Live
The live paper system — full transparency
The AI agent still runs daily on paper money. Its decisions, account, sizing and post-mortem readiness are shown raw — no curation.
Latest run
6/26/2026, 2:00:26 PM
Paper account health
equity chart live — marks have drifted- Cash
- $69,697
- Invested
- $34,633
- Positions
- 4
- Equity marks
- 228
Last mark: 6/26/2026, 2:00:26 PM
Paper only — no real broker order. Equity reflects mark-to-market on persistent paper state.
LOG01What is it holding right now?Every position, with estimated marks.HOLDINGS
| Symbol | Qty | Avg entry | Latest price | Current estimated value | Unrealized P/L | Weight | Sell signal |
|---|---|---|---|---|---|---|---|
| ASML | 5.6260 | $1,631.75 | $1,841.18· Snapshot from the latest run | $10,358.40 | +$1,178.27 | 9.9% | holding — no sell signalNo sell reason emitted by this run yet. |
| TSM | 21.3862 | $428.96 | $434.99· Snapshot from the latest run | $9,302.79 | +$128.85 | 8.9% | holding — no sell signalNo sell reason emitted by this run yet. |
| AMD | 7.9322 | $488.08 | $532.57· Snapshot from the latest run | $4,224.47 | +$352.93 | 4.0% | holding — no sell signalNo sell reason emitted by this run yet. |
| MU | 8.8563 | $1,048.79 | $1,213.56· Snapshot from the latest run | $10,747.71 | +$1,459.27 | 10.3% | buy candidateNo sell reason emitted by this run yet. |
| Positions total | $34,633.37 | +$3,119.32 | 33.2% | ||||
When will this sell?
- Agent decision — the trader emits SELL only when the demo provider's composite analyst score for that symbol drops to ≤ −0.45 (deterministic, offline). In the current universe this rarely fires.
- Stop-loss / take-profit — evaluated before every run. Full exit on −8% from entry (stop-loss) or +25% from entry (take-profit). Trailing stop is OFF in the current risk config.
- Kill switch / drawdown halt — if the daily loss limit or drawdown halt fires, the pipeline liquidates every position before the LLM debate runs.
Paper only — no real broker order. "SELL" means the agent would reduce or close this position in paper mode; it is never advice to you.
Sell / Hold memos
ASMLopen memo
Snapshot
- Unrealized P/L
- —
- Conviction
- 0.50
- Current weight
- 9.9%
- Target weight
- 0.0%
- Fundamental score
- —
- Technical score
- —
- Bull score
- —
- Bear score
- —
Sell case
- Risk brief flagged an elevated condition (cap / halt / drawdown).
Hold case
- Trader agent emitted HOLD this run.
Trader rationale (verbatim)
Deterministic no-token demo decision from analyst score average +0.23. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply.
Risk brief excerpt
- Paper-demo risk: no live broker order will be sent by this provider. - Signal dispersion is 0.62; mixed analyst scores should reduce sizing. - Position caps, sector caps, kill switch, and drawdown h
What would change the decision
- Stop-loss triggers a full exit at$1,501.21
- Take-profit triggers a full exit at$2,039.68
- Trader emits SELL if the composite analyst score drops to ≤ −0.45.
- Kill switch / drawdown halt liquidates every position before the next LLM debate.
Missing evidence
- No delayed quote in this build — valuation falls back to the latest run's snapshot price.
- No news context fetched into the dashboard at build time.
- No fundamentals (P/E, growth, margins) are loaded into the report.
- No benchmark comparison is computed for this position.
- No macro / regime context (e.g. World Monitor) is linked here.
- Per-position thesis history is not tracked across runs.
Original thesis
The most important sell question is whether the original reason to own it is still true. Thesis text is operator-curated in data/thesis_state.json — never LLM-generated, never fabricated.
Original thesis not evaluated yet.Upcoming catalyst / event risk
A Buffett-style memo should know whether a thesis-changing event is near. Earnings dates come from data/event_calendar.json (manual fixture). No paid feed, no fabrication.
World / sector context
Read-only macro regime + sector signal from World Monitor (schema wm-trading-context-v1.0). Context for the memo only — World Monitor publishes signals; this dashboard never lets them place or trigger an order.
