and-rewsmith / and-rewsmith/RNN_LSTM_Stock_Model
Model won't predict for volatile stock prices
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 17
- Forks
- 8
- PR merge metrics
- No merged PRs in 30d
Description
This is not a high priority problem as it will happen for 1/20 stocks.
I suspect that this is because the LSTM can't handle values greater than 1.
This is a problem as prices in a timestep are processed as percent difference relative to the first price in the timestep. If the percent difference goes above 1 then the prediction fails.
My theory of fixing it is to multiply all prices in a timestep by .01 to increase the range of volatility the model will accept.
I'll try this out eventually but if anyone wants to try it before me feel free to try it and submit a PR (with some form of proof of it working) and I'll look at it.
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Locate the timestep percent-difference preprocessing and the LSTM input path, then reproduce the failure with a stock whose relative change exceeds 1. Compare the current scaling theory against the model's behavior and provide proof that volatile-stock prediction succeeds without breaking existing cases.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- fintech-quant, machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100