[go: up one dir, main page]
Scribd
Upload
80 views2 pages

ChatGPT Quant With Genetic Algorithm

This document defines functions for running a genetic algorithm to evolve a population of weight vectors and find optimal weights for a trading strategy. It initializes a population randomly, calculates fitness by measuring returns weighted by each vector, selects parents for reproduction based on fitness, performs crossover and mutation of offspring, and runs the full algorithm over multiple generations to optimize the weights. The algorithm is applied to historical returns data to find the best weighting of assets and evaluate strategy performance.

Uploaded by

lali
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
80 views2 pages

ChatGPT Quant With Genetic Algorithm

This document defines functions for running a genetic algorithm to evolve a population of weight vectors and find optimal weights for a trading strategy. It initializes a population randomly, calculates fitness by measuring returns weighted by each vector, selects parents for reproduction based on fitness, performs crossover and mutation of offspring, and runs the full algorithm over multiple generations to optimize the weights. The algorithm is applied to historical returns data to find the best weighting of assets and evaluate strategy performance.

Uploaded by

lali
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 2

import numpy as np

import pandas as pd
import random

def initialize_population(population_size, num_weights):


population = []
for i in range(population_size):
weights = np.random.rand(num_weights)
population.append(weights)
return population

def calculate_fitness(weights, returns):


strategy_returns = returns.dot(weights)
fitness = np.prod(strategy_returns + 1) - 1
return fitness

def select_parents(population, returns, num_parents):


fitness_values = []
for i in range(len(population)):
fitness_values.append(calculate_fitness(population[i], returns))
sorted_indices = np.argsort(fitness_values)[::-1]
parents = [population[sorted_indices[i]] for i in range(num_parents)]
return parents

def crossover(parents, num_offspring):


offspring = []
for i in range(num_offspring):
parent1 = parents[i % len(parents)]
parent2 = parents[(i + 1) % len(parents)]
offspring.append(np.mean([parent1, parent2], axis=0))
return offspring

def mutate(offspring, mutation_probability):


for i in range(len(offspring)):
for j in range(len(offspring[i])):
if np.random.rand() < mutation_probability:
offspring[i][j] = np.random.rand()
return offspring

def evolve(population, returns, num_parents, num_offspring, mutation_probability):


parents = select_parents(population, returns, num_parents)
offspring = crossover(parents, num_offspring)
offspring = mutate(offspring, mutation_probability)
new_population = parents + offspring
return new_population

def run_genetic_algorithm(returns, population_size, num_parents, num_offspring,


mutation_probability, num_generations):
num_weights = returns.shape[1]
population = initialize_population(population_size, num_weights)
for i in range(num_generations):
population = evolve(population, returns, num_parents, num_offspring, mutation_probability)
return population[0]

# Load the data


data = pd.read_csv("data.csv")

# Calculate returns
returns = data.pct_change().dropna().values

# Run the genetic algorithm


weights = run_genetic_algorithm(returns, population_size=100, num_parents=20,
num_offspring=80, mutation_probability=0.1, num_generations=100)

# Calculate the strategy returns


strategy_returns = returns.dot(weights)

# Plot the strategy returns


plt.plot(np.cumprod(strategy_returns + 1) - 1)
plt.show()

You might also like