Time series classification for stock market

I am looking a solution for predicting if price will go up / down on stock market.

Is it better to use PyTorch or GitHub - timeseriesAI/tsai: Time series Timeseries Deep Learning Machine Learning Pytorch fastai | State-of-the-art Deep Learning library for Time Series and Sequences in Pytorch / fastai or Time Series Made Easy in Python — darts documentation or other tool? The point is I need it for time series for 100%.

I don’t know any of this tool. I just started to learn, so it is hard to understand which one I should choose. Can you recommend the right tool or explain when use which one?

I think you first need to think about the methodology and only then the tools.
Conventional wisdom is that you cannot predict if prices will go up / down on the stock market other than that in the long run it has kept going up for the stock market as a whole so far.

Best regards

Thomas

think about the methodology

Can you precise what information I should add to my question to let you answer which tool I should use?

The assumption is I don’t know what will work if any.

Conventional wisdom is that you cannot predict if prices will go up / down on the stock market other than that in the long run it has kept going up for the stock market as a whole so far.

Probably, then I will prove it doesn’t work :slight_smile: But maybe there are some situations to detect like panic, correlation with delay between others instruments etc.

I have DataSet like this for each pair of crypto. Of course output can be modified to find the one which will perform well. So for example it can be 4 categories like up / down / very up / very down or number if it will work better.

b_open b_close b_min b_max trades a_volume b_volume taker_buy_a_volume taker_buy_b_volume output
open_time
2019-09-19 10:02:00+00:00 9881.43 9881.43 9881.43 9881.43 1.0 0.001113 1.099803e+01 0.001113 10.998032 up
2019-09-19 10:03:00+00:00 9881.43 9881.43 9881.43 9881.43 0.0 0.000000 0.000000e+00 0.000000 0.000000 down
2019-09-19 10:04:00+00:00 9878.36 9878.36 9878.36 9878.36 1.0 0.012940 1.278260e+02 0.012940 127.825978 up
2019-09-19 10:05:00+00:00 9878.36 9878.36 9878.36 9878.36 0.0 0.000000 0.000000e+00 0.000000 0.000000 up
2019-09-19 10:06:00+00:00 9878.36 9878.36 9878.36 9878.36 0.0 0.000000 0.000000e+00 0.000000 0.000000 up
2022-08-20 23:54:00+00:00 21163.60 21163.15 21157.75 21165.00 232.0 7.238290 1.531763e+05 3.244370 68659.152854 down
2022-08-20 23:55:00+00:00 21163.30 21153.18 21148.72 21163.80 376.0 21.761410 4.603385e+05 10.989020 232455.859014 down
2022-08-20 23:56:00+00:00 21153.00 21152.26 21147.74 21155.00 280.0 17.670890 3.737588e+05 8.656380 183098.366984 down
2022-08-20 23:57:00+00:00 21152.40 21150.93 21147.74 21155.48 408.0 30.842110 6.523626e+05 14.203100 300419.935757 down
2022-08-20 23:58:00+00:00 21150.23 21133.44 21131.23 21155.63 586.0 47.517780 1.004668e+06 24.207030 511790.728087 up

PS Of course I can and probably I will convert them to represent the change of the values instead of the raw value. The table is representation of from what I am starting.