vectorbt学习_45DMA之五滑窗网格参数优选

本文在上一篇文章(05dma_03rollingGridParamV2)面临问题
对于双均线，最终采用参数为(20，22),(30，35)这样的参数组合，显然不合理，有对行情进行拟合的嫌疑。
较合理的参数组合方式是，采用快线以及慢线相对快线的倍率。大致认为剔除二者相关性了（正交性）。
此时就无法借助run_combs创建组合计算指标了，需基于vbt创建新技术指标DualMA。

勘误：此篇文章部分截图可能有误，此文章的后继文章“DMA之六滑窗网格参数优选”修复此问题。请查阅后文。

01,基础配置信息#

1
#conda envs:vectorbt_env
2
import warnings
3
import vectorbt as vbt
4
import numpy as np
5
import pandas as pd
6
from datetime import datetime, timedelta
7
import pytz
8
from dateutil.parser import parse
9
import ipywidgets as widgets
10
from copy import deepcopy
11
from tqdm import tqdm
12
import imageio
13
from IPython import display
14
import plotly.graph_objects as go
15
import itertools
16
import dateparser
17
import gc
18
import math
19
from tools import dbtools
20

21
warnings.filterwarnings("ignore")
22

23
pd.set_option('display.max_rows',500)
24
pd.set_option('display.max_columns',500)
25
pd.set_option('display.width',1000)

02,行情获取和可视化#

a,时间交易参数配置#

1
# Enter your parameters here
2
seed = 42
3
symbol = '002594.XSHE'
4
metric = 'total_return'
5

6
start_date = datetime(2020, 1, 1, tzinfo=pytz.utc)  # time period for analysis, must be timezone-aware
7
end_date = datetime(2023,1,1, tzinfo=pytz.utc)
8
time_buffer = timedelta(days=100)  # buffer before to pre-calculate SMA/EMA, best to set to max window
9
freq = '1D'
10

11
vbt.settings.portfolio['init_cash'] = 10000.  # 100$
12
vbt.settings.portfolio['fees'] = 0.0025  # 0.25%
13
vbt.settings.portfolio['slippage'] = 0.0025  # 0.25%

b,获取行情和行情mask#

1
# Download data with time buffer
2
cols = ['Open', 'High', 'Low', 'Close', 'Volume']
3
# ohlcv_wbuf = vbt.YFData.download(symbol, start=start_date-time_buffer, end=end_date).get(cols)
4

5
ohlcv_wbuf=dbtools.MySQLData.download(symbol).get() # 自带工具类查询
6
assert(~ohlcv_wbuf.empty)
7
ohlcv_wbuf = ohlcv_wbuf.astype(np.float64)
8

9
print("ohlcv_wbuf.shape:",ohlcv_wbuf.shape)
10
print("ohlcv_wbuf.columns:",ohlcv_wbuf.columns)
11

12

13
# Create a copy of data without time buffer
14
wobuf_mask = (ohlcv_wbuf.index >= start_date) & (ohlcv_wbuf.index <= end_date) # mask without buffer
15

16
ohlcv = ohlcv_wbuf.loc[wobuf_mask, :]
17

18
print("ohlcv.shape:",ohlcv.shape)
19

20
# Plot the OHLC data
21
ohlcv.vbt.ohlcv.plot().show_svg() # 绘制蜡烛图
22
# remove show_svg() to display interactive chart!

1
ohlcv_wbuf.shape: (978, 5)
2
ohlcv_wbuf.columns: Index(['Open', 'High', 'Low', 'Close', 'Volume'], dtype='object')
3
ohlcv.shape: (728, 5)

svg

20,网格参数-指标计算和可视化#

仅可视化第一列

1
fast_windows = np.arange(10, 50,5)
2
slow_multis = np.arange(1.5, 5.5, 0.5)
3
print("fast_windows:",fast_windows)
4
print("slow_multis:",slow_multis)
5

6
price=ohlcv_wbuf['Close']
7
dualma = vbt.DualMA.run(price, fast_window=fast_windows,slow_multi=slow_multis,param_product=True)
8
dualma = dualma[wobuf_mask]
9
# there should be no nans after removing time buffer
10
assert(~dualma.fast_ma.isnull().any().any())
11
assert(~dualma.slow_ma.isnull().any().any())
12

