- Jacky
- Jerry
- Jumana
In the investment terms options is a derivative which is derived from the price of another security. Another security can be a stock, a currency, rate or a commodity. This means that the price of options moves if the price of another security would move. Understanding how options are priced is important because there are a lot of variables that determine its value.
In this project we are creating a Deep Learning model for the prediction of the price of an option. Options prediction can help in determining the investment better and understanding the stock marketing better. In our model, we pick specific options of the stock. We need to consider the five important characteristics of the option stock-Underlying asset, Call vs. put, Strike price, Expiration date, and American vs. European. These five features are important to our model and output of the model will be the price. As we know there are many different strategies for options trading depending on what we want to get and how much risk we are willing to expound, we will limit the features for the scope of this project.
- https://docs.google.com/presentation/d/1TLyDg_THqW6SLe9ueSo-tYWTkLsFwSMV6-OCWgZKhY0/edit?usp=sharing or
- https://github.com/zjzsu2000/CMPE297_AdvanceDL_Project/blob/main/Documentation/01_slides.pdf
-https://github.com/zjzsu2000/CMPE297_AdvanceDL_Project/tree/main/01_Data_Preprocessing
- LSTM
- Put5 (https://tensorboard.dev/experiment/IN9XplUvRbqhaVBvCPIJSg/)
- Put10 (https://tensorboard.dev/experiment/oQljEuYVSvOLPDi17AtRww/#scalars)
- Put21 (https://tensorboard.dev/experiment/h5KIjLjiQPmdk9cvD9JD6A/)
- Call10 (https://tensorboard.dev/experiment/9tG0Hv68TmyaI7NBnbr2Ow/)
- Call21 (https://tensorboard.dev/experiment/ZE4cErmZRXKKvUJRpHbjHQ/#scalars)
- Model2:
-
Call 5 https://tensorboard.dev/experiment/Bgx7f38NSgOtJQdNMXI9JA/#scalars
-
call10 https://tensorboard.dev/experiment/y9MuyU77Tymi0Uz1Yb0NSA/#scalars
-
Call 21 https://tensorboard.dev/experiment/y9MuyU77Tymi0Uz1Yb0NSA/
-
put5 https://tensorboard.dev/experiment/L208zOrlRziyfgiEm9RjjQ/#scalars
-
Put21 https://tensorboard.dev/experiment/L208zOrlRziyfgiEm9RjjQ/
- Model3:
-
call5 https://tensorboard.dev/experiment/X4dEGe3NQbOBOU5vFzHQuw/
-
call21 https://tensorboard.dev/experiment/V2WF49BDQlK1spQ21eedfA/
-
Put5 https://tensorboard.dev/experiment/RY5C4BGKRIKpndYhO4sdqg/
-
Put21 https://tensorboard.dev/experiment/9zzUE8NcQoySec6DarEo9g/
-
https://github.com/zjzsu2000/CMPE297_AdvanceDL_Project/tree/main/04_models_training
-
Datasets in Google Drive https://drive.google.com/drive/folders/1hBtQBkhc2l0ZTnMOHm49a-bAlLFT8Tl_?usp=sharing
https://github.com/zjzsu2000/CMPE297_AdvanceDL_Project/tree/main/06_TFX
TFX end-to-end Pipeline with Model Training
Example Validator/ Data Anomalies-
Transformed Data
Model Training
Model Evaluation
Pushed model
TFX Model Serving Architecture
StatisticsGen
Artifact
Anomalies
TFX Model Serving
Price Prediction Results
- Stocks filter — Using the stocks with better Sharpe Rate
- Option filter -- Not using the options deep out of the money(The Deep-OTM options are more volatile) -- Not using the options expired in 7 days (The options closer to the expiry date are more volatile)
- More models, More data, More training, More metric
- Apply to the real trading
- Fine-tuning with more hyperparameters