World Monitor context not evaluated yet.Read-only context — no orders, no broker, not financial advice.
Reina narrative context
Read-only editorial brief from Reina (saga-dev-log.dev, schema reina-brief-v1.0). Narrative context for the memo only — Reina publishes commentary; this dashboard never lets it place or trigger an order.
Macro: 全6銘柄が青信号で、まるでAI需要がバグってんのかってくらい爆速成長コミット。NVDA売上+85%、MSFT AI ARR+123%、GOOGL Cloud+63%と、主要数字が軒並み過去最高更新で、もうRTA記録更新待ったなし。
As of: 2026-05-29
Read-only context — no orders, no broker, not financial advice.
Technical context
Deterministic price-action read from daily candles — does the chart confirm or contradict the thesis?
Evidence only — indicators do not place orders, not financial advice.
Buffett-style dimensions
- Business quality / moatnot evaluated yet
- Valuation vs entrynot evaluated yet
- Earnings / financialsnot evaluated yet
- Balance sheet risknot evaluated yet
- Management / capital allocationnot evaluated yet
- Opportunity costnot evaluated yet
- Thesis broken?not evaluated yet
- Risk controlstop-loss 8% / take-profit 25% / kill switch + drawdown halt active
This is a paper-mode reasoning memo, not financial advice. Reasons are derived from the latest run's decision + briefs + the persistent paper state — never from data the dashboard does not actually have.
TSMopen memo
Snapshot
- Unrealized P/L
- —
- Conviction
- 0.50
- Current weight
- 8.9%
- Target weight
- 0.0%
- Fundamental score
- —
- Technical score
- —
- Bull score
- —
- Bear score
- —
Sell case
- Risk brief flagged an elevated condition (cap / halt / drawdown).
Hold case
- Trader agent emitted HOLD this run.
Trader rationale (verbatim)
Deterministic no-token demo decision from analyst score average +0.23. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply.
Risk brief excerpt
- Paper-demo risk: no live broker order will be sent by this provider. - Signal dispersion is 0.62; mixed analyst scores should reduce sizing. - Position caps, sector caps, kill switch, and drawdown h
What would change the decision
- Stop-loss triggers a full exit at$394.65
- Take-profit triggers a full exit at$536.21
- Trader emits SELL if the composite analyst score drops to ≤ −0.45.
- Kill switch / drawdown halt liquidates every position before the next LLM debate.
Missing evidence
- No delayed quote in this build — valuation falls back to the latest run's snapshot price.
- No news context fetched into the dashboard at build time.
- No fundamentals (P/E, growth, margins) are loaded into the report.
- No benchmark comparison is computed for this position.
- No macro / regime context (e.g. World Monitor) is linked here.
- Per-position thesis history is not tracked across runs.
Original thesis
The most important sell question is whether the original reason to own it is still true. Thesis text is operator-curated in data/thesis_state.json — never LLM-generated, never fabricated.
Original thesis not evaluated yet.Upcoming catalyst / event risk
A Buffett-style memo should know whether a thesis-changing event is near. Earnings dates come from data/event_calendar.json (manual fixture). No paid feed, no fabrication.
World / sector context
Read-only macro regime + sector signal from World Monitor (schema wm-trading-context-v1.0). Context for the memo only — World Monitor publishes signals; this dashboard never lets them place or trigger an order.
World Monitor context not evaluated yet.Read-only context — no orders, no broker, not financial advice.
Reina narrative context
Read-only editorial brief from Reina (saga-dev-log.dev, schema reina-brief-v1.0). Narrative context for the memo only — Reina publishes commentary; this dashboard never lets it place or trigger an order.
Conclusion: Q1売上$35.9B、FY26売上30%超成長ガイダンス。CapEx$52-56B上限へ増額、AI需要が牽引。3nm/2nmの価格も3-10%値上げで収益性強化。
Strategy: AI需要が構造的に強く、価格決定力も維持。押し目買い継続でOK。
Macro: 全6銘柄が青信号で、まるでAI需要がバグってんのかってくらい爆速成長コミット。NVDA売上+85%、MSFT AI ARR+123%、GOOGL Cloud+63%と、主要数字が軒並み過去最高更新で、もうRTA記録更新待ったなし。
Upcoming events: 2026-07-16 Q2 FY26
As of: 2026-05-29
Read-only context — no orders, no broker, not financial advice.