13

14
print()
15
print('dualma.fast_ma.head(3)')
16
print(dualma.fast_ma.head(3))
17
print('dualma.slow_ma.head(3)')
18
print(dualma.slow_ma.head(3))
19

20
print()
21
fig = ohlcv['Close'].vbt.plot(trace_kwargs=dict(name='Price'))
22
fig = dualma.fast_ma.iloc[:,0].vbt.plot(trace_kwargs=dict(name="Fast MA col %s"%str(dualma.fast_ma.iloc[:,0].name)), fig=fig)
23
fig = dualma.slow_ma.iloc[:,0].vbt.plot(trace_kwargs=dict(name="Slow MA col %s"%str(dualma.slow_ma.iloc[:,0].name)), fig=fig)
24
fig.show_svg()

1
fast_windows: [10 15 20 25 30 35 40 45]
2
slow_multis: [1.5 2.  2.5 3.  3.5 4.  4.5 5. ]
3

4
dualma.fast_ma.head(3)
5
dualma_fast_window             10                                                                 15                                                                                    20                                                                      25                                                                        30                                                                                      35                                                                                    40                                                                        45
6
dualma_slow_multi             1.5     2.0     2.5     3.0     3.5     4.0     4.5     5.0        1.5        2.0        2.5        3.0        3.5        4.0        4.5        5.0      1.5      2.0      2.5      3.0      3.5      4.0      4.5      5.0      1.5      2.0      2.5      3.0      3.5      4.0      4.5      5.0        1.5        2.0        2.5        3.0        3.5        4.0        4.5        5.0        1.5        2.0        2.5        3.0        3.5        4.0        4.5        5.0      1.5      2.0      2.5      3.0      3.5      4.0      4.5      5.0        1.5        2.0        2.5        3.0        3.5        4.0        4.5        5.0
7
date
8
2020-01-02 00:00:00+00:00  46.665  46.665  46.665  46.665  46.665  46.665  46.665  46.665  45.824667  45.824667  45.824667  45.824667  45.824667  45.824667  45.824667  45.824667  45.3025  45.3025  45.3025  45.3025  45.3025  45.3025  45.3025  45.3025  44.9476  44.9476  44.9476  44.9476  44.9476  44.9476  44.9476  44.9476  44.816667  44.816667  44.816667  44.816667  44.816667  44.816667  44.816667  44.816667  44.594571  44.594571  44.594571  44.594571  44.594571  44.594571  44.594571  44.594571  44.5425  44.5425  44.5425  44.5425  44.5425  44.5425  44.5425  44.5425  44.440222  44.440222  44.440222  44.440222  44.440222  44.440222  44.440222  44.440222
9
2020-01-03 00:00:00+00:00  46.972  46.972  46.972  46.972  46.972  46.972  46.972  46.972  46.128667  46.128667  46.128667  46.128667  46.128667  46.128667  46.128667  46.128667  45.5025  45.5025  45.5025  45.5025  45.5025  45.5025  45.5025  45.5025  45.1420  45.1420  45.1420  45.1420  45.1420  45.1420  45.1420  45.1420  44.964000  44.964000  44.964000  44.964000  44.964000  44.964000  44.964000  44.964000  44.723714  44.723714  44.723714  44.723714  44.723714  44.723714  44.723714  44.723714  44.6265  44.6265  44.6265  44.6265  44.6265  44.6265  44.6265  44.6265  44.555556  44.555556  44.555556  44.555556  44.555556  44.555556  44.555556  44.555556
10
2020-01-06 00:00:00+00:00  47.138  47.138  47.138  47.138  47.138  47.138  47.138  47.138  46.456000  46.456000  46.456000  46.456000  46.456000  46.456000  46.456000  46.456000  45.7310  45.7310  45.7310  45.7310  45.7310  45.7310  45.7310  45.7310  45.3376  45.3376  45.3376  45.3376  45.3376  45.3376  45.3376  45.3376  45.112667  45.112667  45.112667  45.112667  45.112667  45.112667  45.112667  45.112667  44.871143  44.871143  44.871143  44.871143  44.871143  44.871143  44.871143  44.871143  44.7115  44.7115  44.7115  44.7115  44.7115  44.7115  44.7115  44.7115  44.660222  44.660222  44.660222  44.660222  44.660222  44.660222  44.660222  44.660222
