60 sessions
Exactly 60 eligible U.S. regular trading sessions after a prospective Day 1 start.
Transparent AI-assisted trading research and simulation
PRELAUNCH / NOT STARTED
Experiment ID: ATL-PUBLIC-SIM-V1
AI Trading Lab — £10K Transparent Simulation v1
A prospective 60-session research experiment designed to show the full path from strategy research to controlled simulated execution and a public audit trail — including losses, costs, blocked proposals and NO TRADE days.
Planned virtual starting equity
£10,000
£10,000 is virtual starting equity. No real money is being traded, and it is not leveraged broker buying power.
SIMULATED RESEARCH EXPERIMENT — NO REAL MONEY IS BEING TRADED. £10,000 IS VIRTUAL STARTING EQUITY, NOT BROKER BUYING POWER.
This MVP collects no personal data. Email/waitlist capture is deferred until there is an approved privacy-safe storage path.
What this is
Research you can inspect, not a promise you have to trust
AI Trading Lab is a public product for transparent trading research and simulation. The goal is to make the rules, assumptions, risk controls, costs and outcomes visible enough to inspect rather than asking people to trust an “AI trading” claim.
AI-assisted research can be wrong, incomplete or out of date. Important information should be checked against reliable original sources. The experiment is not personalised investment advice and does not place broker orders or manage customer money.
Frozen v1 boundaries
Small enough to understand. Strict enough to audit.
Common shares only
Eligible U.S.-domiciled ordinary common shares, primary-listed on NYSE or Nasdaq.
Long only
No short selling, margin or leverage. Virtual exposure remains inside frozen deterministic limits.
Fixed costs
Simulated fills include the frozen adverse spread/slippage, commission and FX allowance rules.
Human veto, not override
A human may APPROVE or REJECT an eligible proposal, but cannot override a deterministic FAIL or BLOCKED control.
NO TRADE is valid
No candidate, rejection, blocked data, capacity or risk controls can correctly produce NO TRADE.
Control path
How a simulated trade would reach the public record
- 1Research / strategy
Only an eligible frozen strategy version can generate a candidate.
- 2Candidate
The idea becomes an immutable proposal input, not an order.
- 3Deterministic controls
Risk and portfolio rules may reduce, fail or block exposure.
- 4Human approval
APPROVE or REJECT only; no bypass of failed controls.
- 5Simulated execution
A deterministic recheck happens before any simulated fill and cost accounting.
- 6Public audit trail
Proposals, decisions, fills, costs, positions, NO TRADE and incidents are retained.
Future public metrics
Nothing has started, so nothing is pretending to be performance
When the prospective experiment is allowed to start, the public dashboard will read verified public-v1 evidence. Until then, every performance field stays explicitly unstarted.
SIMULATED RESEARCH EXPERIMENT — NO REAL MONEY IS BEING TRADED. £10,000 IS VIRTUAL STARTING EQUITY, NOT BROKER BUYING POWER.
Open the read-only public dashboard →Read the full simulation disclosure →
Methodology
Rules are frozen before results exist
Material v1 changes after Day 1 are not silently edited into history. A material change requires a separately declared experiment version.
Risk controls
Long only, no leverage, a 75% gross-exposure ceiling and 15% new-position ceiling, both measured against starting virtual equity, plus up to five open positions and deterministic loss/drawdown stops.
Execution assumptions
Close-derived decisions can fill no earlier than the next eligible regular-session open, with 15 bps adverse simulated spread/slippage per executed side and USD 1.00 simulated commission per executed side.
Currency assumptions
GBP reporting uses frozen reference-FX rules and explicit simulated FX allowances; missing FX does not justify inventing a GBP value.
Evidence rules
Losses, costs, drawdowns, blocked proposals, human rejections and NO TRADE days remain part of the record.
Day 1 baseline
The start line must be proved before the first result
Status: Not yet created. Before launch, the project must freeze machine-readable evidence for the strategy manifest, fresh zero-history simulation book, market universe, data/source identities, risk and cost configuration, disclosure, metric definitions and reproducibility hashes.
If a mandatory item cannot be proved before the intended Day 1 market open, the correct outcome is NO START, not a weaker rule or a backfilled baseline.
Public disclosure
Simulation context stays next to the experiment
Full disclosure for ATL-PUBLIC-SIM-V1
This is a simulated research experiment. No real money is being traded and the £10,000 starting amount is virtual equity, not cash deposited with a broker and not leveraged buying power.
Results depend on the experiment's stated market-data, timing, execution, commission, spread/slippage, FX and portfolio assumptions. Real trading could differ materially because of liquidity, price gaps, latency, taxes, fees, market impact, order handling, rejected orders, outages and other market conditions.
Simulated or past performance does not predict future results. This experiment may make or lose virtual money, and NO TRADE is a valid outcome. We publish losses, drawdowns, costs, blocked trades and rule-governed inactivity as well as gains.
AI-assisted research can be wrong, incomplete or out of date. AI/model output does not have authority to bypass deterministic risk controls, human approval rules or the simulation ledger.
Nothing in this experiment is personalised investment advice, a recommendation to buy or sell a security, a promise of profit or an invitation to copy the simulated trades. Do not make an investment decision based solely on this experiment; check important information against reliable original sources and consider appropriate professional advice for your own circumstances.
Interested in the experiment?
Inspect the public evidence model now
There is no email capture yet because the repository has no approved personal-data storage path. The read-only dashboard makes the future evidence structure inspectable without inventing a database or marketing stack.