Technical context
Deterministic price-action read from daily candles — does the chart confirm or contradict the thesis?
Evidence only — indicators do not place orders, not financial advice.
Buffett-style dimensions
- Business quality / moatnot evaluated yet
- Valuation vs entrynot evaluated yet
- Earnings / financialsnot evaluated yet
- Balance sheet risknot evaluated yet
- Management / capital allocationnot evaluated yet
- Opportunity costnot evaluated yet
- Thesis broken?not evaluated yet
- Risk controlstop-loss 8% / take-profit 25% / kill switch + drawdown halt active
This is a paper-mode reasoning memo, not financial advice. Reasons are derived from the latest run's decision + briefs + the persistent paper state — never from data the dashboard does not actually have.
AMDopen memo
Snapshot
- Unrealized P/L
- —
- Conviction
- 0.47
- Current weight
- 4.0%
- Target weight
- 0.0%
- Fundamental score
- —
- Technical score
- —
- Bull score
- —
- Bear score
- —
Sell case
- Risk brief flagged an elevated condition (cap / halt / drawdown).
Hold case
- Trader agent emitted HOLD this run.
Trader rationale (verbatim)
Deterministic no-token demo decision from analyst score average +0.18. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply.
Risk brief excerpt
- Paper-demo risk: no live broker order will be sent by this provider. - Signal dispersion is 0.50; mixed analyst scores should reduce sizing. - Position caps, sector caps, kill switch, and drawdown h
What would change the decision
- Stop-loss triggers a full exit at$449.03
- Take-profit triggers a full exit at$610.10
- Trader emits SELL if the composite analyst score drops to ≤ −0.45.
- Kill switch / drawdown halt liquidates every position before the next LLM debate.
Missing evidence
- No delayed quote in this build — valuation falls back to the latest run's snapshot price.
- No news context fetched into the dashboard at build time.
- No fundamentals (P/E, growth, margins) are loaded into the report.
- No benchmark comparison is computed for this position.
- No macro / regime context (e.g. World Monitor) is linked here.
- Per-position thesis history is not tracked across runs.
Original thesis
The most important sell question is whether the original reason to own it is still true. Thesis text is operator-curated in data/thesis_state.json — never LLM-generated, never fabricated.
Original thesis not evaluated yet.Upcoming catalyst / event risk
A Buffett-style memo should know whether a thesis-changing event is near. Earnings dates come from data/event_calendar.json (manual fixture). No paid feed, no fabrication.
World / sector context
Read-only macro regime + sector signal from World Monitor (schema wm-trading-context-v1.0). Context for the memo only — World Monitor publishes signals; this dashboard never lets them place or trigger an order.
World Monitor context not evaluated yet.Read-only context — no orders, no broker, not financial advice.
Reina narrative context
Read-only editorial brief from Reina (saga-dev-log.dev, schema reina-brief-v1.0). Narrative context for the memo only — Reina publishes commentary; this dashboard never lets it place or trigger an order.
Macro: 全6銘柄が青信号で、まるでAI需要がバグってんのかってくらい爆速成長コミット。NVDA売上+85%、MSFT AI ARR+123%、GOOGL Cloud+63%と、主要数字が軒並み過去最高更新で、もうRTA記録更新待ったなし。
As of: 2026-05-29
Read-only context — no orders, no broker, not financial advice.
Technical context
Deterministic price-action read from daily candles — does the chart confirm or contradict the thesis?
Evidence only — indicators do not place orders, not financial advice.
Buffett-style dimensions
- Business quality / moatnot evaluated yet
- Valuation vs entrynot evaluated yet
- Earnings / financialsnot evaluated yet
- Balance sheet risknot evaluated yet
- Management / capital allocationnot evaluated yet
- Opportunity costnot evaluated yet
- Thesis broken?not evaluated yet
- Risk controlstop-loss 8% / take-profit 25% / kill switch + drawdown halt active
This is a paper-mode reasoning memo, not financial advice. Reasons are derived from the latest run's decision + briefs + the persistent paper state — never from data the dashboard does not actually have.
