AI in trading
Agents and algorithms can take real work off a trading desk, on an energy desk or at a fund. Before they write to the books, the controls have to be in place. We build both.
Where agents and algorithms work
Confirmations and settlement exceptions. An agent checks each confirmation against the booked trade and drafts the fix for approval.
Reconciliation. Positions, invoices and exchange statements compared daily. Each difference explained, not only listed.
Surveillance triage. Alerts sorted before an analyst opens them, with the orders and trades behind each.
Bidding and dispatch. Day-ahead and intraday bids for a battery or a portfolio, within limits you set.
The controls that come first
Under REMIT II, software that sets price, volume or timing is algorithmic trading, whatever the vendor calls it. The rules ask for four things.
- Notification. Your regulator and ACER are told that you trade by algorithm.
- Limits that stop erroneous orders. Thresholds stop a bad order before it is sent. A plan covers the day the system fails.
- Testing and monitoring. Fully tested before it goes live, and watched while it runs.
- Records the regulator can check. Including your strategies, limits and test results.
Agents that never place an order still write to the books. So the rules of the book go in before the agent does.
That is the job of Morpholog, our open-source AI security and compliance core. This page is a summary, not legal advice.
Replay your controls against history
A new limit is a guess until you know what it would have done.
We replay your past orders and changes under the controls you have and the ones you want. You see what each would have caught, and what good business it would have blocked. Then each limit is set from your own history.
Morpholog tests a candidate rule against history before it is enforced.
When AI writes the code
AI assistants now write code that changes algorithms and limits. Every such change has to answer two questions. Who approved it? What was it tested against?
We set up change control that keeps both answers with the change. Under REMIT II, the regulator can ask for details of your testing. That record is the answer.
An example: confirmations and settlement
Unmatched confirmations wait in a queue, and settlement breaks pile up behind them. An agent proposes a correction for each. Morpholog refuses one that would breach a limit or reopen a settled period, and says why.
Each accepted change is logged with its reason, so an auditor's question is a query. An illustration of the work, not a client case.
How we work
With your team. Alongside your AI, security and trading technology people. The controls go in first, then the first agent workflow. Your team owns both when we leave.
Through your consultancy. You hold the client. We bring the trading knowledge and the compliance core.