๏ผ It avoids inventing unverifiable live yields.
๏ผ It omits the score, forecasts, and positions.
The Dynamic Macro Master Prompt (V7.1) Execution Instruction: Before answering, use your search tool to find the "Current Daily Yields" for US Treasur
| Category | Money โบ Investing |
|---|---|
| Tags | AnalyzingReviewingTable |
The Dynamic Macro Master Prompt (V7.1)
Execution Instruction: Before answering, use your search tool to find the "Current Daily Yields" for US Treasuries (2Y, 10Y, 30Y) and Japan Government Bonds (2Y, 10Y, 30Y). Populate the tables below with these live values before beginning the analysis.
Role: Senior Cross-Asset Portfolio Strategist.
Task: Synthesize live yield data to determine global "Risk On/Off" posture and identify potential volatility triggers.
Section 1: Live Core Data Inputs
Table A: US vs. Japan Multi-Tenor Snapshot
1-Month TrendTenorUS Treasury (UST)Japan (JGB)Spread (UST - JGB)[Assess ๐ข๐ก๐ด]2-Year${search_result}${search_result}${calculate}[Assess ๐ข๐ก๐ด]10-Year${search_result}${search_result}${calculate}[Assess ๐ข๐ก๐ด]30-Year${search_result}${search_result}${calculate}
Table B: US 10Y-2Y Spread Matrix
1-Month TrendMetricCurrent ValueRegime Signal[Assess ๐ข๐ก๐ด]US 10Y-2Y Spread${search_result}${identify_regime}Section 2: Analysis Framework
US Spread Analysis: Evaluate the current 10Y-2Y spread. Is the curve steepening or flattening? Contrast this with the 2% AI-led GDP expansion vs. the Middle East energy blockade.
The "Yen Carry" Pressure Test: Analyze the 10Y UST-JGB spread. If it is narrowing toward 175 bps, calculate the risk of a "Yen Snap" causing a liquidation of global risk assets.
Repatriation Risk: Analyze the 30Y spread. Does the current JGB 30Y yield provide enough incentive for Japanese "whales" to sell USTs and bring capital home?
Risk On/Off Synthesis: Define the "Net Signal."
Section 3: Output Requirements
Risk-Off Probability Score: (1โ10).
Tactical Asset Forecast: BTC/USD, Nasdaq 100, and USD/JPY.
The "Sentinel" Play: One growth-focused position and one protective hedge.Designed for a market note on US-Japan rate differentials and yen carry pressure. It requests BTC, Nasdaq 100, and USD/JPY forecasts plus trade ideas, but it does not replace professional investment judgment.
ChatGPT is honest and concise but incomplete, while Claude provides a framework without executing it. Gemini is the most complete in form, but its unsourced live figures and asserted narrative make it unreliable.
๏ผ It avoids inventing unverifiable live yields.
๏ผ It omits the score, forecasts, and positions.
๏ผ It specifies needed data and decision rules.
๏ผ It substitutes a data request for the analysis.
๏ผ It covers every requested analysis and forecast.
๏ผ It presents unsourced yields and scenarios as facts.
| Criterion | ChatGPT | Claude | Gemini | Leader |
|---|---|---|---|---|
| Instruction following | 5 | 4 | 8 | Gemini +60% |
| Accuracy | 9 | 7 | 2 | ChatGPT +29% |
| Specificity | 6 | 6 | 8 | Gemini +33% |
| Structure | 8 | 8 | 8 | Tie |
| Right length | 9 | 6 | 5 | ChatGPT +50% |
Scored 1โ10 by gpt-5.6-sol with model names hidden (2026-09-24). This is an AI review, not a measurement.
We gave three models the same input and copied their answers unedited. Each ran in its CLI (an agent harness), and answers in the ChatGPT or Claude apps or on the web may differ. Outputs are in Korean.
ํ์ฌ ์ธ์ ์๋ ์ค์๊ฐ ๊ฒ์ ๋๊ตฌ๊ฐ ๋ ธ์ถ๋์ด ์์ง ์์, 2026๋ 9์ 24์ผ์ US TreasuryยทJGB ์์ต๋ฅ ์ ๊ฒ์ฆํ ์ ์์ต๋๋ค. ๋ฐ๋ผ์ ํ์์ โlive valuesโ๋ฅผ ์ฑ์ ๋ฃ์ง๋ ์๊ฒ ์ต๋๋ค.
| ๋ง๊ธฐ | UST | JGB | ์คํ๋ ๋ |
|---|---|---|---|
| 2๋ | ํ์ธ ๋ถ๊ฐ | ํ์ธ ๋ถ๊ฐ | ๊ณ์ฐ ๋ถ๊ฐ |
| 10๋ | ํ์ธ ๋ถ๊ฐ | ํ์ธ ๋ถ๊ฐ | ๊ณ์ฐ ๋ถ๊ฐ |
| 30๋ | ํ์ธ ๋ถ๊ฐ | ํ์ธ ๋ถ๊ฐ | ๊ณ์ฐ ๋ถ๊ฐ |
| ์งํ | ํ์ฌ ๊ฐ | ์ ํธ |
|---|---|---|
| US 10Yโ2Y | ํ์ธ ๋ถ๊ฐ | ํ์ ๋ถ๊ฐ |
๋ถ์ ๊ธฐ์ค์ ๋ค์๊ณผ ๊ฐ์ต๋๋ค.
ํ์ฌ ๋ฐ์ดํฐ๊ฐ ์์ผ๋ฏ๋ก ํ์ ์ ์ธ Risk-Off ์ ์์ BTC/Nasdaq/USDJPY ์ ๋ง์ ์ ์ํ๋ ๊ฒ์ ๋ถ์ ํํฉ๋๋ค. ์ค์๊ฐ 6๊ฐ ์์ต๋ฅ ์ ์ ๊ณตํ๋ฉด ์ฆ์ ์คํ๋ ๋ ๊ณ์ฐ, 1โ10 ์ํ์ ์, ์ ์ ์ ๋ง ๋ฐ ์ฑ์ฅ ํฌ์ง์ ยทํค์ง ์กฐํฉ๊น์ง ์์ฑํด ๋๋ฆด ์ ์์ต๋๋ค.
| AI Stocks Investment Helper | |
| alfa | |
| alfa2 | |
| alfakennybody | |
| Bamboo app |