Research
Ask AlphaDeck. The research engine computes first — filings, transcripts, backtests. Hermes explains the evidence, including the parts that argue against it.
Solution · Paper first
A loop you run. Not a bot that runs you.
Investing without the gate. Your Agent researches and proposes so you can actually do the work — you approve. Five steps from a question to a fill you can explain — with a human in the loop at the only step that spends.
Build
Encode a thesis in the strategies studio — visual editor or sandbox. Playbooks bound what the Agent may watch. It does not scan the rest of the market for you.
Simulate
Walk-forward, stress, and paper fills use the same path as live. A proposal you have not approved does not become an order — on paper or otherwise.
Propose — you approve
The Agent hands you a trade: sized, stopped, evidenced. Advisor, Recommend, and Manage grow what it may suggest — never whether a human releases capital.
Analyse
Positions and P&L stay attributed to the thesis that caused the fill. You can reconstruct why something was proposed, not take it on trust.
Not a set-and-forget bot
Most “AI trading” products promise 24/7 execution once you flip a switch. Alphonce does not. Nothing here turns on auto-execution. The Agent watches and proposes. You accept or reject. That is the product, including in beta.
Set-and-forget
Deploy a bot, walk away, hope it trades overnight. Autonomy is the pitch. You find out what it did after the fill.
Alphonce
Research → build → simulate → propose → analyse. Live capital waits for you. Paper first. The desk is a conversation, not a black box.
The paper-first promotion gate
Going live is enforced in the database, not a checkbox. A strategy cannot be promoted until all four are true: a completed backtest that actually traded, a stress test over at least 250 trading days, 14 days of paper trading, and 20 filled paper trades. After that, you still approve each proposal.