11
dualma.slow_ma.head(3)
12
dualma_fast_window                10                                                                              15                                                                                      20                                                                                25                                                                                 30                                                                                    35                                                                                      40                                                                                   45
13
dualma_slow_multi                1.5      2.0      2.5        3.0        3.5      4.0        4.5      5.0        1.5        2.0        2.5        3.0        3.5        4.0        4.5        5.0        1.5      2.0      2.5        3.0        3.5        4.0        4.5      5.0        1.5      2.0        2.5        3.0        3.5      4.0        4.5       5.0        1.5        2.0        2.5        3.0        3.5        4.0        4.5      5.0        1.5        2.0        2.5        3.0        3.5        4.0        4.5        5.0        1.5        2.0      2.5        3.0        3.5        4.0        4.5       5.0        1.5        2.0        2.5        3.0        3.5        4.0        4.5        5.0
14
date
15
2020-01-02 00:00:00+00:00  45.824667  45.3025  44.9476  44.816667  44.594571  44.5425  44.440222  44.6384  45.180455  44.816667  44.545676  44.440222  44.717692  45.135167  45.513134  46.025200  44.816667  44.5425  44.6384  45.135167  45.697429  46.307750  46.683111  47.0983  44.545676  44.6384  45.235806  46.025200  46.560460  47.0983  47.997679  48.61136  44.440222  45.135167  46.025200  46.683111  47.425238  48.410917  48.769630  48.8484  44.717692  45.697429  46.560460  47.425238  48.496066  48.803714  48.852357  49.430914  45.135167  46.307750  47.0983  48.410917  48.803714  48.892313  49.622778  50.14240  45.513134  46.683111  47.997679  48.769630  48.852357  49.622778  50.162574  50.375822
16
2020-01-03 00:00:00+00:00  46.128667  45.5025  45.1420  44.964000  44.723714  44.6265  44.555556  44.6660  45.373636  44.964000  44.652162  44.555556  44.741538  45.119167  45.485821  45.984267  44.964000  44.6265  44.6660  45.119167  45.666714  46.291125  46.643333  47.0707  44.652162  44.6660  45.229677  45.984267  46.549080  47.0707  47.936429  48.56848  44.555556  45.119167  45.984267  46.643333  47.349905  48.362083  48.758074  48.8320  44.741538  45.666714  46.549080  47.349905  48.460984  48.784357  48.838471  49.366457  45.119167  46.291125  47.0707  48.362083  48.784357  48.878875  49.584500  50.12260  45.485821  46.643333  47.936429  48.758074  48.838471  49.584500  50.141139  50.379778
17
2020-01-06 00:00:00+00:00  46.456000  45.7310  45.3376  45.112667  44.871143  44.7115  44.660222  44.6908  45.562273  45.112667  44.787297  44.660222  44.773846  45.116667  45.474478  45.950800  45.112667  44.7115  44.6908  45.116667  45.641143  46.267875  46.621889  47.0449  44.787297  44.6908  45.232742  45.950800  46.534598  47.0449  47.864554  48.52880  44.660222  45.116667  45.950800  46.621889  47.278952  48.320667  48.743185  48.8232  44.773846  45.641143  46.534598  47.278952  48.406803  48.770500  48.833885  49.298743  45.116667  46.267875  47.0449  48.320667  48.770500  48.860063  49.552222  50.09115  45.474478  46.621889  47.864554  48.743185  48.833885  49.552222  50.122772  50.388044

svg

21,网格参数-信号计算和可视化#

仅可视化第一列

1
dmac_size.shape: (728, 64)
2

3
dmac_size.iloc[:3,:3]:
4
dualma_fast_window           10
5
dualma_slow_multi           1.5   2.0   2.5
6
date
7
2020-01-02 00:00:00+00:00  True  True  True
8
2020-01-03 00:00:00+00:00  True  True  True
9
2020-01-06 00:00:00+00:00  True  True  True

