Announcement:
(NEW) TRI-SIGNAL: Predictive Equity Analytics v400 features include 1.) ability to detect algorithmic trading for ETFS, mutual funds and stocks and 2.) the ability to predictively partition daily trend into deterministic and variance components.
To learn more visit https://ihaconsultants.com
Equity price prediction machine learning models are hindered by presence of noise in data which degrades predictive accuracy as a function of noise level which varies over time. Noise in external data sets or contained in a data lake presents an additional time varying noise source. Practitioners can use TRI-SIGNAL predictive equity analytics Signals {1,2,3} data set to overcome the effects of noise in equity price series on machine learning models as starting data set to create predictive equity models with consistent high predictive power with accuracy 85%+. Create customized predictive equity models by combing additional external data sets with IHA Consultant's TRI-SIGNAL Signals {1,2,3} for causality attribution modeling or create deep learning machine models.
TRI-SIGNAL Predictive Equity Analytics Paper avaialble at SSRN using the following links: https://papers.ssrn.com/abstract_id=3875632 and http://ssrn.com/abstract=3703297
Each trade day prior to market open, IHA Consultants generates predictive equity analytic signals{1,2,3} for over 10,500 U.S. equities that are predictive for the current trade day available with subscription. IHA Consultants supplies the complete prior trade day TRI-SIGNAL predictive equity analytics data file free of charge (but with required agreement to release of liability) for use under the licensing terms of the Predictive Equity Analytics TRI-SIGNAL Machine Learning Project and IHA Consultant's Release of Liability Agreement.
NEW TRI-SIGNAL v400 Signal {1,2,3} predictive equity analytic data files can be downloaded by clicking here https://ihaconsultants.com/v600/tri_signal_installer.zip.
Sample report is updated daily prior to market open. Coming soon free abridged data set.
TRI-SIGNAL Series (1,2,3) data files are generally available at 8:30 a.m. each trade day.
Using the installer a template Excel user interface macro file is available for download using the link above. The Excel macro file has vba code for data file retrieval, stock filtering and graphical output to the 'Graphs' tab of the Excel file. An original copy of the data is stored on the 'Original Data' tab and a working copy with filters applied is found in 'Sheet1' tab. Module_1 of the vba code is where you will find the TRI-SIGNAL filtering and graphing routines. The other modules contain standard portfolio theory vba code.
Click to zoom in on Excel macro file interface 
Overview video can be viewed at: https://www.youtube.com/embed/ny_alY1kHKo. Signal 1 and Signal 3 are used for reliable prediction with accuracy greater than 85% for future 1 to 3 trade-day stock price behavior. Signal 2 can be thought of as either representing the result of a well-executed marketing plan (temporary higher level of demand that abates in short order with no lasting correlation) or statistical noise (persistent but inconsistent correlation.)
Signal 1 and Signal 3 are jointly predictive for stock selection and timing of increase. Signal 1 and Signal 3 are each independently predictive when used as sole predictor. Predictive accuracy is increased by using both Signal 1 and Signal 3 for predictive rule construction.
Signal 1 is a stock price series event signal that represents a probable change in behavior from the most current price series data. Signal 1 signal strength ranges from 0 to 1.1. Signal 1 signal strength greater than or equal to 0.6 is considered significant in that a stock price event has occurred and analysis is warranted to determine trade action. The higher the Signal 1 signal strength the greater the significance.Signal 1 strength greater than or equal to 0.6 indicates when a stock price series is likely to change behavior as compared to most recent behavior. For example, a stock price series that was moving horizontally is predicted to move up or more down in future 1 to 3 trade-days . Other examples include a stock price series that was moving down and is now predicted to start moving up or moving horizontally. Lastly, a stock price series that was moving up and now is predicted to move down or move horizontally.
Signal 3 analysis (level and recent trend) is required to determine the predicted future price direction behavior.
Signal 3 is a stock price direction change signal that represents a probable change in behavior from the most current price series behavior. Signal 3 signal strength ranges from 0 to 1.12 and changes are measured relative to the average 121 trade-day value. Signal 3 average is typically found to be in the set {0.9, 0.95, 0.97}depending on the particular stock or sector. Signal 3 stock price direction change is measured by comparing the most recent value to the average or using the most recent signal 3 trend.Signal 3 greater than the 121 trade-day average or most recent signal 3 trend increasing and “near the average” signifies probable stock price increase in future 1 to 3 trade-days. Signal 3 less than the 121 trade-day average or most recent signal 3 trend decreasing and “near the average” signifies probable stock price decrease in future 1 to 3 trade-days.
Using Excel, filtering of 4,000+ equities for Signal 1 and Signal 3 for joint predictive rule construction proceeds with selection of stocks with Signal 1 greater than or equal to 0.6 (all stocks with “stock event”) and Signal 3 greater than or equal to 0.97 (stocks with predicted future 1 to 3 trade-day price appreciation.) There is an automated routine that produces predictive rule application as a report as well in graphical form.Users are free to tweak provided template predictive rules or develop customized predictive rules.
We illustrate automated Optimal TRI-SIGNAL Buy/Sell Predictive Rules using Agilent Technologies (A) stock price series.
Use TRI-SIGNAL Signals {1,2,3} on the Original_Data tab together with Scikit-learn or TensorFlow to create deep learning predictive models optionally combine with external data.
Contact Information
This site is maintained by
IHA Consultants Inc. Please contact us at https://ihaconsultants.com.
MIT License
Copyright (c) 2020 IHA Consultants Inc.
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
Publication date: November 2020, updated January 2021