Performance model for thrustworty and low cost money transfers on the Blockchain


Design of a low reliable, fast and low cost money transfer service for developing countries. This service

would prevent users from lining up for hours at the money transfer office and providing oher fast and valuable

services. A Survey conducted in Kinshasa has been conducted to understand the users fears concerning

using a money transfer mobile application to transfer their money using the Blockchain technology. 





The money transfer process, on the blockchain, currently has trouble scaling up when there is a high number of concurrent transactions. Its speed and costs grows too much to be usable for small money transfers. A potential solution to this small cost money transfer is available using the DLT Business Model, proposed by professor Kaiwen Zhang, of l'ÉTS. In this project it is tested by designig an experimental mobile service using the Blockchain.


                               Blockchain based model for low cost money transfer  (Maketa, 2018)                                    



S'assurer d'avoir un prototype d'application mobile fiable qui inclus un mécanisme qui s'assure que le dépôt est effectué soit au fournisseurs identifié pour la personne ou directement à la personne. Le système doit être fiable et rapide pour gagner la confiance des utilisateurs.


             WHAT WE ARE DOING   


A first version will test the money transfer without any special conditions. The sender receives a confirmation when the money is received. The second version sends money to a third party. The money can be given whe the third party accepts.
In the thirs test, a budget is sent and som money can only be used to pay certain third party suppliers. This lat test allows the receiver to obtain services from that third party, for example student fees, directly.


              STUDENTS TEAM     T. Maketa


              TECHNOLOGY     to come

This project is done on Screen-Shot-2016-12-07-at-10-42-18-AM.png


Stock-Market Robots (high-frequency)


Design of a semantic component orchestrator that supplies rich information for decision making. These

components are interconnectable/interchangeable and generate buy/sell/wait signal as well as a context






Stock Brokers and investors are looking more and more into trading bots. They would like to have richer and more diverse information sources for decision making. It is difficult for them to obtain this information at the right moment so it be actionable. We are developing standardized collectors to extract information from specialized Websites (i.e. MarketWatch, Bloomberg, Reuters, ..) influencers (for example @PaulScolardi, @Burns277, @OptionsHawk), specialized databases (i.e. NasTraq, Taq, OptionsMetrics) and technical analysis. See ou paper (in French article in Québec Science).


                                                                Overview of the Prototype                                                          



Experiment with many Web semantic extraction techniques to obtain specific market information.


             WHAT WE ARE DOING   


Here is a few technical reports (in French) of this project: d'analyse de sentiments pour le domaine de la finance, de signaux pour automates de trading et d'enrichissement des signaux.

Currently we are developing the market trend predictor component based on existing techniques 1, 2 and 3. This API takes the market transactions as an input and provides a trend prediction for 2,3 and 5 minutes. 


              STUDENT TEAM     T.Maketa, B.Lebois, D.Méthot, S.Santerre, C.Simon et N.Hubert


              TECHNOLOGY     Anno4J, Alibaba, Marmotta, Camel, Hadoop, Java/Python, pyAlgoTrade, RabbitMQ, SPARQL,
                                            RDF/XML, Interactive Broker, JASON, ElasticSearch/ Kibana


This project is done on Screen-Shot-2016-12-07-at-10-42-18-AM.png


Real Time Stock Market Statistical Engine


This project was proposed by TickSmith, a Montréal startup spécialized in Big Data for FinTech. TickSmith

processes stock market data from many sources and its customers use their platform for trading and also for





The first objective is to scale up and generalize, using Spark/Scala technologies, a first prototype that allows the user to define his own formula and autmatically run it against the stock market data in real-time. The second objective is to validate the results and allow the user to add and run many formulas (volumes, spreads, volatility réalisés, etc.).


             CE QUE NOUS FAISONS  


This figure shows a first version of a prototype parser that generates statistical formulas, using «scala.util.parser.combinator», creating a mathematical grammar that can express each unique element of the formula. Then it is possible to pressent the formula graphically so that he/she can validate it before sending it to processing. 

Four formulas were experimented (Volume, VWAP, VWAS et GK) at large scale on an AWS cluster.

The following AWS configurations were experimented:

Finally, here is an example of the parallel processing results for the execution


              ÉQUIPE D'ÉTUDIANTS     Philippe Grenier-Vallée et Luiz Fernando Santos Pereira


              TECHNOLOGIES     Spark 2.0, Scala, Java, Scala Parser Combinators, JLatexmath, JSON, AWS EMR, Maven,

This project was done on Screen-Shot-2016-12-07-at-10-42-18-AM.png