svg

1
Start                       2020-01-02 00:00:00+00:00
2
End                         2022-12-30 00:00:00+00:00
3
Period                                            728
4
Total                                       474.03125
5
Rate [%]                                    65.114183
6
First Index                 2020-01-15 16:52:30+00:00
7
Last Index                  2022-11-07 20:15:00+00:00
8
Norm Avg Index [-1, 1]                      -0.159967
9
Distance: Min                                     1.0
10
Distance: Max                               82.734375
11
Distance: Mean                               1.464916
12
Distance: Std                                5.175417
13
Total Partitions                             6.671875
14
Partition Rate [%]                           1.510978
15
Partition Length: Min                       41.671875
16
Partition Length: Max                      211.171875
17
Partition Length: Mean                     110.468174
18
Partition Length: Std                       78.523847
19
Partition Distance: Min                      26.78125
20
Partition Distance: Max                     82.734375
21
Partition Distance: Mean                    51.365493
22
Partition Distance: Std                     28.015768
23
Name: agg_func_mean, dtype: object

22,行情,信号的滑窗处理#

注意点：
01，训练集和验证集比例3：1，或者2：1，对应：window_len和set_lens为4<1>(或3<1>)，过大了历史包袱沉重，无法及时响应最新行情，过小了则容易参数跳变，形成类似过拟合效果

a,参数设置和效果预览#

代码中

1
#todo 这里是自然日计算的，但后面训练，验证集个数计算都完全正确，哪里应该和预想的不一致
2
合理的。实测bar_days= 60时
3

4
print(in_indexes[0][0])
5
print(in_indexes[1][0])
6
print(in_indexes[0][53:55])
7

8
2019-01-02 00:00:00+00:00
9
2019-03-25 00:00:00+00:00
10
DatetimeIndex(['2019-03-25 00:00:00+00:00', '2019-03-26 00:00:00+00:00'], dtype='datetime64[ns, UTC]', name='split_0', freq=None)
11
可见第二行第一个位于第一行第53个，不足设置的60,就是由于切分优先保证了数据的足量，但是数据间隔方面则可能有所重叠。

1
# 滚动周期参数设置和大致效果可视化
2
start_end_days=int((end_date-start_date).days) #todo 这里是自然日计算的，但后面训练，验证集个数计算都完全正确，哪里应该和预想的不一致
3
bar_days= 80         # 训练，验证集时间长度，以此为单位
4
test_bar_num=2      # 训练集时间长度
5
verify_bar_num=1    # 验证集时间长度
6
verify_overlap=0 # 验证集重叠时间长度
7
pre_test_days=0    # 由于测试集一部分时间用于计算指标，导致实际训练时间不足，这个是一定程度补充的days周期
8
# n取值需要满足:确保验证集合收尾相接
9
# => (n-1)*(verify_bar_num-verify_overlap)+(verify_bar_num+test_bar_num)=start_end_days/bar_days
10
# => n=(start_end_days/bar_days-test_bar_num-verify_overlap)/(verify_bar_num-verify_overlap)
11
calc_n=(start_end_days/bar_days-test_bar_num-verify_overlap)/(verify_bar_num-verify_overlap)
12

13

14
split_kwargs = dict(
15
    n=int(calc_n),
16
    window_len=int(bar_days*(test_bar_num+verify_bar_num)+pre_test_days),
17
    set_lens=(int(bar_days*verify_bar_num),),
18
    left_to_right=False
19
)  # 10 windows, each 2 years long, reserve 180 days for test
20
# 合理设置n，最好确保验证集，连续且无重复
21
pf_kwargs = dict(
22
    direction='both',  # long and short
23
    freq='d'
24
)
25
print('split_kwargs:',split_kwargs)
26

27
def roll_in_and_out_samples(price, **kwargs):
28
    return price.vbt.rolling_split(**kwargs)
29

30
# 验证：单列数据验证，橘黄色验证集连续且无重复
31
roll_in_and_out_samples(price, **split_kwargs, plot=True, trace_names=['in-sample', 'out-sample']).show_svg()

