pecu / financialvision Goto Github PK

I'd like to generate label8_eurusd_10bar_1500_500_val200_gaf_culr.pkl using code. But I check the content of label8_eurusd_10bar_1500_500_val200_gaf_culr.pkl, it is quite different.

Can you give us the way you serialized the data from
label8_eurusd_10bar_1500_500_val200_gaf_culr.pkl

Hi good afternoon, i am starting a thesis at the university about finance market. And i want to know if there is a jupyter or github where i can use the model the generates the data as a reference for my thesis.

Hi,

I am trying to use this project ("Encoding Candlesticks as Images for Patterns Classification Using Convolutional Neural Networks") to Patterns Classification it looks super cool, however I am not able to understand the label arr's.
Exemple:

ploted candles val_data:

  10.00 ┤        ╭╮
   9.00 ┤ ╭╮     ││
   8.00 ┤ ││   ╭╮││
   7.00 ┤ ││   ││││╭╮
   6.00 ┤ │╰╮  ││││││ ╭
   5.00 ┤ │ │ ╭╯╰╯│││╭╯
   4.00 ┤╭╯ │╭╯   ││││
   3.00 ┼╯  ││    ││││
   2.00 ┤   ╰╯    ╰╯╰╯
-----------------------------------
        0 1 2 3 4 5 6 7 8

val_label_arr

0 - [ 0, 1, 0, 0, 0, 0, 0, 0, 0 ]
1 - [ 0, 1, 0, 0, 0, 0, 0, 0, 0 ]
2 - [ 0, 1, 0, 0, 0, 0, 0, 0, 0 ]
3 - [ 0, 1, 0, 0, 0, 0, 0, 0, 0 ]
4 - [ 0, 0, 1, 0, 0, 0, 0, 0, 0 ]
5 - [ 0, 0, 1, 0, 0, 0, 0, 0, 0 ]
6 - [ 0, 0, 1, 0, 0, 0, 0, 0, 0 ]
7 - [ 0, 0, 0, 0, 1, 0, 0, 0, 0 ]
8 - [ 0, 0, 0, 0, 1, 0, 0, 0, 0 ]

In the val_label_arr there is an array[n, 9], in these 9 collumns what the value 1 means for each one of these collumns? And what's the relation with the val_data array?
Also, where the label8_eurusd_10bar_1500_500_val200_gaf_culr.pkl file come from, and how it's built?

Was there any live testing done on real financial markets?

Hi,

i try your code but i have some trouble in :

Deep Reinforcement Learning for Foreign Exchange Trading/trainer.py"

AttributeError: type object 'Agent' has no attribute 'from_spec'

Can you fix it please ?

Regards
Ludo.

I change the model to the code listed below,
`from utils import util_processor as pro

def get_model(params):
model = Sequential()

# Conv1
model.add(Conv2D(64, (5, 5), input_shape=(10, 10, 4), padding='same', strides=(1,1)))
model.add(Activation('sigmoid'))

# Conv2
model.add(Conv2D(64, (5, 5), padding='same', strides=(1,1)))
model.add(Activation('sigmoid'))

# FC
model.add(Flatten())
model.add(Dense(128, activation='relu'))

model.add(Dense(params['classes']))
model.add(Activation('softmax'))
model.summary()
return model

`

The result improves a lot!

Total results > Label 1 : 1171 / 1500 78.1%

Total results > Label 2 : 1024 / 1500 68.3%

Total results > Label 3 : 1221 / 1500 81.4%

Total results > Label 4 : 1454 / 1500 96.9%

Total results > Label 5 : 783 / 1500 52.2%

Total results > Label 6 : 1055 / 1500 70.3%

Total results > Label 7 : 1312 / 1500 87.5%

Total results > Label 8 : 1280 / 1500 85.3%

Is is possible to change the code for TF2 ?

Hi, I noticed you have released more recent work on this topic, ‘dynamic deep convolutional candle stick learner’ do you plan to add the code and dataset to this repo?