MUopen memo
Snapshot
- Unrealized P/L
- —
- Conviction
- 0.59
- Current weight
- 10.3%
- Target weight
- 10.8%
- Fundamental score
- —
- Technical score
- —
- Bull score
- —
- Bear score
- —
Sell case
- Risk brief flagged an elevated condition (cap / halt / drawdown).
Hold case
- Trader agent emitted BUY — wants more, not less.
Trader rationale (verbatim)
Deterministic no-token demo decision from analyst score average +0.38. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply.
Risk brief excerpt
- Paper-demo risk: no live broker order will be sent by this provider. - Signal dispersion is 0.65; mixed analyst scores should reduce sizing. - Position caps, sector caps, kill switch, and drawdown h
What would change the decision
- Stop-loss triggers a full exit at$964.89
- Take-profit triggers a full exit at$1,310.99
- Trader emits SELL if the composite analyst score drops to ≤ −0.45.
- Kill switch / drawdown halt liquidates every position before the next LLM debate.
Missing evidence
- No delayed quote in this build — valuation falls back to the latest run's snapshot price.
- No news context fetched into the dashboard at build time.
- No fundamentals (P/E, growth, margins) are loaded into the report.
- No benchmark comparison is computed for this position.
- No macro / regime context (e.g. World Monitor) is linked here.
- Per-position thesis history is not tracked across runs.
Original thesis
The most important sell question is whether the original reason to own it is still true. Thesis text is operator-curated in data/thesis_state.json — never LLM-generated, never fabricated.
Original thesis not evaluated yet.Upcoming catalyst / event risk
A Buffett-style memo should know whether a thesis-changing event is near. Earnings dates come from data/event_calendar.json (manual fixture). No paid feed, no fabrication.
World / sector context
Read-only macro regime + sector signal from World Monitor (schema wm-trading-context-v1.0). Context for the memo only — World Monitor publishes signals; this dashboard never lets them place or trigger an order.
World Monitor context not evaluated yet.Read-only context — no orders, no broker, not financial advice.
Reina narrative context
Read-only editorial brief from Reina (saga-dev-log.dev, schema reina-brief-v1.0). Narrative context for the memo only — Reina publishes commentary; this dashboard never lets it place or trigger an order.
Macro: 全6銘柄が青信号で、まるでAI需要がバグってんのかってくらい爆速成長コミット。NVDA売上+85%、MSFT AI ARR+123%、GOOGL Cloud+63%と、主要数字が軒並み過去最高更新で、もうRTA記録更新待ったなし。
As of: 2026-05-29
Read-only context — no orders, no broker, not financial advice.
Technical context
Deterministic price-action read from daily candles — does the chart confirm or contradict the thesis?
Evidence only — indicators do not place orders, not financial advice.
Buffett-style dimensions
- Business quality / moatnot evaluated yet
- Valuation vs entrynot evaluated yet
- Earnings / financialsnot evaluated yet
- Balance sheet risknot evaluated yet
- Management / capital allocationnot evaluated yet
- Opportunity costnot evaluated yet
- Thesis broken?not evaluated yet
- Risk controlstop-loss 8% / take-profit 25% / kill switch + drawdown halt active
This is a paper-mode reasoning memo, not financial advice. Reasons are derived from the latest run's decision + briefs + the persistent paper state — never from data the dashboard does not actually have.
Live quote unavailable for this build; values fall back to the latest run's snapshot price. Not a real-time mark.
Price as of: ASML 6/26/2026, 2:00:26 PM · TSM 6/26/2026, 2:00:26 PM · AMD 6/26/2026, 2:00:26 PM · MU 6/26/2026, 2:00:26 PM
Equity = cash + sum of position market values. The cash drop above is matched dollar-for-dollar by the paper positions below — nothing was lost, the money is held as shares in the in-process broker.