1
split_kwargs: {'n': 11, 'window_len': 240, 'set_lens': (80,), 'left_to_right': False}

svg

b,根据滑窗参数切分行情数据和信号#

1
in_price.shape: (160, 11)
2
out_price.shape: (80, 11)
3

4
in_price.index: RangeIndex(start=0, stop=160, step=1)
5
in_price.columns: Int64Index([0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10], dtype='int64', name='split_idx')
6

7
in_price[0:3]:
8
split_idx     0      1      2      3      4      5       6       7       8       9       10
9
0          49.17  58.15  51.20  43.39  48.15  97.90  167.98  239.52  202.00  251.77  253.14
10
1          48.06  56.16  49.50  43.15  49.73  96.55  164.08  225.00  214.11  252.50  266.49
11
2          50.65  55.36  50.29  43.79  52.25  94.50  168.03  208.99  227.02  246.86  266.08
12

13
###############################
14
in_dmac_size.shape: (160, 704)
15
out_dmac_size.shape: (80, 704)
16

17
in_dmac_size.iloc[:5,:5]:
18
split_idx              0
19
dualma_fast_window    10
20
dualma_slow_multi    1.5   2.0   2.5   3.0   3.5
21
0                   True  True  True  True  True
22
1                   True  True  True  True  True
23
2                   True  True  True  True  True
24
3                   True  True  True  True  True
25
4                   True  True  True  True  True

23,滑窗的收益数据计算#

a,持有参数收益#

在此区间，基础标的物表现

1
def simulate_holding(price, **kwargs):
2
    pf = vbt.Portfolio.from_holding(price, **kwargs)
3
    return pf.sharpe_ratio()
4

5
in_hold_sharpe = simulate_holding(in_price, **pf_kwargs)
6
print(in_hold_sharpe.head(5))
7

8
out_hold_sharpe = simulate_holding(out_price, **pf_kwargs)
9
print(out_hold_sharpe.head(5))

1
split_idx
2
0    0.235446
3
1   -1.630616
4
2    0.598889
5
3    2.647397
6
4    4.501923
7
Name: sharpe_ratio, dtype: float64
8
split_idx
9
0   -0.929956
10
1    2.065991
11
2    4.100300
12
3    4.801291
13
4    0.688785
14
Name: sharpe_ratio, dtype: float64

b,网格参数收益(训练集和验证集)#

1
in_sharpe.shape: (704,)
2
dualma_fast_window  dualma_slow_multi  split_idx
3
10                  1.5                0            0.235446
4
                    2.0                0            0.235446
5
                    2.5                0            0.235446
6
                    3.0                0            0.235446
7
                    3.5                0            0.235446
8
                                                      ...
9
45                  3.0                10           0.663486
10
                    3.5                10           0.663486
11
                    4.0                10           0.663486
12
                    4.5                10           0.663486
13
                    5.0                10           0.663486
14
Name: sharpe_ratio, Length: 704, dtype: float64
15

16
out_sharpe.shape: (704,)
17
dualma_fast_window  dualma_slow_multi  split_idx
18
10                  1.5                0           -0.929956
19
                    2.0                0           -0.820595
20
                    2.5                0           -0.820595
21
                    3.0                0           -0.820595
22
                    3.5                0           -0.820595
23
                                                      ...
24
45                  3.0                10          -0.554763
25
                    3.5                10          -0.554763
26
                    4.0                10          -0.554763
27
                    4.5                10          -0.554763
28
                    5.0                10          -0.554763
29
Name: sharpe_ratio, Length: 704, dtype: float64

c,训练集上的最佳参数用于验证集#

大致思路：
01,获取各split_idx的最佳收益(sharp_radio)的参数组合idxmax,也就是fast_window,slow_window,split_idx，三维索引元组
02,按照split_idx进行聚类，取得各split_idx对应的最佳参数。实际含义就是各滑动窗口的最佳参数

1
def get_best_index(performance, higher_better=True):
2
    if higher_better:
3
        return performance[performance.groupby('split_idx').idxmax()].index
4
    return performance[performance.groupby('split_idx').idxmin()].index
5
in_best_index = get_best_index(in_sharpe)
6

