ForecastQ is built and run by an AI researcher and engineer who first built it for himself: a way to invest with the rigour of genuine analysis, without the hours that rigour normally costs. Underneath it sits an open question. Can quantitative AI methods, applied with statistical discipline, produce something durable in financial markets, or only a pattern that works until it doesn’t?
What started as a personal tool became several years of research and engineering, spent studying the problem, distilling it into principles, turning the principles into mathematics, and coding, testing and recoding until the results genuinely held up. ForecastQ is the platform that came out of that work, now open to anyone who wants to look at the market the way a researcher does: as data, distributions and probabilities rather than opinions.
Note that ForecastQ is, first and foremost, a research platform and a living record of an engineering experiment. It is not a financial adviser, and nothing here is financial advice.
The same five steps run every trading day, from the first scan of the universe to the orders placed before the open.
01
The universe is the US stock market. Every trading day the whole universe is re-analysed: custom features are built from the latest data, and the models derive their signals from them. Yesterday’s result is not carried over. The run starts from the data as it stands this morning.
02
Models developed in-house, built on unique and unorthodox methods, score each stock’s probability of a significant up, down or either-way move. The models are the result of years of research.
03
The portfolios are not hand-picked lists. Each one is an algorithmic optimisation over those signals, adapted to its own scenario: a sector, a market-cap band, a risk profile, a diversification rule.
04
Every probability is a statement about today’s setup. A stock that did not move as expected may carry a different probability tomorrow, because tomorrow’s universe is analysed afresh.
Each of the three is measured independently of the other two. They are not complementary and do not add up to 100%: a stock can score high on all three, or low on all three. Read together, they form a single confidence vector for a position rather than three separate opinions.
Try a shape
The three sum to 212%.
Many more readings follow from the shape of the vector. These are only examples of how to read it.
05
The day’s signals and holdings are published before the market opens. That timing is the point: the orders for the day can be set in the pre-market session, in a few minutes, and the portfolio needs no active intraday management.
Questions about the method, a portfolio that did not behave as you expected, or a reason to think part of this is wrong are all welcome. Messages go to the person who built ForecastQ.
The stock recommendations and any related information furnished by ForecastQ are provided strictly for informational and educational purposes only. This information is derived solely from algorithmic analysis and is not, and should not be construed to be, personalized financial, investment, legal, tax, or any other professional advice tailored to individual circumstances. ForecastQ is not a registered investment advisor, broker-dealer, or financial planner and does not provide personalized financial recommendations or services. Accordingly, ForecastQ explicitly disclaims any responsibility or liability for any investment decisions made, or outcomes achieved, based on the recommendations or information provided. Users should not rely on the information provided herein as a substitute for independent professional advice. Users are strongly advised to consult with appropriately qualified financial advisors, legal counsel, or tax professionals before making any investment decisions or implementing any investment strategy. Past performance, whether actual or hypothetical, is not necessarily indicative of, or a guarantee of, future results. All investments involve significant risk, including, but not limited to, the risk of losing your entire investment. Investment values can fluctuate considerably, and losses may exceed the amount originally invested. Users assume all risk associated with the use of the information provided by ForecastQ.