ChatGPT：你的专属A股量化金融工程师——手把手教你用Prompt驾驭AI大模型

•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
< li>
•  
•  
•  
•  
•  
•  
•  
•  
•  

def calculate_factors(data): """
计算选股因子：市值(MV)、账面市值比(BM)、过去一年的收益率(PR) """ #
计算市值(MV) data['MV'] =
data['close'] * data['total_share']  #
计算账面市值比(BM) data['BM'] = data['total_assets'] / data['MV']  #
计算过去一年的收益率(PR) data['PR'] = data.groupby('ts_code')['close'].pct_change(12)
 return data
def
construct_factor(data): """
构建组合因子
""" #
按市值(MV)排序，划分为Small和Big
data.sort_values(by='MV',
inplace=True) data['Size_Group'] = np.where(data['MV']
<= data['MV'].median(), 'Small', 'Big')  #
在每个市值组内，根据BM和PR分别划分为Low、Medium、High data['BM_Group'] = pd.qcut(data['BM'],
q=3, labels=['Low', 'Medium', 'High']) data['PR_Group'] = pd.qcut(data['PR'],
q=3, labels=['Low', 'Medium', 'High'])  #
计算组合因子 small_low = data[(data['Size_Group'] == 'Small') & (data['BM_Group'] == 'Low') & (data['PR_Group'] == 'Low')]['MV'].sum() small_medium = data[(data['Size_Group'] == 'Small') & (data['BM_Group'] == 'Medium') & (data['PR_Group'] == 'Medium')]['MV'].sum() small_high = data[(data['Size_Group'] == 'Small') & (data['BM_Group'] == 'High') & (data['PR_Group'] == 'High')]['MV'].sum() big_low = data[(data['Size_Group'] == 'Big') & (data['BM_Group'] == 'Low') & (data['PR_Group'] == 'Low')]['MV'].sum() big_medium = data[(data['Size_Group'] == 'Big') & (data['BM_Group'] == 'Medium') & (data['PR_Group'] == 'Medium')]['MV'].sum() big_high = data[(data['Size_Group'] == 'Big') & (data['BM_Group'] == 'High') & (data['PR_Group'] == 'High')]['MV'].sum()  factor =
(1/3) * (small_low + small_medium + small_high) - (1/3) * (big_low + big_medium +
big_high)
 return factor

•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
< li>
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  
•  

def calculate_macd(data, short_window=12, long_window=26, signal_window=9): """
计算MACD指标
""" # 计算短期EMA short_ema = data['close'].ewm(span=short_window, min_periods=1, adjust=False).mean() #
计算长期EMA long_ema =
data['close'].ewm(span=long_window,
min_periods=1, adjust=False).mean()  #
计算MACD macd =
short_ema - long_ema
 # 计算Signal signal = macd.ewm(span=signal_window, min_periods=1, adjust=False).mean()  return macd,
signal
def
calculate_factors(data): """
计算Fama-French因子
""" # 假设计算Fama-French因子的方法 #
这里用随机数代替
 np.random.seed(0) data['Fama_French'] =
np.random.rand(len(data))
 return data
def
select_candidates(data, n=50): """
选取MACD因子值最高的股票作为买入候选
""" top_candidates =
data.nlargest(n, 'MACD') return
top_candidates
def
select_final_targets(data, n=10): """
在买入候选中，选取Fama-French因子值最高的股票作为最终买入标的 """
final_targets = data.nlargest(n, 'Fama_French') return
final_targets