat it usually meansWhy it fails
AI predicts the next moveA model trained on price history emitting directional callsNo published retail product has demonstrated a durable, fee-surviving edge from this. You are paying for confidence, not information.
AI trades for youAutomated execution against a black-box signalTransfers your risk decisions to something you cannot inspect, usually with no record of why a position was taken.
AI reviews what you didA model reading your own closed trades and journal entriesThis one works, because the data is yours, the claims are checkable, and being wrong is cheap.

A weekly review with a partner in the loop

  1. Close the week and let the history sync

    Connected accounts sync on a schedule; manual accounts take a statement import. Start from complete closed trades, not a partial week.

  2. Ask for the shape of the week before the total

    “Where did the loss concentrate?” is a better opening than “how did I do?”. The total is the last thing you need, not the first.

  3. Take the answer to the analytics

    Open the strategy or instrument it pointed at, check the trade count, and look at the two or three trades doing the work.

  4. Read your own entries for those trades

    This is where the cause lives. The model can see that you lost on one pair; only your note says you entered before the trigger because you were down on the day.

  5. Write one rule and a sample size

    End with a change and the number of trades after which you will judge it. Put both in the journal so next week’s review has something to check.

Why grounding matters more than model quality

An assistant with no access to your history can only produce generic trading advice, which you can get for free and which has never made anyone money. An assistant with access to your closed trades can be specific, and specific claims can be wrong in ways you can detect. That is the property you actually want.

It also sets the honest limit. Orderblock’s assistant is bounded by what your account data can support. Where the data cannot answer the question, the right response is to say so rather than to produce a plausible paragraph.

AI partner questions

Can an AI trading partner make me profitable?

No tool can. What this one can do is shorten the distance between a pattern existing in your history and you noticing it. The decisions, and the risk, remain entirely yours.

Is my trading data used to train a model?

Orderblock sends account context to OpenAI only after you consent, to answer your question. See the data-use page for the categories involved and how to raise a request with us.

What does it do when the data is thin?

It should qualify the answer with the sample size rather than generalising from a handful of trades. If an answer does not mention how many trades it is based on, ask.