CaerusLite
Share
Link to this App:
Embed this App on your site:
Description
This is the free version of Caerus, the only difference is this app has fewer supported stocks. Named after the Greek god of chance and fortune, \"Caerus\" is a Artificial Neural Network Stock Market Predictor. It predicts the price of a stock based solely on historical data. Once a network is trained its performance is then measured with a few metrics. These metrics include NN Error, Error Average, and Error Standard Deviation. Unlike people who only remind you of their successes, Cearus keeps honest statistics.
The AIs get trained for days on a 8-core MacPro. The iphone does not have enough power to train worth wild neural networks. These networks will be periodically updated. These networks come with its report cards in hand. For example, over the last 180 trading days the Neural Network that predicts GOOGLE (GOOG) has been accurate within $6.02 and has a standard deviation of $3.59.
Caerus lives and breaths these numbers and seems to have developed an intuitive understanding about the market. It is impossible to know exactly why it makes its predictions. Neural networks can find very deep subtle patterns in chaotic data, however it is only as good as its training data. The best we can do is make honest and concrete metrics, then see how the network scores.
Please note that fundamentally nothing can predict the stock market 100% of the time. The very existence of a prediction can change the market because the market is a chaotic system that works off of information. Any introduction of information can drastically change the outcome. Please take the information presented here in its proper context. These neural networks are trained entirely on the historical data of other stocks. The neural network may pick up on trends that are no longer relevant. There is a reason that the program is named after the Greek god of chance.
The AIs get trained for days on a 8-core MacPro. The iphone does not have enough power to train worth wild neural networks. These networks will be periodically updated. These networks come with its report cards in hand. For example, over the last 180 trading days the Neural Network that predicts GOOGLE (GOOG) has been accurate within $6.02 and has a standard deviation of $3.59.
Caerus lives and breaths these numbers and seems to have developed an intuitive understanding about the market. It is impossible to know exactly why it makes its predictions. Neural networks can find very deep subtle patterns in chaotic data, however it is only as good as its training data. The best we can do is make honest and concrete metrics, then see how the network scores.
Please note that fundamentally nothing can predict the stock market 100% of the time. The very existence of a prediction can change the market because the market is a chaotic system that works off of information. Any introduction of information can drastically change the outcome. Please take the information presented here in its proper context. These neural networks are trained entirely on the historical data of other stocks. The neural network may pick up on trends that are no longer relevant. There is a reason that the program is named after the Greek god of chance.
Screen Shots
