Description
In the foreign exchange market, the smallest changes in international affairs and economics can have a huge effect on the conditions of a financial transaction, sometimes resulting in gigantic losses without the involved parties even being aware of it.
This possibility of financial losses resulting from the variation of exchange rates between local and foreign currency is called "exchange rate risk." Because the values of foreign currencies relative to a local currency are so unstable, their exchange rates are extremely difficult even for investors and exchange markets to accurately predict.
To find a solution to this issue, SoftBank has commissioned this competition as part of their research and development efforts. The successful submissions will be utilized in an exchange position transaction and management algorithm.
Guidelines
Predict the end-of-month exchange rate with given variables: market and economic news data.
The data set contains: training set, test set, and text features folders.
<train & test>
id: encrypted date
span: days to target date (end of month)
target: exchange rate on target date / current exchange rate
feature_00~feature_59_type5: various market data
<text features>
id: encrypted date
feature_0~feature_299: text features
Outputs should follow the format:
id, target
z, 0.9976
y, 1.0245
x, 1.07
etc.
*Please note that the file should be sorted by id in descending order.
The accuracy rate you'll see on the submission page will not be accurate due to the time we take to get the right accuracy. It will be refreshed as soon as possible.
You will be able to see the leaderboard later that shows the rankings of competitors based on the accuracy on a subset of test data.
Rules
This competition is governed by a Non-Disclosure Agreement. Participants must agree to and comply with this NDA in order to participate.
Please do not contact SoftBank directly in regards to this competition and if you have any queries regarding this competition, please contact us at: develop@bitgrit.net