Macro3M 4+

Macro3M uses mathematical statistics and machine learning models to analyze the impact of U.S. economic indicators on the market, find the rules and build a generalized model. Through the model, you can enter specific economic indicator data to predict the market performance next month. You can use the predicted value of the model to help you analyze the impact of economic indicators on the market. In the long run, the market always fluctuates around the economy and tends to the same direction.

Economic Indicators:
Economic indicators are statistics about economic activity. The dataset analyzed by Macro3M contains 14 U.S. economic indicators from 1967 to 2023, 4 of which are highly correlated with the U.S. market. The 4 indicators are "M2 Money Supply", "Producer Price Index", "Industrial Production Index" and "Nonfarm Payrolls". These indicators help analyze the overall performance of the economy.

Algorithms and Models:
Macro3M uses three "Deep learning algorithms" to build three generalization models. The evaluation metric for these models is to minimize the mean absolute error (MAE) between the predicted value and the target value. Macro3M has long tracked the MAE performance of nine "Machine learning models" and the final results show that "Deep learning models" outperform traditional "Machine learning models”.

MLP Model:
MLP is very flexible and can usually be used to learn the mapping from input to output.
A multilayer perceptron (MLP) is an artificial neural network that can be used to classify data or predict outcomes based on the input characteristics provided by each training example. It is also known as the basic architecture of deep neural networks.

RNN Model:
RNN mainly deals with the prediction of sequence or time series data.
The difference between RNNs and other neural networks is that they consider time and sequence and have a time dimension. For sequential data, RNNs are favored because their patterns allow the network to discover dependencies on historical data.

LSTM Model:
LSTM is a special kind of RNN that can learn long-term dependencies between data.
LSTMs are essentially an improved version of RNNs. LSTMs add a way to pass information across multiple time steps to interpret longer sequences of data.