The Big Data Combine engineered by BattleFin are rapid fire, live tryouts for computer scientists with elite predictive analytic skills intent on monetizing their models. The first stage of the competition is a predictive modeling competition that requires participants to develop a model that predicts stock price movements using sentiment data provided by RavenPack. Traders, analysts and investors are always looking for techniques to better predict price movements. Knowing whether a security will increase or decrease allows traders to make better investment decisions and manage risk more effectively.
\r\nThis competition is designed to identify people with the talent to create a predictive model using financial data. Competitors are given intraday trading data showing stock price movements at 5 minute intervals and asked to predict the change two hours in the future. The winners of the predictive modeling phase are invited to the \"live\" Big Data Combine tryouts in Miami, FL. Up to 12 finalists will be selected to compete in the live event in Miami. The lucky few will pitch their predictive model to expert judges and an engaged audience. They have only three minutes to present in non-technical terms three items: personal background, predictive model description, and how they would use there model to make money in finance. If their model and presentation impresses our judges, they will be eligible to work with BattleFin and Deltix to convert their predictive model into a trading strategy.
\r\nMatt Iseman - Actor, Comedian, Host of American Ninja Warrior
\r\nMr. Iseman has hosted the game shows Scream Play on E! and Casino Night on the GSN. He currently appears as a regular cast member on the home makeover show Clean House, and its companion outtakes show, Clean House Comes Clean, both on the Style Network. Additionally, he hosted season 2 and 3 of American Ninja Warrior on the channel G4. He also has worked episodically in television shows including The Drew Carey Show, NCIS, and General Hospital. He has appeared on the syndicated MAD TV, Comedy Central’s Premium Blend, Fox’s The Best Damn Sports Show Period, and Fox News Channel’s Red Eye w/ Greg Gutfeld. Mr. Iseman was also the host of Sports Soup, a spin-off of E!'s The Soup, on Versus. Style Network and Versus are owned by Comcast. Iseman began working with American Ninja Warrior (G4) in 2010. He uses his great athleticism and work as a comedian to add his style to the show with Johnny Moseley (American Professional Freestyle Skier), and Angela Sun (Sideline Correspondent).
\r\nIlya Gorelik, CEO of Deltix Inc. CEO, Ilya Gorelik is responsible for setting the strategic direction of the company, as well as overseeing global product development, sales and marketing.Ilya has more than 15 years of experience managing large-scale software projects and teams. He was one of the key development leaders of PTC, where he worked from 1989 to 1998, attaining the position of Senior VP of Engineering and Chief Technology Officer. From 1998 until 2000, Ilya was Senior VP of Product Strategy and Development and Chief Scientist for FirePond. From 2000 until founding Deltix in 2005, Ilya worked as Advisory CTO for HighRoads and several other software technology companies. Ilya has a Ph.D. in ComputationalMechanics from Moscow Technical University, he received an MS in Mechanical Engineering from Minsk Technical University.
\r\nPeter Hafez, Head Quantitative Research at RavenPack
\r\nPeter is an award-winning expert in the field of applied news analytics and has consulted numerous leading trading and investment firms on how to take advantage of news analytics in financial markets. Peter has more than 10 years of experience in quantitative finance with companies such as Standard & Poor's, Credit Suisse First Boston, and Saxo Bank. He is a recognized speaker at conferences on behavioral finance and algorithmic trading. Peter holds a Master's degree in Quantitative Finance from City University's Cass Business School along with an undergraduate degree in Economics from Copenhagen University.
\r\nNabyl Charania, Managing Director at Rokk3r Labs
\r\nNabyl is a Co-Founder and Managing Director at Rokk3r Labs. Rokk3r Labs is Miami based Venture Capital firm with 35 employees. Rokk3r Labs fuses entrepreneurial and professional talent to help entrepreneurs create ideas, prototypes, products and generally invests in disruptive companies designed for the modern hyper-connected world. Prior, he was a Founder and Managing Director at Decipher Labs Inc. He graduated from University of Waterloo in 2000.
