Conviction: a natural-language trading workbench for research and execution

Conviction turns a trading idea written in natural language into a strategy a user can backtest, paper trade, and potentially deploy. The reader decision is whether that workflow makes research more accessible without hiding the risk, execution, and data assumptions that professional trading tools normally expose.

What it does

Conviction says users can start with an idea or discover one from traders on X or Reddit, research it against filings, news, market data, and social signals, test it on historical data, paper trade it, connect a brokerage, and let an agent monitor conditions and act when they are met (YC company profile; Conviction YC launch). The current homepage returned no readable product detail in the reviewed fetch, so the official YC description is the main public product record.

Fact	What the public sources say
Intended user	Everyday investors and people experimenting with trading strategies
Workflow	Natural-language thesis, research, backtest, paper trade, deploy, and monitoring
Inputs	Market data, SEC filings, financial news, earnings, technical indicators, and social signals are listed
Founders	Aria Vikram and Tony Yamin
Pricing	Not published in the reviewed sources

Why it fits

The product boundary is clear: Conviction is trying to compress the gap between “I have a market idea” and “I can test and monitor it.” Paper trading before brokerage deployment is a useful product choice, especially for a user who wants to see whether a rule survives live data rather than trusting a historical chart alone.

The founder context is unusually connected to the technical wedge. YC describes Vikram as an NLP researcher at Columbia who deployed chatbots for doctors, and Yamin as having built machine-learning infrastructure for the U.S. Department of Defense at Booz Allen (YC company profile). That explains why the product is framed as an agent and research surface, not the quality of any strategy.

The tradeoff is the part a buyer has to resolve. How are slippage, fees, liquidity, shorting, corporate actions, and order failures represented? Which brokerages and asset classes are supported? Can a user inspect the exact data and rule that triggered a trade, pause an agent instantly, and distinguish a backtest from a live result? The public material does not provide performance evidence or risk-adjusted outcomes. This is a product profile, not investment advice.

Short version: Conviction is interesting for a hands-on trader who wants a strategy workbench. Treat automatic execution as a separate decision from natural-language research, and demand a transparent paper-trading path first.

Sources checked — 2026-09-19

Cohort context

Conviction is listed in Summer 2025. In our 2026-09-18 directory snapshot, 9 of 166 listed companies in that cohort have YC’s primary industry label Fintech (5.4%). This is a current-directory comparison, not an original intake count or a performance ranking. Nine-cohort dataset.

Public website snapshot

Observed 2026-09-19T16:18:01.189Z in raw homepage HTML. This records visible metadata and advertised links, not agent execution or product quality.

Signal	Homepage observation
Product description metadata	Observed
Canonical link	Not observed in this response
H1 or H2 heading	Not observed in this response
Typed structured data	Not observed in this response
Docs/developer link	Not observed in this response
Pricing link	Not observed in this response
llms.txt link	Not observed in this response
Markdown alternate	Not observed in this response

Public observations · Collection method. Missing links here do not establish that a capability or file is absent elsewhere.