LOG02How is the portfolio composed?The full mix, cash included.MIX
- Cash$69,69766.8%
- ASML$10,3589.9%
- TSM$9,3038.9%
- AMD$4,2244.0%
- MU$10,74810.3%
Cash plus each position, sized by weight. This is the live state of the in-process paper broker for the latest run. Bar widths reflect what you actually hold — not a recommendation, not a target.
True paper equity (mark-to-market across runs)
Cash plus mark-to-market positions, sampled once per hourly paper run from the persistent broker state. Real movement only — never a fake horizontal line when marks are flat. 228 marks
Paper account state only — no real broker order has been placed. The series describes what the agent would have done; it is never a recommendation to you.
LOG03What did the last run decide — and why?Action, conviction and rationale, per symbol.DECISIONS
Action legend
- BUY= buy candidate
- HOLD= no trade
- SELL= sell / reduce if held
Demo mode only — no real broker orders are placed. BUY / HOLD / SELL describe what the trader agent would do; they are never a recommendation to you.
| Symbol | Action | Conviction | Target | Rationale |
|---|---|---|---|---|
| NVDA | HOLD（no trade） | 0.43 | 0.0% | Deterministic no-token demo decision from analyst score average +0.13. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply. |
| MSFT | HOLD（no trade） | 0.38 | 0.0% | Deterministic no-token demo decision from analyst score average +0.04. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply. |
| GOOGL | HOLD（no trade） | 0.43 | 0.0% | Deterministic no-token demo decision from analyst score average +0.13. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply. |
| META | HOLD（no trade） | 0.38 | 0.0% | Deterministic no-token demo decision from analyst score average +0.04. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply. |
| AAPL | HOLD（no trade） | 0.43 | 0.0% | Deterministic no-token demo decision from analyst score average +0.13. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply. |
| AMZN | HOLD（no trade） | 0.40 | 0.0% | Deterministic no-token demo decision from analyst score average +0.08. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply. |
| TSLA | HOLD（no trade） | 0.45 | 0.0% | Deterministic no-token demo decision from analyst score average -0.15. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply. |
| AMD | HOLD（no trade） | 0.47 | 0.0% | Deterministic no-token demo decision from analyst score average +0.18. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply. |
| AVGO | HOLD（no trade） | 0.43 | 0.0% | Deterministic no-token demo decision from analyst score average +0.13. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply. |
| PLTR | HOLD（no trade） | 0.40 | 0.0% | Deterministic no-token demo decision from analyst score average +0.07. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply. |
| SMCI | HOLD（no trade） | 0.35 | 0.0% | Deterministic no-token demo decision from analyst score average -0.01. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply. |
| ARM | HOLD（no trade） | 0.47 | 0.0% | Deterministic no-token demo decision from analyst score average +0.18. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply. |
| TSM | HOLD（no trade） | 0.50 | 0.0% | Deterministic no-token demo decision from analyst score average +0.23. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply. |
| ASML | HOLD（no trade） | 0.50 | 0.0% | Deterministic no-token demo decision from analyst score average +0.23. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply. |
| MU | BUY（buy candidate） | 0.59 | 10.8% | Deterministic no-token demo decision from analyst score average +0.38. No Anthropic/OpenAI/Gemini API was called; paper guardrails still apply. |
Agent backtest (answer-check)
Replay the real agent over a past week and grade it against known forward prices — comparison without waiting for closed windows.
View answer-check →LLM tracks (per provider)
This view is the demo track. Real-model decisions are separated per provider:
LOG05Is this experiment ready to conclude anything?Outcome-window progress — and what can't be said yet.READINESS
Research readiness
Can we actually test whether this framework has edge? An honest read on the data — not a backtest.
Data on hand
- Runs
- 254
- Days
- 30
- Decisions
- 3730
- Sell decisions
- 0
- Filled / orders
- 71/80
- Equity marks
- 228
- Provider
- demo ×246
Known gaps
- Too little history — walk-forward needs months of trading days.(have 30 / need 60)
- No sell decisions yet — the sell framework cannot be validated without sell events.(have 0 / need 20)
- Demo provider only — decisions are not real model output yet.(have 0 / need 1)
Signal attribution (schema)
Which signal carried the edge? Decided per signal once outcomes close — unknown until then.
| Signal | Supported | Later outcome | Status |
|---|---|---|---|
| Bullish equity research analyst | unknown | unknown | Unknown — insufficient history |
| Bearish equity research analyst | unknown | unknown | Unknown — insufficient history |
| Technical / price-action analyst | unknown | unknown | Unknown — insufficient history |
| Fundamental analyst | unknown | unknown | Unknown — insufficient history |
| Risk manager (stop / take-profit / halt) | unknown | unknown | Unknown — insufficient history |
No backtest results are claimed. Attribution stays “unknown” until enough out-of-sample outcome windows close.
