README.md

metadata

icon: chart-line-up
description: >-
  A dataset with daily updated predictions of price breaking upwards for US
  Equities.

Price Breakout

Data Notice: This dataset provides academic research access with a 6-month data lag. For real-time data access, please visit sov.ai to subscribe. For market insights and additional subscription options, check out our newsletter at blog.sov.ai.

from datasets import load_dataset
df_breakout = load_dataset("sovai/price_breakout", split="train").to_pandas().set_index(["ticker","date"])

Daily predictions arrive between 11 pm - 4 am before market open in the US for 13,000+ stocks.

Tutorials are the best documentation — Price Breakout Prediction Tutorial

Category	Details
Input Datasets	Historical Stock Prices, Trading Volumes, Technical Indicators, Order Book.
Models Used	Classification Algorithms, Regression Models, Conformal Predictors
Model Outputs	Price Movement Predictions, Probability Scores, Confidence Intervals

Description

This datasets identifies potential price breakout stocks over the next 30-60 days for US Equities. This dataset provides daily predictions of upward price breakouts for over 13,000 US equities.

The accuracy is around 65% and ROC-AUC of 68%, it is one of the most accurate breakout models on the market. It is retrained on a weekly basis.

Several machine learning models are trained using the prepared dataset:

• Calibrated Classifier: A classification model trained on the engineered features to predict the binary target.
• Proprietory Regressor: A proprietory regression model is used to predict the probability of a price increase.
• Conformal Regressor: Used to provide calibrated confidence intervals around the predictions, offering an additional measure of uncertainty.

Data Access

Retrieving Data

Latest Data

import sovai as sov
df_breakout = sov.data("breakout")

Full history

import sovai as sov
df_breakout = sov.data("breakout", full_history=True)

Specific Ticker

df_msft = sov.data("breakout", tickers=["MSFT"])

Plots

Line Predictions

df_breakout.plot_line(tickers=["TSLA", "META", "NFLX"])

Breakout Predictions

Visualize breakout predictions using the SDK's plotting capabilities:

sov.plot("breakout", chart_type="predictions", df=df_msft)

Prediction Accuracy

Assess the accuracy of breakout predictions:

sov.plot("breakout", chart_type="accuracy", df=df_msft)

Data Dictionary

Column	Description	Type	Example
ticker	Stock ticker symbol.	object	"AAPL"
date	Date when the data was recorded.	datetime64[ns]	2023-09-30
target	Target variable for predictions.	float64	0.05
future_returns	Future returns of the stock.	float32	0.10
prediction	Predicted probability from the model.	float64	1.25
bottom_prediction	Lower bound of the prediction interval.	float64	1.20
top_prediction	Upper bound of the prediction interval.	float64	1.30
standard_deviation	Standard deviation of the predictions.	float64	0.02
bottom_conformal	Lower bound of the conformal prediction interval.	float64	1.18
top_conformal	Upper bound of the conformal prediction interval.	float64	1.32
slope	Slope derived from the rolling regression of predictions over a window.	float64	0.003

---

Use Case

Understood. I'll focus on the use cases that would be most relevant to professional investors. Here's the refined list:

• Portfolio optimization:

• Identify potential new additions to diversified stock portfolios
• Rebalance existing holdings based on breakout predictions

• Risk management:

• Use confidence intervals and standard deviations to assess potential downside risk
• Implement more precise hedging strategies based on predicted price movements

• Sector and market analysis:

• Identify trends across industry sectors or the broader market
• Compare breakout potentials across different stock categories (e.g., large-cap vs. small-cap)

• Market timing:

• Use aggregate predictions across multiple stocks to gauge overall market sentiment
• Time entry and exit points for broader market positions