Better Abstract

Online Portfolio Selection/Online Portfolio Selection - Pattern Matching.ipynb

@@ -33,13 +33,6 @@
  "source": [
  "Pattern matching locates similarly acting historical market windows and make future predictions based on the similarity. Traditional quantitative strategies such as momentum and mean reversion focus on the directionality of the market trends. The underlying assumption that the immediate past trends will continue is simple but does not always perform the best in real markets. Pattern matching strategies combine the strengths of both by exploiting the statistical correlations of the current market window to the past.\n",
  "\n",
- "In the following notebook, we will examine various Correlation Driven Nonparametric Strategies. Using the Pearson correlation coefficient as a threshold for similarity, CORN identifies similar periods to maximize returns for the contained set.\n",
- "\n",
- "Symmetric and Functional CORN strategies are further extensions that not only look at similar periods but also inversely correlated periods. As markets tend to have symmetric properties, using both ends of the correlation improves these strategies to maximize returns for similarly performing one but also reduce drawdowns for anticorrelated periods.\n",
- "\n",
- "Further on, we will examine a further improvement of these strategies by using the top-K selection process for the universal portfolio. By selecting the top-K best-performing strategies each period, the universal portfolio employs an ensemble method that reduces variability and can effectively choose the strategy with the highest returns.\n",
- "\n",
- "\n",
  "**1. Correlation Driven Nonparametric Learning**\n",
  " - CORN\n",
  " - CORN-U\n",