LSTM Price Prediction

Introduction

This is my initial implementation of the LSTM (Long Short-Term Memory) model for price prediction. I am releasing this code for public use to help others utilize and build upon it. Additionally, I have used other models for similar projects, and their source code is available on my GitHub repository.

The model predicts the Open, High, Low, and Close prices for various financial assets, including indices, cryptocurrencies, commodities, and currency pairs. The following sections present the performance metrics of the model.

---

Results

BTC-USD (Bitcoin)

Metric	Open	High	Low	Close
Mean Squared Error	0.0008489510	0.0009795529	0.0010661283	0.0011113042
Mean Absolute Error	0.0239363016	0.0253854565	0.0252731072	0.0263076988
R-squared	0.9618445615	0.9568933098	0.9511450073	0.9510090004
Median Absolute Error	0.0209806155	0.0213325995	0.0199003023	0.0207255720
Explained Variance Score	0.9694517512	0.9652160792	0.9537688306	0.9563388589

GC=F (Gold Futures)

Metric	Open	High	Low	Close
Mean Squared Error	0.0102292391	0.0127224254	0.0086396166	0.0091730457
Mean Absolute Error	0.0858521072	0.0955619372	0.0768300133	0.0785435058
R-squared	0.5042878632	0.3761326958	0.5818894131	0.5498634680
Median Absolute Error	0.0786604464	0.0893523749	0.0630110043	0.0646407916
Explained Variance Score	0.8563785943	0.8134662843	0.8579252476	0.8404204121

EURUSD (Euro/US Dollar)

Metric	Open	High	Low	Close
Mean Squared Error	0.0004678367	0.0004778843	0.0007748126	0.0005051806
Mean Absolute Error	0.0179689022	0.0175163507	0.0231784207	0.0182077065
R-squared	0.8915972721	0.8914169532	0.8243772347	0.8831130975
Median Absolute Error	0.0148790617	0.0143152889	0.0205483317	0.0162848522
Explained Variance Score	0.9207051173	0.9007113662	0.8749229833	0.9102028001

GSPC (S&P 500 Index)

Metric	Open	High	Low	Close
Mean Squared Error	0.0085264583	0.0081207582	0.0079891423	0.0076892149
Mean Absolute Error	0.0805443270	0.0776077142	0.0769452674	0.0745821802
R-squared	0.3710538626	0.4289697536	0.4101209639	0.4603227313
Median Absolute Error	0.0754465075	0.0712171335	0.0714640949	0.0656625427
Explained Variance Score	0.8424518479	0.8469419318	0.8395034957	0.8419850381

---

Websites

These results and similar projects have been a part of my experience working on predictive models. I have developed free tools that provide valuable insights into financial markets. You can access these tools at the following websites:

• Predict Price: Free AI-powered short-term (5/10/30 days) & long-term (6 months/1/2 years) forecasts for cryptocurrencies, stocks, ETFs, currencies, indices, and mutual funds.
• Magical Prediction: Get free trading signals generated by advanced AI models. Enhance your trading strategy with accurate, real-time market predictions powered by AI.
• Magical Analysis: Discover free trading signals powered by expert technical analysis. Boost your forex, stock, and crypto trading strategy with real-time market insights.

---

Conclusion

This LSTM implementation represents an initial effort to predict price movements across multiple asset classes. The model is open for public use, and I encourage everyone to build upon it. For further improvements and related projects, check out my GitHub repository.