Market-Direction-Predictor 0.1.0__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
|
@@ -0,0 +1,48 @@
|
|
|
1
|
+
Metadata-Version: 2.4
|
|
2
|
+
Name: Market-Direction-Predictor
|
|
3
|
+
Version: 0.1.0
|
|
4
|
+
Summary: simple python code for predicting markets
|
|
5
|
+
Author: navidpgg
|
|
6
|
+
Description-Content-Type: text/markdown
|
|
7
|
+
Requires-Dist: scikit-learn
|
|
8
|
+
Requires-Dist: numpy
|
|
9
|
+
Dynamic: author
|
|
10
|
+
Dynamic: description
|
|
11
|
+
Dynamic: description-content-type
|
|
12
|
+
Dynamic: requires-dist
|
|
13
|
+
Dynamic: summary
|
|
14
|
+
|
|
15
|
+
# Market Direction Predictor (MDP)
|
|
16
|
+
predict UP or DOWN days(candles) based on previous sources/indicators on a crypto/forex... markets using neural networks
|
|
17
|
+
## why i should use it when there are other libraries for this task?
|
|
18
|
+
well...
|
|
19
|
+
1. the code is simple,understandable and easy to use.
|
|
20
|
+
2. use this if you don't want a headache on figuring out how to use neural network for market prediction or even reading the documentation page
|
|
21
|
+
## will this 100% work.
|
|
22
|
+
IDK. the markets no matter crypto,stock,forex... have non stationary behaviour and mostly influenced by economics,news... but if i were to make a code that trades markets i would have 2 choices:
|
|
23
|
+
1. is to make a very complicated program that includes many inputs. economics,news,candlestick... which is very difficult to make
|
|
24
|
+
2. use a simple method to trade the market like the one written in this code which only trains a neural network on candlesticks (%changes of sources) to predict next candle direction. (BTW i wouldn't suggest you to make a deep neural network it will easily overfit the data and perform poorly. just need a nn that learns the general pattern this model worked fine but don't take my word `MLPClassifier((8,8),"tanh",solver="lbfgs",max_iter=2000)`)
|
|
25
|
+
|
|
26
|
+
so to answer your question, no it won't.
|
|
27
|
+
|
|
28
|
+
## so how do i use it ?
|
|
29
|
+
the library is a single file called `mdp.py` and
|
|
30
|
+
i've put some tests on the file too. this is the simples way of using the code:
|
|
31
|
+
```python
|
|
32
|
+
d=[1,2,3 ,2,3,4 ,3,4,5 ,4,5,6] # this is your closing data provided by the Exchange
|
|
33
|
+
mdpc=MDPC(wl=2,clf=MLPClassifier()) # wl stands for window length and it controls how far the model sees the past bars to predict next
|
|
34
|
+
ac,cm,npp=mdpc.train_predict(d) # ac:accuracy(number),cm:confusion table(2x2 array),npp:list of 2 numbers first one is the probability in which model thinks it is going down and UP for the second one.
|
|
35
|
+
print(ac)
|
|
36
|
+
print(cm)
|
|
37
|
+
print(npp)
|
|
38
|
+
assert npp[0]>npp[1]
|
|
39
|
+
```
|
|
40
|
+
there are 3 classes. `MDPC,MDPHLC,MDPMORE`. MDPHLC accepts HLC and MDPMORE accepts infinity of features (note that all of them compute %change on the data)
|
|
41
|
+
|
|
42
|
+
|
|
43
|
+
|
|
44
|
+
|
|
45
|
+
|
|
46
|
+
|
|
47
|
+
|
|
48
|
+
btw my github is https://github.com/navidpgg
|
|
@@ -0,0 +1,4 @@
|
|
|
1
|
+
market_direction_predictor-0.1.0.dist-info/METADATA,sha256=mwSnOx_cNlESSkHoA_61-zLIA9mBJKX_he3nI7d9idw,2507
|
|
2
|
+
market_direction_predictor-0.1.0.dist-info/WHEEL,sha256=lTU6B6eIfYoiQJTZNc-fyaR6BpL6ehTzU3xGYxn2n8k,91
|
|
3
|
+
market_direction_predictor-0.1.0.dist-info/top_level.txt,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
|
|
4
|
+
market_direction_predictor-0.1.0.dist-info/RECORD,,
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
|