Read-only research audit. No backtest has been run and no profitability is claimed.
LOG06How did the closed outcome windows score?Calls don't float free — they get graded.OUTCOMES
Research outcomes — paper replay
For each past paper decision, what did price do afterward — measured only from later trading days.
- Evaluated
- 3730
- Pending
- 30
- Partial
- 341
- Closed
- 3359
- No data
- 0
Price data as of: 2026-06-26
Raw close-to-close price movement after the decision, before fees and slippage. Not P&L, not direction-adjusted, not a backtest.
Read-only paper replay. No profitability is claimed; outcomes only mature into evidence once enough windows close.
LOG07How are position sizes decided?Kelly-based sizing with hard caps.SIZING
Position sizing preview
Suggested target sizing bands from the evidence — research/display only, not an instruction.
Preview only — does not place orders.
| Symbol | Current | Suggested band | Confidence | Top reasons | Blockers |
|---|---|---|---|---|---|
| ASML | 9.9% | Hold current | Medium |
| — |
| TSM | 8.9% | Hold current | Medium |
| — |
| AMD | 4.0% | Hold current | Medium |
| — |
| MU | 10.3% | Starter (2%)2% | Low |
|
|
Sizing is research-only until enough validated history exists. Bands are conservative and capped while closed outcome windows are scarce; no order is ever placed or resized here.
LOG08Where are the current positions weakest?Adversarial memos against our own book.RED TEAM
Red-team memo
An adversarial stress-test per position: how could this investment fail? Read-only critique from existing evidence.
ASMLOpen red-team / bear-case memo
The goal is to break the thesis before the market does.
Bear case
No thesis is tracked for this position, so the biggest risk is that we cannot say what would invalidate it.
Failure modes
- highThesisNo tracked thesis — we cannot state what would invalidate the position.
- medEvent / catalystEvent calendar unknown — a catalyst could hit unmodeled.
- lowMacro / sectorMacro / sector regime unknown — broad context not available.
Invalidation watchlist
- Stop-loss breach at −8% from entry.
- Daily close below SMA20 ≈1783.90
- No invalidation triggers recorded for this thesis yet.
- Next earnings date unknown.
- Macro regime / sector signal unknown.
Strongest counterargument
No strong bull signal is currently evident in the data.
What would change my mind
- Record an explicit thesis with invalidation triggers.
- Populate the event calendar for this symbol.
- Add a World Monitor regime snapshot.
- Wait for the technical trend to confirm.
This is a paper-mode critique, not financial advice.
TSMOpen red-team / bear-case memo
The goal is to break the thesis before the market does.
Bear case
No thesis is tracked for this position, so the biggest risk is that we cannot say what would invalidate it.
Failure modes
- highThesisNo tracked thesis — we cannot state what would invalidate the position.
- medEvent / catalystEvent calendar unknown — a catalyst could hit unmodeled.
- lowMacro / sectorMacro / sector regime unknown — broad context not available.
Invalidation watchlist
- Stop-loss breach at −8% from entry.
- Daily close below SMA20 ≈433.90
- No invalidation triggers recorded for this thesis yet.
- Next earnings date unknown.
- Macro regime / sector signal unknown.
Strongest counterargument
No strong bull signal is currently evident in the data.
What would change my mind
- Record an explicit thesis with invalidation triggers.
- Populate the event calendar for this symbol.
- Add a World Monitor regime snapshot.
- Wait for the technical trend to confirm.
This is a paper-mode critique, not financial advice.
AMDOpen red-team / bear-case memo
The goal is to break the thesis before the market does.
Bear case
No thesis is tracked for this position, so the biggest risk is that we cannot say what would invalidate it.
