Analyse et développement d'un outil statistique pour entrer dans un marché boursier survendu

Abstract

In this thesis, we proposed an innovative avenue based on industrial engineering tools, in order to predict the stock market movement: The Six-Sigma methodology. The objective of this thesis will be to verify its validity in order to maximise short term yields (1 week, 2 weeks and 1 month). Variation patterns of this methodology have been tested on various Exchange-Traded funds, but more specifically on SPY, which is correlated with the S&P500 index. After rigorous analysis and with the use of Z and Student tests, we concluded that two patterns offer a yield that is statistically more significant than the 1 month expected average yield (with a confidence intervalle of 97,5%). It would be beneficial for an investor to buy SPY when one of the two following patterns occur, pattern 1: Nine consecutive closing prices are below the 20-day simple moving average and/or, pattern 2, when six consecutive closing prices are in a downward trend. Both variation patterns were then used on 9 other asset classes. For the nine values below the moving average, the yield was significantly higher than the control yield for GLD, XLE and EEM for a confidence intervalle of 75%, XIU and TLT with 90% and finally, QQQ and DIA with 97,5%. As for the six values in the downward trend, yield was significantly above the control for IWM, FXE and XLE with a confidence intervalle of 75%, TLT for 90% and DIA for 95%. On of the most important highlights on technical analysis is that results are based on probabilities. Using those two Six-Sigma patterns would represent an asset of great value for the investor as a decision-making tool. Nevertheless, the investor will understand that the technical analysis methodology and by respect, its results are subject to probability and statistics and that potential gains are not a guaranteed result.

Date	18 Oct 2017
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Edmond T. Miresco (Supervisor)