MIA Last updated: 2022-07-01

Here you can find all the information about Sennet MIA. We are constantly working on creating better and more powerful APIs, so we recommend that you periodically check this documentation for new changes and updates.

What is MIA?

MIA is a Deep Learning model that predicts market trends based on news activity. It has been trained with millions of different data points and constantly learns from new situations in the markets, improving its results over time.

How does MIA work?

With MIA, you can follow the news and its impact on the market. You can predict how individual articles impact the market or monitor all news activity for any asset. MIA predicts market trends in 7 different time intervals ('1d', '2d', '3d', '4d', '5d', '6d', '7d').

A quick example:

We have 4 different articles from top sources published on a given day, and we want to know how they will affect the market. We use MIA to predict the impact these articles will have on the market. For each article, we make a request to get the trend forecast. Here are the results:

article1 = [0.97, 0.81, 0.52, 0.86, 0.78, 0.72, 0.94]
article2 = [0.96, 0.75, 0.51, 0.87, 0.82, 0.78, 0.96]
article3 = [0.91, 0.77, 0.54, 0.89, 0.86, 0.78, 0.93]
article4 = [0.96, 0.74, 0.54, 0.93, 0.85, 0.76, 0.91]
The predictions are returned as a probability, in ranges from 0 to 1, with 0 being a negative market trend and 1 being a positive market trend. In this example, we want to know the predictions for each day taking into account the four published articles. To achieve this we can calculate the average between the predictions for each article:
average_predictions = [0.95, 0.77, 0.53, 0.89, 0.83, 0.76, 0.94]
average_predictions contains the market predictions calculated by MIA for the 7 supported intervals ('1d', '2d', '3d', '4d', '5d', '6d', '7d').

Note

The time intervals start from the time the article is published.

In this example, if the articles were published on July 01, the predictions for a 1d interval will be for July 02:
01 of July: The articles are published.
02 of July: The predicted value is 0.95, or a strong positive market trend.
03 of July: The predicted value is 0.77, or a strong positive market trend.
04 of July: The predicted value is 0.53, or a positive market trend.
05 of July: The predicted value is 0.89, or a strong positive market trend.
06 of July: The predicted value is 0.83, or a strong positive market trend.
07 of July: The predicted value is 0.76, or a strong positive market trend.
08 of July: The predicted value is 0.94, or a strong positive market trend.

Market Impact Analysis

Predicting the market impact of any article.

GET api/news/market-impact/

Parameters


content string

The full content of the article. Recommended content structure: title + description + body. The maximum allowed length is 5000 characters.


Response object

content string

The full content of the article.


predictions array

Market predictions for each interval since the publication of the article ('1d', '2d', '3d', '4d', '5d', '6d', '7d'). Prediction values range from 0 to 1, with 0 being a negative market trend and 1 a positive trend.

Warning

Predictions are generated regardless of the source or publication date of the article analyzed. Keep this in mind before making any interpretation of the results.

Request example

/api/news/market-impact/?content=Lorem ipsum dolor...

Response example

{
    "content": "Lorem ipsum dolor..."
    "predictions": [{
        0.62, 
        0.56, 
        0.41, 
        0.58, 
        0.66, 
        0.48, 
        0.44
    ]}
    "from": "content"
}