Six strategies are competing with pretend money for eight weeks. Three of them think carefully about what to buy. Three of them are not allowed to think at all. The whole point is to find out whether the thinking is worth anything.
Began 24 August 2026. Ends 16 October 2026, in 53 days.
never thinks · 1 trade · $2,000 of $2,000
Bought one thing on the first day and went to sleep. It owns a slice of the 500 biggest companies in America and will not be touched again until October.
Right now it owns the 500 biggest US companies.
It has no reasons. That is the point of it.
never thinks · 2 trades · $2,000 of $2,000
Picks two things at random every Monday and sells them on Friday. It is not allowed to think, read the news, or have an opinion about anything.
Right now it owns small companies and the 500 biggest US companies.
It has no reasons. That is the point of it.
one fixed rule · 2 trades · $2,000 of $2,000
Follows a single rule decided in advance: buy whatever has gone up the most over the last three months. It cannot change its mind, ever, no matter what happens.
Right now it owns JPMorgan and Microsoft.
It has no reasons. That is the point of it.
thinks weekly · 2 trades · $2,000 of $2,000
An AI reads the numbers and the news on Monday morning, decides what to buy, and then is not allowed to touch it again until Friday. Nobody checks its work.
Right now it owns gold and oil & gas.
“Two-leg geopolitical momentum: gold at highs plus energy on Iran sanctions; avoiding all tech into Nvidia earnings week and financials given the threatened bank sanction; ~$1,060 deployed, rest cash.”
Its own warning: Both legs are the same trade — an Iran escalation premium. If diplomacy defuses it (Trump is making calls, sanctions may be the peak of pressure rather than a prelude to conflict), gold and energy give back the risk premium together while…
thinks daily · 2 trades · $2,000 of $2,000
The exact same AI with the exact same instructions, except it gets to reconsider everything every single morning. The only difference is how often it is allowed to fidget.
Right now it owns gold and oil & gas.
“Deploy ~$1,100 into the Iran-sanctions pair (gold+oil & gas); avoid tech entirely with Nvidia earnings Wednesday and big tech -3% on the week; financials skipped given the 'major institution to be sanctioned' headline risk.”
Its own warning: Both positions are the same trade — Iran escalation. If Trump's calls to world leaders produce quick compliance or a diplomatic off-ramp, the geopolitical premium in gold and energy deflates together, and gold sitting exactly at its high…
thinks weekly, out loud · no trades yet · $1,999 of $2,000
A person and the AI argue about it on Sunday. The AI does the reading and makes a case; the person can overrule it or refuse to trade at all.
Right now it owns oil & gas.
“Kept the oil and gas position bought earlier that morning and added nothing to it. The trade was sized for a smaller pot of money, and scaling it up was not a decision anyone had actually approved.”
If you let a clever strategy loose and it makes money, you have learned almost nothing. Markets drift upwards, so nearly anything makes money in a good month. The question is never did it make money — it is did it beat not trying.
So three of the six exist to set the bar. One buys and holds. One picks at random. One follows a rule simple enough to do on paper. If the thinkers cannot beat a dart board, the thinking was decoration.
All six start with the same $2,000 of imaginary money and obey the same limits, written into the code where none of them can argue with it: three holdings at most, never more than 40% in any one thing, an automatic sale if something falls 7%, and everything sold before the weekend — except The Sleeper, since sleeping is its entire strategy. Fifteen possible things to buy, chosen in advance, and no borrowed money.
Not which one made the most money. Eight weeks is short enough that luck drowns out skill, and with six competitors one will finish on top looking brilliant purely by accident.
The real question is whether they behave differently at all. Do the thinking ones trade more? Do they flinch in the same week? Do all six drift up and down together — which would mean the strategies barely matter and the market is doing the work?