SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillsreview

review

Data & analytics

View repository

Updated 10-15 16:47

About this skill

/review - Periodic Review

Conduct periodic reviews of trading decisions, verify the accuracy of recommendations, and consolidate investment experience.

How to Use

  • /review - Automatically determine whether a weekly review or monthly review should be done
  • /review week - Force a weekly review
  • /review month - Force a monthly review
  • /review 2026-01-08 - Review a specific trade

Execution Steps

Step 1: Determine the review scope

Read the following files to determine what needs to be reviewed:

  • 股市信息/Records/trades.md - Find trades that need verification
  • 股市信息/Daily/ - Find the briefings and recommendations for this period

Automatic determination logic：

  • If it is the weekend (Saturday/Sunday) → weekly review
  • If it is the last trading day of the month → monthly review
  • Otherwise → ask the user what they want to do

Step 2: Obtain verification data (core step)

Run the data retrieval script to obtain the latest prices of the relevant instruments:

cd "股市信息" && python3 scripts/fetch_market_data.py

The script output includes：

  • holdings: Current holdings market data/net value (including ETF/A-shares/Hong Kong stocks/funds, including cost, profit and loss, holding days)
  • indices: Market indices (used to compare benchmark returns)
  • north_flow: Northbound capital data (prefer the intraday real-time value of the day, may include a time field)

Must strictly comply：

  1. Wait for the script to finish executing(about 1-2 minutes)
  2. All prices must come from the script JSON output, and must not be fabricated
  3. If the script fails, inform the user; the use of default values is prohibited
  4. Price display format:X.XXX（来源：AKShare）

Step 3: Perform the review analysis

Weekly review content

# 周复盘 YYYY-WXX (MM-DD ~ MM-DD)

## 一、本周交易回顾

| 日期 | 操作 | 标的 | 操作价 | 现价 | 涨跌 | 初步判断 |
|------|------|------|--------|------|------|----------|
| ... | ... | ... | ... | ... | ... | ... |

## 二、建议执行情况

| 日期 | 建议内容 | 是否执行 | 结果 |
|------|----------|----------|------|
| ... | ... | 是/否 | 对了/错了 |

### 执行了的建议
- 分析执行后的结果

### 没执行的建议
- 分析如果执行了会怎样
- 用户当时为什么没执行
- 这个决策对不对

## 三、持仓变化

| 标的 | 周初价 | 周末价 | 周涨跌 | 贡献 |
|------|--------|--------|--------|------|
| ... | ... | ... | ... | ... |

**本周组合收益**：+/-X.XX%

## 四、经验总结

### 做对了什么
- ...

### 做错了什么
- ...

### 需要改进的
- ...

## 五、下周关注

- ...

Monthly review content

# 月复盘 YYYY年MM月

## 一、本月交易汇总

### 买入操作
| 日期 | 标的 | 买入价 | 现价 | 浮盈/亏 | 决策质量 |
|------|------|--------|------|---------|----------|

### 卖出操作
| 日期 | 标的 | 卖出价 | 现价 | 卖对了吗 | 决策质量 |
|------|------|--------|------|----------|----------|

### 决策质量评分
- A：完美决策，时机和方向都对
- B：方向对，时机可以更好
- C：方向错，但及时纠正
- D：方向错，且没及时纠正

## 二、建议准确性统计

| 类型 | 总数 | 正确 | 错误 | 准确率 |
|------|------|------|------|--------|
| 止盈建议 | | | | |
| 止损建议 | | | | |
| 加仓建议 | | | | |
| 观望建议 | | | | |

### 错误建议分析
- 哪些建议是错的
- 错在哪里
- 如何改进分析逻辑

## 三、组合表现

### 收益统计
- 月初市值：XXX
- 月末市值：XXX
- 本月收益：+/-X.XX%
- 对比指数：沪深300 X.XX%，中证500 X.XX%

### 持仓变化
| 标的 | 月初仓位 | 月末仓位 | 变化 |
|------|----------|----------|------|

## 四、经验沉淀

### 本月学到的
1. ...
2. ...

### 需要形成的规则
（如果有值得固化的经验，建议添加到 Principles.md）
- ...

### 下月改进计划
1. ...
2. ...

## 五、准则更新建议

基于本月经验，建议添加/修改以下投资准则：
- ...

Step 4: Save the review report

  • Save the weekly review to:股市信息/Records/reviews/YYYY-WXX.md
  • Save the monthly review to:股市信息/Records/reviews/YYYY-MM.md

Step 5: Update investment principles (if applicable)

If experience worth consolidating has been distilled from the review, ask the user whether to add it to 股市信息/Config/Principles.md

Step 6: Update trades pending verification

In trades.md , update the verification results of verified trades.

Step 7: Update user insights (Insight.md)

Read and update 股市信息/Config/Insight.md：

  1. Update the growth trajectory：

    • Add any progress made in this period to the "progress record"
    • Add any newly discovered issues to the "items to be improved"
    • Update the number of occurrences of the issue

  2. Verify behavioral patterns：

    • Compare the weaknesses the user describes with actual performance
    • Update the confidence level of the "behavioral pattern observations"
    • Record newly discovered behavioral patterns

  3. Summarize adoption status：

    • Count the adoption rate of each type of recommendation in this period
    • Analyze the patterns of adoption/non-adoption
    • Verify the results of adopted recommendations (after 30 days/90 days)

  4. Distill key insights：

    • Record important findings from the review in "key insights"
    • Organize past insights and categorize them into the corresponding sections

30-day/90-day verification logic

For each trade, the system sets a verification time point when it is recorded. During the review, check:

  1. Find verifications that are due: Check which trades have reached the 30-day or 90-day verification point
  2. Get real prices: Get the current real price of the instrument
  3. Calculate results: Compare the trade price with the current price
  4. Judge right or wrong：
    • Sell action: price fell after selling = sold correctly, price rose after selling = sold incorrectly
    • Buy action: price rose after buying = bought correctly, price fell after buying = bought incorrectly
  5. Record judgment: Update the verification results in trades.md

Data authenticity assurance

  1. All prices must be obtained in real time: Obtain via AKShare, must not fabricate
  2. Clearly state when data cannot be obtained: If some data cannot be obtained, inform the user, do not guess
  3. Transparent calculation process: Show the return calculation process so the user can verify it
  4. Data source labeling: Label the data acquisition time and source in the report

Example output

User：/review

Assistant: Today is Saturday, let me do this week's review for you.

First, get the latest data...

Weekly Review 2026-W02 (01-06 ~ 01-10)

1. This week's trade review

You executed the following operations this week:

Date Operation Instrument Trade price Current price Change Preliminary judgment
01-08 Liquidate Consumer ETF 0.805 0.812 +0.87% Sold too early for now
01-09 Reduce position by 50% Hang Seng Internet 0.538 0.541 +0.56% Sold too early for now

(Note: Short-term price movements do not mean the decision was right or wrong; a longer period is needed for verification)

2. Implementation of recommendations

Date Recommendation Executed or not In hindsight
01-07 Liquidate Consumer ETF Yes ✓ Correct execution; the stop-loss logic holds
01-07 Reduce Industrial Metals by half No Good decision! It later rose 8%
01-08 Reduce Hang Seng Internet by half Yes ✓ To be verified

...

Data source: AKShare, acquisition time: 2026-01-11 10:00