\r\nZeid Barakat, Co-Founder at Flyberry Capital LLC
\r\nZeid is a Co-Founder of Flyberry Capital a Boston based hedge fund that utilizes a Big Data strategy. Zeid works with ‘best-in class’ MIT & Harvard computer scientists and machine learning experts to develop proprietary trading strategies, focused on event-based arbitrage. In charge of dfining corporate growth strategy,business development, marketing, managing Board of Advisers, and fundraising. He is an Entrepreneur in high-tech companies, focused on novel approaches to biotechnology, healthcare and financial services. He earned MBA degree in General Management and Entrepreneurship from MIT in 2008.
\r\nBattleFin is a tournament platform that crowdsources the world's best investment talent. The firm uses rapid fire, real capital tournaments to democratically identify up and coming investment talent. BattleFin has recently been featured in Bloomberg Business Week in an article titles \"The Hedge Fund Hunger Games\". The firm is passionate about leveling the playing field in finance so that anyone with an internet connection can participate in its tournaments. The firm specializes in finding the hedge fund managers of tomorrow. To learn more about BattleFin visit BattleFin.com.
\r\nFounded in 2005 and with more than 50 staff, Deltix has established itself as a leader in the growing domain of software and services for quantitative research and systematic automated trading. The Deltix Product Suite is an end-to-end platform for all phases of the alpha discovery and trading life-cycle; including data collection and aggregation, model development, back-testing, optimization, simulation, and deployment to production trading.
\r\nFinancial firms are overloaded with information and have turned to computers to read news and other media. What would take days for an investment professional to read and interpret takes computers only a few milliseconds. Now financial institutions can react much faster to the ever increasing amounts of news and information available for making investment decisions. Powered by a proprietary text analysis platform, RavenPack the tournament data sponsor, analyzes novel and relevant stories published by major news sources to look for key scheduled and unexpected geopolitical, macro-economic and corporate events, topics and opinions that indicate changes in market sentiment. Sampled news sources represent the most reliable and authoritative publishers of business and financial news.RavenPack continuously analyzes relevant information from Dow Jones Newswires, regional editions for the Wall Street Journal, and Barron’s to produce real time news sentiment scores and events from entities across multiple asset classes, including currencies, commodities, organizations, companies, sectors, and industries.
", "isPublished": false, "hasIsPublished": false, "postTitle": "", "hasPostTitle": false, "mimeType": "", "hasMimeType": false }, { "name": "Evaluation", "content": "Submissions are evaluated by the mean absolute error between the predicted percentage change and the actual percentage change.
$$
\\textrm{MAE} = \\frac{1}{n} \\sum_{i=1}^n | y_i - \\hat{y}_i |
$$
Each line should have the predicted percentage change for the given FileId. A header is required. Refer to the sample submission to see the exact format. The format looks like this:
\r\nFileId,O1,O2,O3,...,O198\r\n201,0,0,0,...\r\n202,0,0,0,...\r\n...\r\n510,0,0,0...\r\n", "isPublished": false, "hasIsPublished": false, "postTitle": "", "hasPostTitle": false, "mimeType": "", "hasMimeType": false }, { "name": "data-description", "content": "
For this competition, you are asked to predict the percentage change in a financial instrument at a time 2 hours in the future. The data represents features of various financial securities (198 in total) recorded at 5-minute intervals throughout a trading day. To discourage cheating, you are not provided with the features' names or the specific dates.
\r\ndata.zip - contains features for 510 days worth of trading, including 200 training days and 310 testing days
trainLabels.csv - contains the targets for the 200 training days
sampleSubmission.csv - shows the submission format
Each variable named O1, O2, O3, etc. (the outputs) represents a percent change in the value of a security. Each variable named I1, I2, I3, etc. (the inputs) represents a feature. The underlying securities and features represented by these anonymized names are the same across all files (e.g. O1 will always be the same stock).
