基于路径依赖的Python股票池每周动态维持模拟需求

Hey there! Let's walk through how to implement this stock portfolio simulation exactly as you've outlined. Here's a structured breakdown with actionable steps and example logic:

First, let's align on the ground rules we're working with:

• 6 years of weekly stock data, with ~1000 stocks per week (slight variance)
• Start at t0 with a random selection of 75 stocks
• Each subsequent week t_n:
  • Some stocks exit the portfolio with probability p (due to structural reasons like delisting/merger)
  • Rebalance the portfolio to always maintain exactly 75 stocks by replacing exited stocks with new picks from the week's non-portfolio stocks

Step 1: Initialize the Portfolio at t0 #

• Pull the full list of stocks available in the first week (t0)
• Use a true random sampling method to pick 75 distinct stocks from this pool. Avoid any bias (e.g., don't favor large-cap or high-volume stocks unless specified)

Step 2: Weekly Exit Logic #

For each stock in your current 75-stock portfolio:

• Independently check if it exits: generate a random number between 0 and 1, and if it's less than p, mark the stock for removal
• Optional edge case handling: If a stock no longer exists in the current week's full stock pool (e.g., it delisted outside the probability-based exit), automatically mark it as exited
• Count the total number of exiting stocks (exit_count), which will range from 0 to 75

Step 3: Rebalance to 75 Stocks #

• Calculate how many stocks you need to add: need_to_add = exit_count (since we start with 75, subtract the remaining after exits to get the gap)
• Filter the current week's full stock pool to get available candidates: all stocks not already in your current portfolio
• Randomly sample need_to_add stocks from this available pool and add them to your portfolio
• Verify the final portfolio size is exactly 75 to catch any edge cases (e.g., if the available pool is unexpectedly small)

Key Edge Cases to Anticipate #

• Insufficient available stocks: While your setup says ~1000 stocks per week (so 925+ available after excluding the portfolio), in rare cases you might hit a week with fewer available stocks than needed. If this happens, you could:
  • Adjust the exit probability dynamically for that week
  • Prioritize adding stocks that have been present in recent weeks (to avoid one-off new listings)
• Repeat selections: Ensure your random sampling doesn't pick the same stock multiple times in one rebalance (most standard random sample functions handle this automatically)
• Consistent randomness: If you need reproducible results, set a fixed random seed at the start of the simulation

Example Pseudocode (Python) #

Here's a concrete snippet to illustrate the logic:

import random

# Prepped data: Dictionary mapping each week number to its list of stocks
# (Replace with your actual 6-year weekly stock data)
total_weeks = 52 * 6
all_stocks_by_week = {
    week: [f"stock_{i}" for i in range(random.randint(950, 1050))]
    for week in range(total_weeks)
}

# Step 1: Initialize t0 portfolio
initial_stocks = all_stocks_by_week[0]
current_portfolio = random.sample(initial_stocks, 75)
exit_probability = 0.02  # Replace with your desired probability p

# Step 2 & 3: Iterate through each subsequent week
for week in range(1, total_weeks):
    current_week_stocks = all_stocks_by_week[week]
    
    # Determine exiting stocks
    exiting_stocks = []
    for stock in current_portfolio:
        # Auto-exit if stock isn't in the current week's pool, plus probability-based exit
        if stock not in current_week_stocks or random.random() < exit_probability:
            exiting_stocks.append(stock)
    
    # Remove exited stocks from portfolio
    current_portfolio = [s for s in current_portfolio if s not in exiting_stocks]
    
    # Calculate how many to add and select candidates
    need_to_add = 75 - len(current_portfolio)
    available_stocks = [s for s in current_week_stocks if s not in current_portfolio]
    added_stocks = random.sample(available_stocks, need_to_add)
    
    # Update portfolio
    current_portfolio.extend(added_stocks)
    
    # Sanity check
    assert len(current_portfolio) == 75, f"Week {week}: Portfolio size mismatch! Expected 75, got {len(current_portfolio)}"

内容的提问来源于stack exchange，提问作者Martin Richter