Quantitative Research Engineer / Quantitative Researcher / Quant Developer / Algorithmic Trading Researcher / Machine Learning Engineer for Financial Markets
Remote project-based work / development of a quantitative research & prediction system
Relevant areas of experience:
Quantitative Research / Quant Development / Algorithmic Trading / Machine Learning for Financial Markets / Time-Series Research / Trading Systems Development / Systematic Trading / Quantitative Trading
We are looking for an individual specialist / a team of specialists / a specialized company to develop a system for forecasting intraday movements of U.S. stocks listed on the NYSE and NASDAQ.
The system's primary objective is to use historical and current market data to identify stocks with an elevated probability of a persistent, sufficiently strong, and relatively smooth directional LONG or SHORT move within a specified future time window.
We are not simply looking to predict whether the final price will be above or below the starting price. We are interested in the quality of the price path: predominantly one-directional movement without significant adverse moves, repeated reversals, or chaotic price action.
The system must distinguish between:
LONG โ a high-quality directional move upward;
SHORT โ a high-quality directional move downward;
NO_TREND โ the absence of a sufficiently high-quality directional move.
Truth
We already have our own methodology for defining LONG / SHORT / NO_TREND.
At the first stage, it will be provided to the contractor for an independent methodological review, discussion, and final agreement.
Once agreed, the Truth methodology is frozen and must not be changed to fit the results of a specific model or the final TEST/HOLDOUT.
Final Project Goal
In production mode, the system must analyze the stock universe on a daily basis and select
5 to 10 of the most promising stocks with a LONG or SHORT forecast.
Primary target KPI:
average monthly accuracy of 40%+ among the signals actually selected by the system.
In other words, over the course of a month, at least 40% of all evaluated LONG/SHORT forecasts actually selected by the system for use must match the pre-defined Truth.
The prediction time and the future prediction horizon must be configurable system parameters.
In real-time mode, a forecast is generated at time ฯ, and its actual outcome is evaluated after the corresponding horizon H has elapsed, using the same Truth methodology.
The specific statistical / ML methods and the model's internal architecture are not prescribed by the client in advance.
Example of the Type of Movement We Are Looking For
Real intraday candlestick charts are included in the project description.
Collaboration and Compensation
Our primary focus is on results, rather than hourly billing.
The first key threshold is confirmed forecast accuracy of at least 35% under an agreed validation protocol, with no look-ahead, data leakage, or fitting to the control data.
The result will be evaluated on a pre-defined independent period and/or in live/paper-trading mode using the agreed Truth methodology.
The target operating mode is 5โ10 selected LONG/SHORT forecasts per trading day.
A separate bonus is provided for achieving confirmed accuracy of 35%+ under the agreed methodology.
The final target is average forecast accuracy of 40%+ per calendar month, with a sufficient number of evaluated signals.
After the 35% level has been confirmed, we are ready to discuss one or a combination of several collaboration models:
Performance-based compensation. As a guideline โ from USD 100 for each correctly evaluated forecast, subject to pre-agreed rules regarding the maximum number of signals, accuracy, Truth, and the method used to calculate results.
License / subscription โ recurring payment for continuous access to the system or daily forecasts. Ownership of the software may remain with the contractor.
System buyout โ transfer to the client of full rights to the software and all required components: source code, data pipeline, prediction/research pipeline, configurations, documentation, and instructions for independent operation and further development.
To apply:
Please send a short note including:
Your experience
Your general approach (you do not need to disclose your algorithms)
When you would be available to start working on the project
Project-based, performance-oriented compensation; the final format and budget are agreed individually.