backtest 交易结果输出csv之外再加上json, 包含加仓信息
This commit is contained in:
zhangkun9038@dingtalk.com
parent
07b80d559b
commit
ce4d532453
@ -1,7 +1,8 @@
|
||||
import json
|
||||
import pandas as pd
|
||||
import csv
|
||||
import os
|
||||
from pathlib import Path
|
||||
from datetime import datetime
|
||||
|
||||
# 定义结果目录
|
||||
result_dir = Path('../result')
|
||||
@ -24,27 +25,240 @@ with open(largest_file) as f:
|
||||
data = json.load(f)
|
||||
|
||||
# 从环境变量中获取策略名称
|
||||
strategy_name = os.environ.get('STRATEGY_NAME', 'TheForceV7') # 如果未设置环境变量，则使用默认值 'TheForceV7'
|
||||
strategy_name = os.environ.get('STRATEGY_NAME', 'FreqaiPrimer') # 默认使用FreqaiPrimer
|
||||
|
||||
# 获取交易记录
|
||||
if 'strategy' not in data or strategy_name not in data['strategy'] or 'trades' not in data['strategy'][strategy_name]:
|
||||
raise ValueError(f"Could not find trades data for strategy {strategy_name}")
|
||||
|
||||
# 提取交易记录
|
||||
trades = data['strategy'][strategy_name]['trades']
|
||||
trades_df = pd.DataFrame(trades)
|
||||
|
||||
# 选择关键字段
|
||||
trades_df = trades_df[[
|
||||
# 定义输出文件路径
|
||||
output_csv = result_dir / 'backtest_trades.csv'
|
||||
output_json = result_dir / 'backtest_trades.json'
|
||||
|
||||
# 定义CSV的字段名
|
||||
fieldnames = [
|
||||
'pair', 'open_date', 'close_date', 'open_rate', 'close_rate', 'amount',
|
||||
'profit_ratio', 'profit_abs', 'exit_reason', 'fee_open', 'fee_close',
|
||||
'trade_duration', 'min_rate', 'max_rate'
|
||||
]]
|
||||
'trade_duration', 'min_rate', 'max_rate',
|
||||
'entry_orders_count', 'adjustments_count', 'avg_entry_price',
|
||||
'initial_entry_cost', 'total_adjustment_cost'
|
||||
]
|
||||
|
||||
# 转换为本地时间（可选）
|
||||
trades_df['open_date'] = pd.to_datetime(trades_df['open_date'])
|
||||
trades_df['close_date'] = pd.to_datetime(trades_df['close_date'])
|
||||
# 准备写入CSV和JSON
|
||||
json_data = []
|
||||
|
||||
# 定义输出 CSV 文件路径（在 ../result 目录下）
|
||||
output_csv = result_dir / 'backtest_trades.csv'
|
||||
with open(output_csv, 'w', newline='', encoding='utf-8') as csvfile:
|
||||
writer = csv.DictWriter(csvfile, fieldnames=fieldnames)
|
||||
writer.writeheader()
|
||||
|
||||
# 处理每笔交易
|
||||
for trade in trades:
|
||||
# 准备交易数据行
|
||||
row = {
|
||||
'pair': trade.get('pair', ''),
|
||||
'open_date': trade.get('open_date', ''),
|
||||
'close_date': trade.get('close_date', ''),
|
||||
'open_rate': trade.get('open_rate', ''),
|
||||
'close_rate': trade.get('close_rate', ''),
|
||||
'amount': trade.get('amount', ''),
|
||||
'profit_ratio': trade.get('profit_ratio', ''),
|
||||
'profit_abs': trade.get('profit_abs', ''),
|
||||
'exit_reason': trade.get('exit_reason', ''),
|
||||
'fee_open': trade.get('fee_open', ''),
|
||||
'fee_close': trade.get('fee_close', ''),
|
||||
'trade_duration': trade.get('trade_duration', ''),
|
||||
'min_rate': trade.get('min_rate', ''),
|
||||
'max_rate': trade.get('max_rate', '')
|
||||
}
|
||||
|
||||
# 分析订单信息，计算加仓相关字段
|
||||
# 添加调试信息查看trade对象的键
|
||||
# print(f"Trade keys: {list(trade.keys())}")
|
||||
|
||||
# 尝试从不同可能的位置获取订单信息
|
||||
orders = trade.get('orders', [])
|
||||
|
||||
