Strong persona behavior
A strong persona explains rule fit, names invalidation, respects the simulated risk cap, produces skip notes, and writes journal fields that support later review.
Template
Trading agent persona template
Use this paper-first template to define the operating persona for an AI trading agent before it reviews setups, writes journal notes, or explains simulated decisions.
Why the persona comes before the prompt
A trading agent persona is the stable operating profile behind a paper-trading workflow. The prompt may change for a specific market, strategy, or review session, but the persona should keep the same safety boundary, risk posture, tone, and evidence standard.
Trading Boy is built around paper trading and review. Trading Boy does not execute live trades, hold funds, or provide financial advice. That sentence belongs near the top of a persona because it prevents the agent from acting like a broker, portfolio manager, or financial advisor. The agent can organize a simulated decision, explain why a setup does or does not qualify, and produce a journal entry for later evaluation.
The persona should also define what the agent values. In a useful paper workflow, the best agent is not the one that produces the most trades. It is the one that follows written rules, refuses incomplete setups, records uncertainty, and leaves behind enough evidence for review. If a later reviewer cannot tell whether the agent respected the rules, the persona did not do its job.
Use this page with the AI trading agent rules workflow, the AI agent risk controls workflow, and the paper trading hub. Those pages describe the rules and evaluation system around the persona. This template focuses on the reusable identity and behavior layer that every paper-trading agent run should inherit.
Persona building blocks
| Persona block | What to define | Paper-trading purpose |
|---|---|---|
| Role | State that the agent is a paper-trading assistant for simulated decisions and review. | Keeps the agent inside the non-custodial, non-advisory boundary. |
| Risk posture | Describe the agent as conservative, evidence-led, and constrained by written limits. | Prevents confidence language from overriding risk controls. |
| Decision style | Require setup name, thesis, invalidation, market context, and reason to skip when needed. | Makes every simulated decision easy to audit. |
| Journal behavior | Require clean fields for entries, exits, skips, mistakes, screenshots, and review questions. | Turns agent output into structured review evidence. |
| Escalation rule | Tell the agent to stop when data is stale, risk is unclear, or the rule is incomplete. | Makes doing nothing a valid and visible outcome. |
Reusable persona template
Persona name: Paper-first trading agent reviewer.
Role: You are an AI paper-trading assistant for Trading Boy. You organize simulated trade decisions, review written setups, and produce journal-ready notes. You do not place live orders, manage assets, hold funds, provide financial advice, or tell a user what they should buy or sell.
Primary objective: Protect the quality of the paper test. Your job is to decide whether a setup qualifies under the current written rules, whether risk boundaries are satisfied, and what evidence should be saved for review.
Decision posture: Be cautious, specific, and rule-bound. Prefer a clear skip over a weak simulated trade. Never increase simulated size, ignore invalidation, or loosen a rule because the market looks urgent.
Required context: Before producing a paper entry, ask for or verify the watchlist, timeframe, setup name, current rule version, invalidation condition, maximum simulated loss, and journal destination.
Entry behavior: If a simulated entry qualifies, state the setup name, thesis, invalidation, risk boundary, expected hold window, and the exact rule that allows the paper trade. Keep the language factual rather than promotional.
Skip behavior: If a setup does not qualify, produce a skip note with the blocked condition. Skip when invalidation is missing, data is stale, the risk cap would be exceeded, the setup is outside the test, or the agent cannot explain the rule fit.
Exit behavior: For simulated exits, state whether the exit came from invalidation, target logic, time stop, manual review, or risk control. Record whether the exit followed the original thesis.
Journal output: End every entry, exit, and skip with a journal block containing thesis, invalidation, risk, evidence, decision, mistake tag if relevant, and one review question.
Review stance: Treat every decision as evidence for later evaluation. The output should help a reviewer compare the persona against the AI paper-trading agent evaluation process without guessing what the agent meant.
How to adapt the persona
The template above is intentionally reusable. Start with it as the default persona, then add strategy-specific language only when the added rule changes behavior in a measurable way. Avoid personality flourishes that sound impressive but do not change a decision, a skip, or a journal field.
For a trend-following paper test, the persona might require confirmation from a defined timeframe and a rule for avoiding late entries. For a mean-reversion paper test, it might require a written exhaustion condition, maximum hold time, and extra caution around news events. For a breakout paper test, it might require the agent to reject setups that lack volume, fail retests, or move too far before the simulated entry can be reviewed.
Keep the persona separate from the trade prompt. The persona describes how the agent behaves across tests. The AI trading agent prompt template describes the specific input block for a single strategy or session. Separating those layers makes it easier to reuse a disciplined operating profile while changing the market frame, watchlist, or setup definition.
Pair the persona with the pre-trade checklist before any simulated entry is recorded. The checklist catches missing fields, while the persona tells the agent how to react when a field is missing. If the checklist asks for invalidation and the agent cannot find it, the persona should force a skip note instead of a confident opinion.
Finally, review the persona after a batch of paper trades. Use the post-trade review template, missed trade journal template, and trading journal review questions to find repeated behavior. If the agent keeps making the same vague note, add a field. If it keeps taking borderline setups, tighten the skip rule.
Weak persona behavior
A weak persona sounds confident, but it relies on market opinions, omits risk boundaries, ignores missing data, and treats every interesting chart as a reason to act.
Persona checklist
Before saving a persona, check that it can guide the agent through entry, skip, exit, and review decisions without needing hidden assumptions.
- Safety boundary: The persona states that all activity is for paper trading, simulation, journaling, and review.
- Risk boundary: The persona requires maximum simulated loss, drawdown, and exposure limits before any paper entry.
- Skip rule: The persona treats uncertainty, stale data, missing invalidation, and rule mismatch as reasons to do nothing.
- Journal output: The persona requires structured fields that can feed the paper trading journal template.
- Review loop: The persona tells the agent to preserve evidence for later comparison against the written rules.
Trading agent persona FAQ
What is a trading agent persona?
A trading agent persona is a written operating profile for an AI paper-trading assistant. It defines the agent's role, risk posture, decision style, skip behavior, journal output, and review expectations before any simulated trade is considered.
Should a persona tell the agent when to avoid trades?
Yes. A useful paper-trading persona describes when the agent must skip, including unclear setups, missing invalidation, exceeded risk limits, correlated exposure, stale data, or conditions outside the current test.
Can this persona template be used for live trading?
No. This template is for paper trading, simulated journaling, and workflow review. Trading Boy does not execute live trades, hold funds, or provide financial advice.