7
print('in_best_index[:5]')
8
print(in_best_index[:5])
9

10
# 参数走势可视化
11
def get_best_params(best_index, level_name):
12
    return best_index.get_level_values(level_name).to_numpy()
13
in_best_fast_windows = get_best_params(in_best_index, 'dualma_fast_window')
14
in_best_slow_multi = get_best_params(in_best_index, 'dualma_slow_multi')
15
in_best_slow_windows = in_best_fast_windows*in_best_slow_multi
16
in_best_window_pairs = np.array(list(zip(in_best_fast_windows, in_best_slow_windows)))
17
print()
18
print('in_best_window_pairs[:5][:]:')
19
print(in_best_window_pairs[:5][:])
20
pd.DataFrame(in_best_window_pairs, columns=['fast_window', 'slow_window']).vbt.plot().show_svg()

1
in_best_index[:5]
2
MultiIndex([(40, 5.0, 0),
3
            (10, 3.0, 1),
4
            (10, 1.5, 2),
5
            (10, 1.5, 3),
6
            (10, 1.5, 4)],
7
           names=['dualma_fast_window', 'dualma_slow_multi', 'split_idx'])
8

9
in_best_window_pairs[:5][:]:
10
[[ 40. 200.]
11
 [ 10.  30.]
12
 [ 10.  15.]
13
 [ 10.  15.]
14
 [ 10.  15.]]

svg

将滚动获取的最佳参数用于验证集，统计收益信息

1
in_best_index.shape: (11,)
2

3
in_best_index:
4
MultiIndex([(40, 5.0,  0),
5
            (10, 3.0,  1),
6
            (10, 1.5,  2),
7
            (10, 1.5,  3),
8
            (10, 1.5,  4),
9
            (10, 1.5,  5),
10
            (10, 1.5,  6),
11
            (45, 2.5,  7),
12
            (10, 1.5,  8),
13
            (25, 2.0,  9),
14
            (10, 2.5, 10)],
15
           names=['dualma_fast_window', 'dualma_slow_multi', 'split_idx'])
16

17
out_dmac_size.shape: (80, 704)
18

19
out_dmac_size_reindexed[in_best_index].shape: (80, 11)
20

21
dmac_pf_out.trades.records[:5]
22
   id  col        size  entry_idx  entry_price  entry_fees  exit_idx  exit_price  exit_fees           pnl    return  direction  status  parent_id
23
0   0    0  199.762836          0    49.934525   24.937656        79       46.85        0.0   -641.111119 -0.064271          0       0          0
24
1   1    1  222.599259          0    44.811750   24.937656        79       58.80        0.0   3088.836429  0.309656          0       0          1
25
2   2    2  182.338041          0    54.706425   24.937656        79       88.73        0.0   6178.854345  0.619430          0       0          2
26
3   3    3  114.462060          0    87.147325   24.937656        79      183.53        0.0  11007.221874  1.103474          0       0          3
27
4   4    4   59.581957          0   167.417500   24.937656        79      176.88        0.0    538.856616  0.054020          0       0          4
28

29
out_test_sharpe.head(5)
30
dualma_fast_window  dualma_slow_multi  split_idx
31
40                  5.0                0           -0.929956
32
10                  3.0                1            2.065991
33
                    1.5                2            4.100300
34
                                       3            4.801291
35
                                       4            0.688785
36
Name: sharpe_ratio, dtype: float64

24,sharp ratio的汇总可视化#

1
cv_results_df = pd.DataFrame({
2
    'in_sample_hold': in_hold_sharpe.values,
3
    'in_sample_median': in_sharpe.groupby('split_idx').median().values,
4
    'in_sample_best': in_sharpe[in_best_index].values,
5
    'out_sample_hold': out_hold_sharpe.values,
6
    'out_sample_median': out_sharpe.groupby('split_idx').median().values,
7
    'out_sample_test': out_test_sharpe.values
8
})
9

