立方体参数优化
构建参数搜索空间
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
| # Define hyper-parameter space
# 49 fast x 49 slow x 19 signal
fast_windows, slow_windows, signal_windows = vbt.utils.params.create_param_combs(
(product, (combinations, np.arange(2, 51, 1), 2), np.arange(2, 21, 1)))
pd.DataFrame({"fast_windows":fast_windows, "slow_windows":slow_windows, "signal_windows":signal_windows})
Out[6]:
fast_windows slow_windows signal_windows
0 2 3 2
1 2 3 3
2 2 3 4
3 2 3 5
4 2 3 6
... ... ... ...
22339 49 50 16
22340 49 50 17
22341 49 50 18
22342 49 50 19
22343 49 50 20
|
计算指标和信号
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
| # Run MACD indicator
macd_ind = vbt.MACD.run(
price,
fast_window=fast_windows,
slow_window=slow_windows,
signal_window=signal_windows
)
# Long when MACD is above zero AND signal
entries = macd_ind.macd_above(0) & macd_ind.macd_above(macd_ind.signal)
# Short when MACD is below zero OR signal
exits = macd_ind.macd_below(0) | macd_ind.macd_below(macd_ind.signal)
exits
Out[16]:
macd_fast_window 2 ... 49
macd_slow_window 3 ... 50
macd_signal_window 2 3 ... 19 20
split_idx 0 1 2 0 ... 2 0 1 2
0 False False False False ... False False False False
1 False False False False ... False False False False
2 False False False False ... False False False False
3 True True True True ... False False False False
4 True False True True ... False False False False
5 True False True True ... False False False False
6 True True False True ... False False False False
7 False False False False ... False False False False
8 False True True False ... False False False False
9 True False True True ... False False False False
|
可见这里的信号为持续性信号，不同与常规的crossxx信号
组合收益评估
1
2
3
| # Build portfolio
pf = vbt.Portfolio.from_signals(
price.vbt.tile(len(fast_windows)), entries, exits, fees=0.001, freq='1D')
|
原始的1列price，经过range_split后变为3列，但是依然和entries列数不同，所以需要tile对齐列数据。
1
2
3
4
5
6
| price.shape
Out[26]: (16, 3)
len(fast_windows)
Out[27]: 22344
entries.shape
Out[28]: (16, 67032)
|
tile函数效果
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
| Out[21]:price.head()
split_idx 0 1 2
0 0.374540 0.304242 0.065052
1 0.950714 0.524756 0.948886
2 0.731994 0.431945 0.965632
3 0.598658 0.291229 0.808397
4 0.156019 0.611853 0.304614
price.vbt.tile(2)
Out[22]:
split_idx 0 1 2 0 1 2
0 0.374540 0.304242 0.065052 0.374540 0.304242 0.065052
1 0.950714 0.524756 0.948886 0.950714 0.524756 0.948886
2 0.731994 0.431945 0.965632 0.731994 0.431945 0.965632
3 0.598658 0.291229 0.808397 0.598658 0.291229 0.808397
4 0.156019 0.611853 0.304614 0.156019 0.611853 0.304614
5 0.155995 0.139494 0.097672 0.155995 0.139494 0.097672
|
可视化
1
2
3
4
5
6
7
8
9
10
11
12
13
14
| # Draw all window combinations as a 3D volume
fig = pf.total_return().vbt.volume(
x_level='macd_fast_window',
y_level='macd_slow_window',
z_level='macd_signal_window',
slider_level='split_idx',
trace_kwargs=dict(
colorbar=dict(
title='Total return',
tickformat='%'
)
)
)
fig.show()
|
会导致电脑卡顿，效果上也没觉得多么直观。
