This thesis explores US stock market seasonality and the development of graphical and automated trading systems to leverage its profitability. In both cases, seasonal patterns are detected and examined in terms of their historical returns, current returns, and their reliability as stock market indicators. A fixed frequency graphical tool is used to evaluate certain wellknown seasonal periods and to identify emerging seasonal patterns over the last 20 years. Certain of these patterns are found to still provide consistent results, such as exiting the market in the month of September, whereas others are no longer as reliable, such as the Hirsch Cycle. A graphical approach using seasonality curves based on trading days of the year is also tested over the same period. This method is then modulated using a weighted average favouring recent years, by detrending the seasonality curve, and by combining it with the MACD technical indicator. Next, using an automated seasonality detection tool, several high-performance seasonal patterns are identified for the SPY and a selection of stocks. This same automated seasonal analysis is then applied to the over 500 stocks making up the S&P500 for 10 consecutive analysis periods of 10, 15 and 20 years. These results are then grouped and processed in order to identify the characteristics that are shared by the most reliable seasonal patterns in terms of their stock market forecasting. Finally, the use of ETFs is also studied as an alternative the previous exhaustive analysis. These produced favourable results, notably for the Health sector.
| Date | 26 Feb 2024 |
|---|
| Original language | French |
|---|
| Awarding Institution | - École de technologie supérieure
|
|---|
| Supervisor | Edmond T. Miresco (Supervisor) |
|---|
Brault, D. (Author),
Miresco (Supervisor),
26 Feb 2024Student thesis: Master's thesis › Master in Engineering: Engineering