|
|
| (One intermediate revision by the same user not shown) |
| Line 1: |
Line 1: |
| contact Elisa van der Linden (Elisa.vanderLinden@imc.nl). | | contact Elisa van der Linden (Elisa.vanderLinden@imc.nl). |
|
| |
|
| Note that clients would be traveling from the Netherlands for project meetings | | Note that clients would be traveling from the Netherlands for project meetings - they would proceed with two of these at most. |
|
| |
|
| ==Initial proposals== | | ==Initial proposals== |
|
| |
|
| [[News Analysis]]: Fundamental statistics and news about a company is one of the biggest drivers of the stock price of that company. On the other hand, getting traders to act on information is in practice not tractable. Many companies employ methods to estimate the effect of company news for its stock price. In this project you will take on this challenge and create a recommender system based on high frequency news data. The system should use techniques from Natural Language Processing in order to come up with its recommendation. The research will be done on historical data, and will be evaluated on a separate training set. | | [[News Analysis]]: |
| | |
| Owner: Julien Mattei (Julien.mattei@imc.nl) | | Owner: Julien Mattei (Julien.mattei@imc.nl) |
|
| |
|
| [[Game trading engine]]: Data mining and statistical modeling is an essential part of generating successful models. There are various approaches to generate a statistical model, but verification of ideas are usually standardized to a couple of criteria. In this project, the aim is to create a back-testing engine that is capable of evaluating different models according to well-known success metrics such as P&L, and risk metrics like returns, alpha, beta, max drawdown and sharp ratio. The project involves efficient parallelization of experiments with parameter sweeps and visualization of the success of the models. Succinctly put, the input to this engine is a strategy and the output is a vector of success metrics mentioned above. | | [[Game trading engine]]: |
| | |
| Owner: Radmilo Racic (Radmilo.racic@imc.nl) | | Owner: Radmilo Racic (Radmilo.racic@imc.nl) |
|
| |
|
| [[Trading input visualization]]: In this project you will be constrained to a single desktop grade machine with a high-end GPU. Most traders need to either monitor or take in a very large number of, often numerical, inputs and process them quickly. The aim of this project is to amalgamate this data in an easily digestible format that a human can effectively process. The visualization will consist of different views on the same data, focusing on different aspects. Apart from standard plots like scatterplots, histograms and cdfs, on-the-fly filtering and brushing techniques are an important part of this project. New gadgets like 3D goggles(oculus rift) or myo can be used for achieving extraordinary results. | | [[Trading input visualization]]: |
| | |
| Owner: Taylan Toygarlar (Taylan.toygarlar@imc.nl) | | Owner: Taylan Toygarlar (Taylan.toygarlar@imc.nl) |
|
| |
|
| [[Network queuing analysis]]: The Objective of this project is to measure network queuing in a distributed network, by correlating packets between locations and between different identifiers domains. Currently large financial networks are monitored by means of network packet captures that are pushed to central or distributed databases. These databases contain the full lifetime of an opportunity, from receiving a public tick packet in location A all the way to send a private order packet in a different location B, including all the datacenters and all the transformations/processing in between. Assuming that these captures are accurately time synchronized, and there is auxiliary databases that contain the exact or approximate mappings between identifiers domains, it is then possible to accurately reconstruct the full lifetime of all the packets that took part in an opportunity, and statistically compare these to all other billions of packets, of all the other opportunities that passed thought the same network equipment or trading applications. Doing this correlation will enable to measure every single queuing effect in every single step of every single packet, which will enable to optimize and tune the current systems by removing the bottlenecks or increase capacity at the choke points. | | [[Network queuing analysis]]: |
| | |
| Owner: Pedro Estrela (Pedro.estrela@imc.nl) | | Owner: Pedro Estrela (Pedro.estrela@imc.nl) |
contact Elisa van der Linden (Elisa.vanderLinden@imc.nl).
Note that clients would be traveling from the Netherlands for project meetings - they would proceed with two of these at most.
Initial proposals
News Analysis:
Owner: Julien Mattei (Julien.mattei@imc.nl)
Game trading engine:
Owner: Radmilo Racic (Radmilo.racic@imc.nl)
Trading input visualization:
Owner: Taylan Toygarlar (Taylan.toygarlar@imc.nl)
Network queuing analysis:
Owner: Pedro Estrela (Pedro.estrela@imc.nl)
