Alan Hutton, Xinfinit Sales & Marketing Director

David Kedmey, EidoSearch President & Co-Founder

Introduction

Any portfolio manager will tell you that the process of managing financial positions around quarterly announcements is still considered a “black art”. How do you respond to new information? Are adjustments and tweaks necessary? For large positions where you have a long-term bullish or bearish outlook do short term market moves count?

The emergence of predictive analytic tools interacting with “big data” in the past few years provide investors the ability to perceive new signals and make better predictions around asset price movements. The promise of sentiment analysis using social data – internet traffic, twitter feeds and chat rooms – has fallen short. Alternative data sets of all stripes hold promise but they require a more analytical tear down to quantify the amount of predictive power contained within.

EidoSearch is an example of a fintech start-up that is advancing the data analytics revolution by quantifying the predictive power of traditional and alternative data. Their numeric search engine allows investors to get their hands around future price movements based on repeating historical patterns and their associated outcomes. 

Predicting Profitable Outcomes

The EidoSearch engine essentially uses algorithms across stock prices, economic statistics and market conditions to generate predictive analytics that leads to better decisions. Using historical and current global macro conditions this engine can accurately create links across all asset classes to determine future market directions. Their methodology is extremely useful for equity traders, analysts and portfolio managers trying to get their heads around projected returns and probabilities.

Markets are driven by investor behaviour, evidenced by patterns of overvaluation and/or undervaluation. These conditions reflect a complex sequence of reactions to events (and reactions to reactions). Quantifying the behavioural responses of investors allows asset managers to anticipate and respond to market and macroeconomic events using the return probabilities that are most relevant to current conditions. Asset managers can reason about the risk premium for every investment using these analytic tools (Figure 1).

The four key steps in the EidoSearch engine execution are :

• Data Identification – search a host of data sets simultaneously to identify current conditions that may contain information.
• Data Search – a proprietary search algorithm finds similar historical incidents to the identified conditions and returns the most relevant results.
• Result Analysis – generate a probabilistic range of outcomes associated with similar historical incidents.
• Fully Automated – as new information becomes available the engine automatically updates and uncovers conditions that contain maximum information to use in an investment decision.

One way that asset managers and hedge funds benefit from EidoSearch is by using the objective statistics and return probabilities to provide analytic insight into the right time and level to enter and exit trades. This, in turn, offers a unique monitoring and risk management scenario to maximize return profitability on positions. Using historical return patterns also allows options traders to forecast market volatility and better quantify the probability of tail events. The built-in back test capabilities can forecast return distributions for any emerging pattern, adding a new layer to classic momentum and reversion strategies.

Figure 1 : Using EidoSearch Analytics to Forecast Return Distributions and Compare to Implied Volatility

Xinfinit access the EidoSearch platform using RestAPI’s located on their production server to create numeric searches and forecasting available within the Xinfinit web container and visualized using the Xinfinit widget library. 

The EidoSearch Widget Cloud Library inside Xinfinit’s Container

The Xinfinit open ecosystem platform provides EidoSearch with multiple HTML5 based applications all located within a comprehensive widget cloud. The container provides multiple layout managers enabling the widgets to be flexibly arranged in layouts of choice and all governed by the Xinfinit Admin Centre (Figure 2). This gives EidoSearch the capability of managing their permissions and determines which widgets can be made available to EidoSearch and their own end client user groups.

Figure 2 : Xinfinit’s Administration Centre

Conclusion

While traditional methods using sentiment analysis via the internet and chat rooms will continue to play a role in price prediction, it is expected that software analytic tools will have a far larger role to play. During the past few years, EidoSearch has advanced beyond traditional analytic approaches, helping to usher in a paradigm shift in the analysis of current market conditions and cross comparing them against history to generated forecasted return distributions. Capturing these events to analyse the links within and across asset classes using multiple variables gives investors a powerful tool to support their investment portfolios and trade decisions.

The relationship between Xinfinit and EidoSearch can be compared to the workings of an engine; that is, vendor agnostic data feeds (the “fuel”) is converted by the EidoSearch pattern recognition software (the “pistons”) into information. This information-based power is transferred to the Xinfinit cloud (acting as a “connecting rod” from the pistons), which animates the Xinfinit’s charts and visualization widgets (the “crankshaft”). Without these interacting parts of the engine it would be impossible for any strategic investor to visualize the powerful impact of using predictive software algorithms for financial data analytics.

The EidoSearch engine is a powerful partner inside the Xinfinit open cloud, web based HTML5 container. Coupled with Xinfinit widgets, investors can now quickly and easily visualize their key asset classes, indices and market sectors that are most sensitive to market conditions and global macros trends from both an historical and forward-looking perspective.