NeuroDynamicRunning now
Can an AI beat just holding?
We gave an AI system pretend money and real market prices, and let it invest by its own rules. Here’s how it’s doing, compared with simply buying a world fund and leaving it alone. See the full evidence ↓
Loading the latest figures…
The pots
Each pot tries a different way of investing, side by side, with the same starting money.
Pretend money, real prices. These pots hold no real money and you can’t invest in them. Values are worked out from market prices and don’t fully count dealing costs, fees or tax.
The race so far
Each line starts at 100. Above 100 means it’s making money; below means it’s losing. The dashed line is just holding a world fund: the yardstick every pot is trying to beat. Drag the timeline, or point at the chart, to read any date; dots mark the pots’ decisions.
Is its “good time to buy” signal any good?
Every time the AI flags a good moment to buy, we wait 20 days and check what really happened.
Latest decisions
What the pots did recently, and why. Every move follows the written rules.
Shown to explain how the experiment works, not as tips. Please don’t copy these trades: they’re made with pretend money, may already be out of date, and some will turn out to be wrong.
Market weather
The general mood of the markets, from live prices. Background only: the pots’ own rules do the reacting.
How it works, in three steps
The full evidence
Open any section to see what the system checks, and how we keep it honest.
Loading the latest evidence…
What are the markets telling us?Traffic lights worked out by fixed rules, not guessed by the AI
Loading…
Bitcoin and Ethereum, last 90 daysEach coin starts at 100, so you can compare them fairly
Market prices aren’t the pots’ results. This shows what the crypto pot was up against.
Is new money arriving?Money flowing into and out of Bitcoin and Ethereum funds
Fund flows show money going into or out of exchange-traded funds. They don’t identify every buyer or predict the next move. Figures from Farside Investors.
Checks, limits and proofEvery call is saved and scored later, including the wrong ones
A missing input shows as unknown. A paused experiment stays visible. Results are never reset to make the record look better.
What does green actually mean?How to read the traffic lights
Green means both Bitcoin and Ethereum pass the trend rules on completed daily prices, recent fund flows are positive for both, and independent exchanges agree on the price. Amber means the evidence is mixed or incomplete. Red means neither trend passes. Grey means something essential can’t be checked or is overdue.
Risk is judged separately: unusual borrowing or stretched prices can warn that a move is fragile. These are clear monitoring thresholds, not a calculated chance of profit.
How we keep it honestWhat’s recorded, when it’s checked, and what the AI is allowed to do
The first assessment each day is saved. Once enough time has passed, it’s checked against what happened 7 and 20 days later. No past calls are invented, and dealing costs aren’t included, so a good early score isn’t proof of an edge.
The AI helps explain the facts in plain English, but it can’t change the rules or the money. Anything it says is checked, and fixed wording is used if it invents a detail.
A separate, higher-risk experiment uses its own pretend money and stop limits. It isn’t mixed into the pots above.
What this doesn’t claim
- It hasn’t beaten the market yet. A few months is far too short to prove anything.
- It’s pretend money. No real trades, and costs and taxes aren’t fully counted.
- It isn’t advice. Nothing here is a recommendation to buy or sell anything.