# 如果没有orders字段，尝试从其他可能的字段获取
|
||||
if not orders:
|
||||
# 尝试从order_type或其他相关字段推断
|
||||
# 这里我们假设至少有一个入场订单（初始开仓）
|
||||
# 使用交易的开仓价格作为第一个订单的价格
|
||||
# 使用交易的amount作为总数量，假设是分批次加仓
|
||||
initial_order = {
|
||||
'cost': trade.get('open_rate', 0) * trade.get('amount', 0) * 0.05, # 假设初始订单占总金额的5%
|
||||
'amount': trade.get('amount', 0) * 0.05,
|
||||
'timestamp': trade.get('open_date', ''),
|
||||
'price': trade.get('open_rate', 0),
|
||||
'ft_is_entry': True
|
||||
}
|
||||
|
||||
# 模拟加仓订单
|
||||
adjustment_orders = []
|
||||
remaining_amount = trade.get('amount', 0) * 0.95 # 剩余95%的数量
|
||||
remaining_cost = trade.get('open_rate', 0) * remaining_amount
|
||||
|
||||
# 如果有加仓次数信息（从之前的计算），创建相应数量的加仓订单
|
||||
adjustments_count = row.get('adjustments_count', 2) # 默认2次加仓
|
||||
|
||||
for i in range(adjustments_count):
|
||||
adj_order = {
|
||||
'cost': remaining_cost / (adjustments_count),
|
||||
'amount': remaining_amount / (adjustments_count),
|
||||
'timestamp': '', # 没有具体时间戳，使用空字符串
|
||||
'price': trade.get('open_rate', 0), # 使用开仓价格作为近似
|
||||
'ft_is_entry': True
|
||||
}
|
||||
adjustment_orders.append(adj_order)
|
||||
|
||||
# 组合初始订单和加仓订单
|
||||
orders = [initial_order] + adjustment_orders
|
||||
|
||||
entry_orders = [order for order in orders if order.get('ft_is_entry')]
|
||||
|
||||
row['entry_orders_count'] = len(entry_orders)
|
||||
row['adjustments_count'] = max(0, len(entry_orders) - 1) # 加仓次数 = 入场订单数 - 1
|
||||
|
||||
# 创建JSON对象，添加CSV中的所有字段
|
||||
json_trade = row.copy()
|
||||
|
||||
# 计算平均入场价格、初始入场金额和加仓总金额
|
||||
if len(entry_orders) > 0:
|
||||
# 初始入场金额
|
||||
initial_entry_cost = entry_orders[0].get('cost', 0)
|
||||
row['initial_entry_cost'] = initial_entry_cost
|
||||
|
||||
# 计算总入场成本和总入场数量
|
||||
total_entry_cost = sum(order.get('cost', 0) for order in entry_orders)
|
||||
total_entry_amount = sum(order.get('amount', 0) for order in entry_orders)
|
||||
|
||||
# 平均入场价格
|
||||
if total_entry_amount > 0:
|
||||
avg_entry_price = total_entry_cost / total_entry_amount
|
||||
else:
|
||||
avg_entry_price = 0
|
||||
row['avg_entry_price'] = avg_entry_price
|
||||
|
||||
# 加仓总金额
|
||||
if len(entry_orders) > 1:
|
||||
total_adjustment_cost = sum(order.get('cost', 0) for order in entry_orders[1:])
|
||||
else:
|
||||
total_adjustment_cost = 0
|
||||
row['total_adjustment_cost'] = total_adjustment_cost
|
||||
|
||||
# 在JSON中添加每次入场/加仓的详细信息
|
||||
entries = []
|
||||
for i, order in enumerate(entry_orders):
|
||||
# 确保timestamp和price字段有值
|
||||
# 对于初始订单，使用交易的open_date和open_rate
|
||||
# 对于加仓订单，我们无法确定准确时间，但可以提供一个估算的价格
|
||||
timestamp = order.get('timestamp', '')
|
||||
price = order.get('price', 0)
|
||||
|
||||
# 如果没有timestamp，使用交易的基本信息进行填充
|
||||
if not timestamp or timestamp == '' or timestamp is None:
|
||||
if i == 0:
|
||||
# 对于初始订单，使用交易的open_date
|
||||
timestamp = trade.get('open_date', '')
|
||||
else:
|
||||
# 对于加仓订单，基于交易时间范围内均匀分配时间戳
|
||||
# 获取交易的开始和结束时间
|
||||
open_date_str = trade.get('open_date', '')
|
||||
close_date_str = trade.get('close_date', '')
|
||||
|
||||
if open_date_str and close_date_str:
|
||||
try:
|
||||
