Can Convolutional Neural Networks Predict Stock Prices?

Рет қаралды 10,333

Күн бұрын

Пікірлер: 24

@Vlad-e9u Ай бұрын

Is it possible to train a model using computer vision to recognize trading patterns and candlesticks on screen, and then use reinforcement learning to train an agent that can trade based on what's happening in real-time on the chart?

@depodtech 2 ай бұрын

Hey Jin, if I want to add volume tick & RSI, what the best way to shape the data?

@zhangjason2879 10 ай бұрын

Thanks for the great video Jin! Would like to see a future video about how to incorporate CNN with bi-directional LSTM to predict the stock prices. Keep up with the great content, all the best!

@jinchoi-moneygeek 10 ай бұрын

Thanks for your support! I'll see about combining CNNs and LSTMs in a future video.

@amirghorbani7922 9 ай бұрын

Predict stocks using ICDST AI PREDICT.

@Scalykams 6 ай бұрын

Damn, this is the project I'm currently working on. Hahaha. Ltsm for direction and CNN for the entry pattern. Mine is for trading. Im just happy to see that it's being thought about

@Ownermode Жыл бұрын

The advantage we have compared to asset managers is that we have way less money in the market. So some strategies are out of the question for asset managers but can work for a retail investor.

@jinchoi-moneygeek Жыл бұрын

Definitely

@OkSid300 9 ай бұрын

Good catch! This is a different kind of game if you are a whale.

@dumbpandabear7947 11 ай бұрын

Hi Jin! Thanks for the great video!! I’m new to the ML Finance space but have been really interested recently, do you have good resources that you recommend? What does training in batch sizes mean? Is the gradient for the cost function adjusted after looking at 64 instances of data instead of a single row ? Also would love to see more videos exploring maybe higher dimensional CNNs within finance and exploring LSTMs and RNNs!!

@jinchoi-moneygeek 11 ай бұрын

Thanks for the support. There are some books on the intersection of ML and finance written by Ernest Chan and Marcos Lopez de Prado. But personally, I've found the general ML books more helpful. I like Kevin Murphy's book, for instance. Training in batches mean exactly what you said. I'll consider making a video on using more elaborate ML structures in the future. It's hard to do in finance because the data is so noisy, but it could be worth a shot.

@murtuzakhan3655 Жыл бұрын

Hey Jin , is there any way to create targets for images created by first converting a 1D close price column into a recurrence plot and then from the main recurrence plot we are creating subsets out of it eg 10*10 size and applying Gadf image processing on to the subsets but i'm confused how to create targets for the final gadf images

@jinchoi-moneygeek Жыл бұрын

Sorry for the late reply. Have you looked at papers.ssrn.com/sol3/papers.cfm?abstract_id=3756587 ?

@muhammednihal2410 10 ай бұрын

Can you explain what you mean by overfitting

@jinchoi-moneygeek 10 ай бұрын

Take a look at this other video I made kzbin.info/www/bejne/oZOmp2eGrt5jZq8

@lokeshkumar-dq7yl 4 ай бұрын

Over fitting is too much bias on training data and not performing well in testing data

@rkedits_alone 4 ай бұрын

Over fitting means memorizing if give 1-1=0 next if give any other its doesn't able to do it

@OkSid300 9 ай бұрын

In other words. I need to develop the solution myself. And it have to be out of the box, groundbreaking strategy that nobody would guess for. Interesting challenge even though that sounds impossible. But still the biggest question is still remains. How do regular traders are still earning money on market if mathematicians and data scientists are swarming WallStreet with all fancy tehnology?

@jinchoi-moneygeek 9 ай бұрын

That's a good question. I have two answers to your question. One, very few traders actually make money consistently. I read one statistic that says only 3% of day traders make money - though I don't know how accurate that statistic is. Two, we should give more credit to the human mind, which can reason much better than machines can. Machines are really good at detecting repeated patterns, but they're not so good at contextualizing the patterns. Are stocks falling because of Federal Reserve action or because of Covid? Such contexts are hard for machines to comprehend, especially if the contexts are new.