Atlas Investment Technologies, LLC
Applying statistical analysis and specialized algorithms to securities trading.
Altas Investment Technologies is a pass through entity which primarily manages the Atlas Fund. The fund utilizes trade secret machine learning models and data acquisition techniques to amplify the earning potential of human judgements with regards to macroeconomic trends. This novel technique ensures that the trading algorithm maximizes upside during bull runs and reacts rapidly to reduce exposure during more bearish time periods. This means at worse, should human analysis prove inaccurate, the trading algorithm will suffer minimal losses on that particular bet while still generating profits where possible on others. As this strategy works best when maximizing total returns, that is ultimately the fund's aim.
The Atlas Fund is a closed fund. To invest in it requires a formal invitation from the fund manager. There may be future financial products Atlas Investment Technologies offers which will be open to the general public, however, for now, all resources are private. For inviduals who are able to invest, the fee structure is negotiable, though by default it will be set to the standard 2% base fee and 20% performance fee (taken from the positive marginal difference from the benchmark returns). Any alterations to this arrangement should be visible in your profile. The minimum amount necessary to invest may vary but it is typically set at $200,000. Due to the high risk nature of any fund we advise that you do not invest any more than you are willing to lose.
Withdrawals are restricted to the end of the year. In order to make a withdrawal from the fund you must provide advance notice via the app, should the amount you requested be available at the end of Q4 your money will be withdrawn and deposited into your specified account. Funds may automatically be deposited into your account in the case of intentional liquidation or in order to provide you with the necessary amount you may need for your taxes. In extraordinary circumstances the fund manager reserves the right to restrict the withdrawal of funds for a period he/she deems necessary. Deposits can be made at any time but they must be confirmed by the fund manager before they are added to your account and are visible on your investment performance dashboard.
Gregory is a Google Machine Learning Engineer with a Master's of Science degree in Computer Science with a focus in Machine Learning and an undergraduate degree in Computer Science and Economics from Brown University. He has been at Google for two years working on consumer facing machine learning models at Google Cloud. These models aim to reduce user workload and recommend cost saving actions regarding cloud infrastructure. At Google he created models which generated over $11 million in ARR and worked on Gemini for Cloud Console.
The strategy implemented in the Atlas Fund is, in effect, a more automated version of the algorithms Gregory was using to make trades for two years after graduate school. The thesis is simple, utilize AI to amplify or augment human intuition. In this model the human would provide overarching guidance on stocks to monitor while the AI, which has been trained on millions of data points, is responsible for reacting to interday market movements. This proved to be generally profitable during its early implementation phases, yielding upwards of 30% in returns during a time when the market was generally returning around 20%. Increasing automation and further improvements to the algorithm allowed the model to exceed expectations, however, leading to the creation of Atlas on March 6th, 2024.
As fund manager, Gregory is responsible for monitoring the performance of the model as well as modifying hyperparameters or the model as a whole when deemed necessary. He is also the primary individual responsible for researching new ways to improve on the model to maintain its competitive edge. On top of this, Gregory is also responsible for the creation of the Atlas Fund application which allows investors to monitor the performance of their investments, request deposits or withdrawals, and receive regular updates.
Performance data is difficult to acquire as this fund's methodology is relatively new. In backtesting, with no market frictions and ideal conditions, the algorithm performed exceedingly well. While it could fluctuate from time to time it consistently achieved well above the benchmark rate in terms of returns (often by an order or magnitude). While we recognize such returns are likely fantastical, preliminary forward testing results show consistent returns even in bearish markets. From the period of March 6th, 2024 to April 30th, 2024, a generally bearish window, the model achieved 10% returns while the benchmark (S&P 500) dropped over 1%. With all that in mind the target average return for the Atlas Fund is set at 50% on an annual basis, pre-fees. The above images show some of the more modest backtesting results, the orange lines show the portfolio value if an individual were simple to buy and hold SPY. While the returns, as stated before, may vary, the topology of these results typically remain consistent, giving us confidence that the model is performing as expected and that such variance is a natural artifact of training.