Collect market data
Ay Trade is currently connected to three markets. Incoming market data is processed in real time across the relevant strategies, timeframes and lanes.
R&D - Agent & Systems Engineering
Ay Trade
Ay Trade is a trading system inside my personal agent, Aygent Jay. The system collects market data and processes it through different strategies and lanes. It looks for potential trades and follows them from the first detection through controlled execution and monitoring. Aygent Jay is based on OpenClaw and is something I am building and expanding as a long-term agent system. Ay Trade is currently the first major productive part of it.
The next major part is Ay Search, a research and scraping agent built around how I work. It collects knowledge sources, news and other information and makes them usable for me. In the long run, it can also become another source of input for the development of Ay Trade.
System flow
From market data to execution
Ay Trade deliberately separates data collection, trade discovery, lifecycle monitoring and actual execution.
Find trade candidates
Different strategies examine the processed data for possible setups. A concrete trade candidate is only created when the rules and conditions defined for a lane are met.
Promotion to live eval
A confirmed trade candidate is promoted into the live-eval system. From that point on, the setup is no longer just being searched for but is tracked as a concrete state.
One source of truth per lane
Within the live-eval system, each lane has one canonical state. This acts as the central source of truth for the rest of the lifecycle and prevents conflicting parallel states.
Trigger or invalidation
An active setup continues to be monitored. The system checks whether the required trigger occurs or whether other conditions invalidate the setup first.
Controlled execution
When a trigger is confirmed and the execution rules are satisfied, a request/ticket is created for the exchange. The trade is then executed and continues to be managed by the system.
Engineering
More than finding trades
A large part of the development is about making the overall system reliable, traceable and safe.
A major part of Ay Trade is its architecture. It is there to make sure data is processed correctly, states stay unambiguous and errors are caught early.
For that, I built different validators, guards, lifecycle rules and tests. Processing steps are checked before a setup can move into the next state or trigger a real execution.
Analysis and execution are separated. Each active lane can be granted permission to execute real orders.
Overall, the system is stable. The current focus is long-term evaluation and finding a robust edge in the market.
- Clear lane and lifecycle states
- Validators and guards between critical steps
- Controlled promotion into the live-eval system
- Separated analysis and execution paths
- Testing, debugging and reconciliation
- Long-term live evaluation
Reporting & automation
Information where I need it
Besides the trading logic itself, Aygent Jay also prepares and forwards relevant information.
Daily reports
A daily report is generated automatically from the collected system information. Its structure, information density and priorities are tailored to how I use it.
Relevant events
Important state changes, trades and other system events can be forwarded immediately without me having to watch the system all the time.
Automated workflows
Depending on the task, processing is event-based or handled through regularly scheduled cron jobs. Relevant results are then sent to a device I choose.
BTC, ETH and SOL
four groups per market
twelve long and twelve short lanes
4h core long and short for each market
Lane Browser
Market → lane group → direction
First choose a market, then a lane group and finally the long or short direction.
The lane data could not be loaded.
Step 1
BTC
Lane group
Step 2
–
Select lane
Step 3
Not selected
Lane details
Select a lane group and then choose Long or Short.
Dashboard
What this view shows
The public view deliberately shows only a limited part of the internal system.
Lane
A lane represents a defined evaluation path made up of market, strategy family, timeframe and direction.
Waiting
There is currently no active setup. The lane continues to watch the market and waits for the next relevant state.
Watchlist
A setup has been detected and is now in the live-eval system. It waits for confirmation or gets invalidated before that happens.
Shadow
A shadow lane can be fully analyzed and evaluated, but it has no permission for real execution.
Trade
A trade can only reach actual execution after the defined checks and state transitions have been completed.
Historical values
Entry, stop and target values are only shown publicly for completed processes. Active price levels stay private.