Knowledge Pattern Search (KPS) For Market Anomaly
We show an example of using the KPS to search for market anomalies. In the financial market, it is long believed that the market is efficient, namely the price of an asset at any given time reflects all the information known to that moment. In other words, when an event happens, the price of the asset is adjusted to the level affected by the event instantaneously, because the "market", which is an ultimate smart and optimal "being", is extremely efficient to dissimilate the information. This is the market efficiency theory.
However, in real-life, market anomalies do exist, which seem to be inconsistent with this maintained theory, and often indicate rare profit opportunities for investors. One of the interesting phenomena is that anomalies seem disappear after the publications of the anomalies. The hypothesis is that more market participants implement the published strategies to make profit and cause the anomalies go away quickly.
Market participants constantly and diligently seek anomalies or profit opportunities. In consequence, the market becomes more efficient. However, once these anomalies are known to more people, profit opportunities disappear. Therefore, in order to profit frequently, investors should constantly look for new and different anomalies.
New concepts, technologies and products are fresh suppliers of market anomalies. What are these anomalies that might become popular quickly? Naturally, an anomaly can have a true value only when more people know it and consider it valuable, i.e. becomes popular. Such a highly potential anomaly has to be intrinsically valuable, known to fewer people and deviated from the current market norms. Therefore, popular information can not meet these criteria. Only an unique piece of information might have such a potential. However, an unique piece of information does not always generate a true value.
Figure 1: Market anomaly search represents significant opportunities
As an investor, you want to discover a valuable anomaly as early as possible; As a marketer of a new concept and product, you want to know how to realize the value of an anomaly as soon as possible. Market anomaly search represents significant opportunities as shown in Figure 1. The KPS network will help both as follows
Go to Knowledge Pattern Search, input a keyword "investing". The returned results are shown in Figure 2 containing the related keywords in red.
Figure 2: Associated keywords are highlighted with the search input.
The results are sorted by their anomalousness, namely the first result is more interesting and unique information than the second or third one, so on. This is different from a conventional search engine where the results are usually sorted according to a page's popularity. For example, if you enter the same search in a popularity-based search engine (e.g. Google, Yahoo or MSN), you get the highest ranked information are the popular pages explaining "investing", but no specifics. Our market anomaly search returns the anomaly pattern related to the area of "investing", is a current market anomaly pattern that has a profit potential.
Figure 3: Meaning clusters are sorted according to their correlations with the input search context
In Figure 3, the first result is associated with a meaning cluster with a correlation 46.15% from the news Server 2 which contains 13 results; the second result with the news Server 3 with a correlation 6.45% which contains 31 results, and etc. The first meaning cluster is the market anomaly pattern where the unique information with respect to the context search "investing".
Join the KPS network, discover your investment opportunities and become the suppliers of valuable anomalies! The dashboard demo is a continuously market anomaly monitoring of real-time financial news. We use four search indexes in four servers. Each server learns knowledge patterns from a single source of business daily news of public companies. The total number of companies is about 4000. Each server mines about one day of business news for 1000 companies.
|Site Map | Privacy | Terms|
|Copyright © 2001-2008 Quantum Intelligence Inc. All rights reserved.|