Failure modes
- highThesisNo tracked thesis — we cannot state what would invalidate the position.
- medEvent / catalystEvent calendar unknown — a catalyst could hit unmodeled.
- lowMacro / sectorMacro / sector regime unknown — broad context not available.
Invalidation watchlist
- Stop-loss breach at −8% from entry.
- Daily close below SMA20 ≈512.23
- No invalidation triggers recorded for this thesis yet.
- Next earnings date unknown.
- Macro regime / sector signal unknown.
Strongest counterargument
Price action is in an uptrend, which argues for staying.
What would change my mind
- Record an explicit thesis with invalidation triggers.
- Populate the event calendar for this symbol.
- Add a World Monitor regime snapshot.
This is a paper-mode critique, not financial advice.
MUOpen red-team / bear-case memo
The goal is to break the thesis before the market does.
Bear case
No thesis is tracked for this position, so the biggest risk is that we cannot say what would invalidate it.
Failure modes
- highThesisNo tracked thesis — we cannot state what would invalidate the position.
- medEvent / catalystEvent calendar unknown — a catalyst could hit unmodeled.
- lowMacro / sectorMacro / sector regime unknown — broad context not available.
- lowSizing / risk mgmtSizing preview confidence is low.
Invalidation watchlist
- Stop-loss breach at −8% from entry.
- Daily close below SMA20 ≈1036.21
- No invalidation triggers recorded for this thesis yet.
- Next earnings date unknown.
- Macro regime / sector signal unknown.
Strongest counterargument
The trader emitted a conviction call here — the bull case is the agent's own thesis.
What would change my mind
- Record an explicit thesis with invalidation triggers.
- Populate the event calendar for this symbol.
- Add a World Monitor regime snapshot.
This is a paper-mode critique, not financial advice.
LOG09How do missed calls get autopsied?Losses get documented hardest — the post-mortem system.POST-MORTEM
Post-mortem loop
Reviewing decisions after their outcome windows close — to learn, not to score. Nothing is reviewed until a window closes.
- Closed
- 3359
- Pending
- 371
- Review queue
- 3359
Status: Ready for review
What will be reviewed once windows close
- Each closed decision against its t+1/t+3/t+7d outcome (later prices only).
- Whether each evidence lane's stance lined up with the realized move.
- Whether the red-team memo's failure modes actually mattered.
- A short human-readable lesson per decision (no automatic grade).
Future signal grades (not applied yet)
- Correct warning
- False alarm
- Missed risk
- Useful support
- Noise
Read-only. No win rate, profitability, or edge is computed; review begins only once outcome windows close.
LOG10How is this experiment automated?The plumbing and schedule behind the daily runs.AUTOMATION
Automation
Paper only — no real orders are placed.
Paper automation
ON — paper runs may execute when the scheduler is activeTrading mode
TRADING_MODE = paper🔒 Live trading is locked
Live trading is hard-disabled in this dashboard. The paper broker, the persistence layer, and every helper refuse TRADING_MODE=live. No control here can enable real broker orders.
LLM provider
Active: demo (deterministic, no API)
| LLM provider | Notes | Required env | Status |
|---|---|---|---|
demo (deterministic, no API)Active | no-token / free / deterministic — always available, no paid LLM calls | — | OK |
Claude / Anthropic | API key configured — selecting this would call a paid LLM | ANTHROPIC_API_KEY | OK |
OpenAI | API key not configured | OPENAI_API_KEY | API key not configured |
Gemini | API key configured — selecting this would call a paid LLM | GEMINI_API_KEY | OK |
Ollama (local) | Local — no API key needed; runs against your local Ollama (start the daemon) | OLLAMA_HOST | OK |
How to change these
This panel is read-only: it reflects the most recent build. To toggle paper automation or pick a different provider, edit data/automation.json in the repo (paper_automation_enabled / active_provider), then redeploy. To enable a paid provider, set its API key in the environment that runs the build — never paste secrets into chat or UI.
Generated: 2026-06-26 19:00:35 UTC
Follow the research
Short notes when a new backtest, tear sheet, or post-mortem ships. Research-only — never investment advice.