10
color_schema = vbt.settings['plotting']['color_schema']
11

12
cv_results_df.vbt.plot(
13
    trace_kwargs=[
14
        dict(line_color=color_schema['blue']),
15
        dict(line_color=color_schema['blue'], line_dash='dash'),
16
        dict(line_color=color_schema['blue'], line_dash='dot'),
17
        dict(line_color=color_schema['orange']),
18
        dict(line_color=color_schema['orange'], line_dash='dash'),
19
        dict(line_color=color_schema['orange'], line_dash='dot')
20
    ]
21
).show_svg()

svg

关注点：

蓝色部分正常排序是(从上到下)：点线，实现，线段，

橘色部分

实线对实线
说明测试集和验证集的周期收益情况，二者同时出现0轴同侧较好（同时上涨，同时下跌，保持行情的稳定性or延续性）

线段对线段
二者一方面随着各自颜色的实线趋势变化（受各自实线影响较大），其他应该无必然联系

点线对点线
蓝色点高于橘色点线，蓝色是训练集内最佳，橘色则是训练集得到最优参数用于验证集结果收益，大概率低于验证集。

测试，验证集时间长度差异，引入偏差
由于测试集一般是验证集的2-3倍（或更多），对于单边行情(假如上涨)，则(测试集的)实线收益。蓝色线大概率位于橘色线上方。
如果下跌，则相反。蓝色由于时间长，大概率位于橘色下方。

注意： 01，202406，对于当前case，y周取值为sharp ratio夏普比，而非收益率。所以数据点高低并不反映收益率。所以，以上结论需要稍斟酌，并不完全准确。

25,滚动回测收益可视化#

1
# 验证集：原始价格变动
2
out_price_org=out_price.iloc[-1, :]/out_price.iloc[0, :]
3
print('out_price_org shape:',out_price_org.shape)
4
print('out_price_org.head(5)')
5
print(out_price_org.head(5))
6

7
# 验证集：持有收益率
8
def simulate_holding(price, **kwargs):
9
    pf = vbt.Portfolio.from_holding(price, **kwargs)
10
    return pf.total_return()
11

12
out_hold_return = simulate_holding(out_price, **pf_kwargs)
13
print()
14
print('out_hold_return shape:',out_hold_return.shape)
15
print('out_hold_return.head(5) + 1')
16
print(out_hold_return.head(5)+1)
17

18

19
print()
20
print('out_test_return shape:',out_test_return.shape)
21
print('out_test_return.head(5) + 1')
22
print(out_test_return.head(5)+1)
23

24
cv_results_df = pd.DataFrame({
25
    'out_price_org':  out_price_org.cumprod(),
26
    'out_hold_return': (out_hold_return.values+1).cumprod(),
27
    'out_test_return': (out_test_return.values+1).cumprod()
28
})
29

30
color_dmac_pfschema = vbt.settings['plotting']['color_schema']
31

32

33
cv_results_df.vbt.plot(
34
    trace_kwargs=[
35
        dict(line_color=color_schema['blue']),
36
        dict(line_color=color_schema['blue'], line_dash='dash'),
37
        dict(line_color=color_schema['blue'], line_dash='dot')
38
    ]
39
).show_svg()

1
out_price_org shape: (11,)
2
out_price_org.head(5)
3
split_idx
4
0    0.940574
5
1    1.315436
6
2    1.625985
7
3    2.111239
8
4    1.059162
9
dtype: float64
10

11
out_hold_return shape: (11,)
12
out_hold_return.head(5) + 1
13
split_idx
14
0    0.935889
15
1    1.308884
16
2    1.617885
17
3    2.100722
18
4    1.053886
19
Name: total_return, dtype: float64
20

21
out_test_return shape: (11,)
22
out_test_return.head(5) + 1
23
dualma_fast_window  dualma_slow_multi  split_idx
24
40                  5.0                0            0.935889
25
10                  3.0                1            1.308884
26
                    1.5                2            1.617885
27
                                       3            2.100722
28
                                       4            1.053886
29
Name: total_return, dtype: float64

svg

可见，在上次降低技术指标的预热时间优化的基础上，整体收益进一步提升。改进后的策略避免了参数过拟合，挺高了鲁棒性，带来了收益改善。

26,计算正确性验证(略)#

1
a,准备校验数据，数据展示
2
b,行情->指标 计算正确
3
c,指标->信号 计算正确
4
d,信号->交易 计算正确

黄金矿工