One AI model guessing isn't a committee.
Turion is an autonomous investment committee — not a single model with an opinion. Every trade has to survive a debate, a skeptic, and two independent risk gates before it ever reaches the market, and every decision it makes is explainable after the fact, in plain English.
What happens before a single trade is placed
Read the market
Classifies the regime, runs technical indicators, searches history for similar setups, and reads the news — before anyone forms an opinion.
Market Regime · Technical · Historical Analog · News Sentiment
Argue about it
A Bull Case Analyst and a Bear Case Analyst — each a separate AI model — build the strongest possible case for and against the trade.
Bull Case Analyst · Bear Case Analyst
Stress-test the argument
A Skeptic Agent checks both cases for confirmation bias, overfitting, hype cycles, and narrative traps — and can dock confidence directly.
Skeptic Agent
Size the risk
A dedicated Risk Agent sets the maximum position size, sets a stop-loss, and can veto the trade outright — before it ever reaches a decision.
Risk Agent · veto authority
Decide
A CIO Agent synthesizes every stage above — regime, debate, skeptic review, risk limits — into one weighed call, with its reasoning kept intact.
CIO Agent · multi-model synthesis
Final sign-off
A Chief Risk Officer has the last, binding word. It can only hold or shrink a position — never increase it beyond what Risk already capped.
Chief Risk Officer · binding authority
Act, and explain
Only then does an order reach the market. Every stage above — including a stand-down — is logged and explainable in plain English, one click away.
Execution · full audit trail
Only after all seven stages does an order reach Alpaca for execution — or the committee stands down.
Why a committee, not a bot
The unique bet behind Turion: a trading decision is only as trustworthy as the argument that survived to make it.
Built to argue with itself
Most AI trading tools are one model, guessing with confidence. Turion runs a Bull case, a Bear case, and a Skeptic whose entire job is to find the holes in both — before a decision ever forms.
Two independent risk gates
A Risk Agent sets position caps and stop-losses with veto power. A separate Chief Risk Officer reviews everything again downstream and can only hold or shrink the trade — never override risk upward.
Nothing is a black box
Click any trade — or any decision the committee chose not to act on — and see the plain-English reasoning behind it, built from the decision's own stored data. No fabricated explanations, ever.
It doesn't hide the bad news
A veto, a skeptic's concern, or a flat "no trade" is stated as plainly as a winning call. Explaining why the system correctly stood down matters as much as explaining why it acted.
See the committee run
Every stage above is real, working code — not a mockup. Start free and watch it argue, gate, decide, and explain itself on your own dashboard.