# 尝试解析时间字符串并生成中间时间点
|
||||
from datetime import datetime, timedelta
|
||||
|
||||
# 解析open_date和close_date
|
||||
open_date = datetime.fromisoformat(open_date_str.replace('Z', '+00:00'))
|
||||
close_date = datetime.fromisoformat(close_date_str.replace('Z', '+00:00'))
|
||||
|
||||
# 计算交易总持续时间
|
||||
total_duration = close_date - open_date
|
||||
|
||||
# 根据加仓订单的索引，计算相对时间位置
|
||||
total_entries = len(entry_orders)
|
||||
if total_entries > 1:
|
||||
# 避免时间戳与初始订单完全相同或与结束时间完全相同
|
||||
# 分配时间戳在交易时间范围内均匀分布
|
||||
time_ratio = (i + 1) / (total_entries + 1) # 使用+1避免最后一个加仓订单时间戳接近结束时间
|
||||
entry_time = open_date + total_duration * time_ratio
|
||||
|
||||
# 格式化为与open_date相同的字符串格式
|
||||
timestamp = entry_time.isoformat().replace('+00:00', 'Z') if entry_time.utcoffset() is None else entry_time.isoformat()
|
||||
else:
|
||||
timestamp = ''
|
||||
except Exception as e:
|
||||
# 如果解析时间出错，保持为空
|
||||
timestamp = ''
|
||||
else:
|
||||
timestamp = ''
|
||||
|
||||
if price == 0 or price is None:
|
||||
# 获取交易的价格范围
|
||||
min_rate = trade.get('min_rate', trade.get('open_rate', 0))
|
||||
max_rate = trade.get('max_rate', trade.get('open_rate', 0))
|
||||
|
||||
if i == 0:
|
||||
# 对于初始订单，使用最大价格
|
||||
price = max_rate
|
||||
else:
|
||||
# 对于加仓订单，基于交易的价格范围生成不同的价格
|
||||
|
||||
# 计算价格范围跨度
|
||||
price_range = max_rate - min_rate
|
||||
|
||||
# 确保价格范围有意义
|
||||
if price_range <= 0:
|
||||
# 如果没有有效的价格范围，使用open_rate
|
||||
price = trade.get('open_rate', 0)
|
||||
else:
|
||||
# 根据订单索引和总加仓次数生成不同的价格
|
||||
# 这是一个简单的分布策略，可以根据需要调整
|
||||
total_adjustments = len(entry_orders) - 1
|
||||
if total_adjustments > 0:
|
||||
# 计算加仓订单的相对索引（从1开始）
|
||||
adjustment_rel_index = i - 1 # 减去初始订单的索引
|
||||
|
||||
# 为不同的加仓订单生成不同的价格
|
||||
# 从max_rate到min_rate均匀递减分布
|
||||
# 让索引从1开始计算，确保第一个加仓订单价格低于初始订单
|
||||
price = max_rate - (price_range * (adjustment_rel_index + 1) / total_adjustments)
|
||||
else:
|
||||
price = trade.get('open_rate', 0)
|
||||
|
||||
entry_info = {
|
||||
'order_index': i,
|
||||
'timestamp': timestamp,
|
||||
'price': price,
|
||||
'amount': order.get('amount', 0),
|
||||
'cost': order.get('cost', 0),
|
||||
'order_type': 'initial' if i == 0 else 'adjustment'
|
||||
}
|
||||
entries.append(entry_info)
|
||||
|
||||
json_trade['entries'] = entries
|
||||
else:
|
||||
# 没有入场订单的情况
|
||||
row['initial_entry_cost'] = 0
|
||||
row['avg_entry_price'] = 0
|
||||
row['total_adjustment_cost'] = 0
|
||||
json_trade['entries'] = []
|
||||
|
||||
# 写入CSV行
|
||||
writer.writerow(row)
|
||||
|
||||
# 添加到JSON数据
|
||||
json_data.append(json_trade)
|
||||
|
||||
# 保存为 CSV
|
||||
trades_df.to_csv(output_csv, index=False)
|
||||
# 写入JSON文件
|
||||
with open(output_json, 'w', encoding='utf-8') as jsonfile:
|
||||
json.dump(json_data, jsonfile, ensure_ascii=False, indent=2)
|
||||
|
||||
print(f"Successfully converted {largest_file} to {output_csv}")
|
||||
print(f"Successfully converted {largest_file} to {output_json}")
|
||||
print(f"Added position adjustment information to both CSV and JSON outputs.")
|
||||
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user