\r\nWithin each trading day, you are provided the outputs as a relative percentage compared to the previous day's closing price. The first line of each data file represents the previous close. For example, if a security closed at $1 the previous day and opened at $2 the next day, the first output would be 0, then 100. All output values are computed relative to the previous day's close. The timestamps within each file are as follows (ignoring the header row):
\r\nLine 1 = Outputs and inputs at previous day's close (4PM ET)
Line 2 = Outputs and inputs at current day's open (9:30AM ET)
Line 3 = Outputs and inputs at 9:35AM ET
...
Line 55 = Outputs and inputs at 1:55PM ET
You are asked to predict the outputs 2 hours later, at 4PM ET.
", "isPublished": false, "hasIsPublished": false, "postTitle": "", "hasPostTitle": false, "mimeType": "", "hasMimeType": false }, { "name": "rules", "content": "\r\n\t\r\nYou cannot sign up to Kaggle from multiple accounts and therefore you cannot submit from multiple accounts.
\r\n\r\n Privately sharing code or data outside of teams is not permitted.\r\n It's okay to share code if made available to all participants on the forums.\r\n
\r\nKaggle and the competition host have the right to publicly disseminate any entries or models.
\r\n\r\n Winning solutions need to be made available under\r\n a popular OSI-approved license\r\n in order to be eligible for recognition and prize money.\r\n
\r\n\r\n Prizes will be awarded after the winners have posted their solutions to the competition forum.\r\n Winners must post or link to their solutions within seven (7) days of being notified of their winning status.\r\n
\r\nTeam mergers are not allowed in this competition.
\r\n\r\nThe maximum size of a team is 1 participant.
\r\nYou may submit a maximum of 5 entries per day.
\r\nYou may select up to 2 final submissions for judging.
\r\nStart Date: 8/16/2013 5:25 PM UTC
\r\nMerger Deadline: None
\r\nEntry Deadline: None
\r\nEnd Date: 10/1/2013 11:59 PM UTC
\r\n\t\r\nAugust 16, 2013: Competition launched
\r\nOctober 1, 2013: Predictive modeling phase of the Competition ends. Finalists are selected to compete at the Big Data Combine in Miami.
\r\nOctober 4, 2013: 3 minute video presentation on model and data scientists background to be uploaded to BattleFin website.
\r\nOctober 24, 2013: Hotel check in at the Loews.
\r\nOctober 25, 2013: Combine starts at 1pm at 320 Lincoln Rd, Miami Beach, FL 33139. The Combine will end at 7pm.
\r\nOctober 25, 2013: Big Data Combine Tournament Trophy is presented to the Combine winner.
\r\nOctober 27, 2013: Big Data Combine winner receives log in and password for Deltix's award winning Quant-Office strategy development and backtesting software. If a strategy is designed using the predictive model by December 15th, 2013 that meets the goals and objectives of the BattleFin's capital provider than BattleFin and the Big Data Combine Winner may enter into an agreement to license the strategy from the data scientist.
", "isPublished": false, "hasIsPublished": false, "postTitle": "", "hasPostTitle": false, "mimeType": "", "hasMimeType": false }, { "name": "Prizes", "content": "Receipt of prizes is contingent on participation in the Big Data Combine in Miami on October 25, 2013. You may still participate if you do not wish to attend, but will not be eligible for prizes.
\r\nThe top 10 finishers who are interested in attending the Combine will receive a complimentary one-night stay at the prestigious Loews hotel in South Beach, plus a travel stipend of $500 to attend the Miami event. This is intended to cover airfare and transportation to and from the airport.
\r\nIn addition, the top 5 finishers will split a $10,000 prize pool:
\r\nU.S. participants will then be eligible to license their strategy for a $1,000,000 investment account based on their predictive models. The agreement for the licensing deal will be mutually agreed between the Big Data Combine Winner and BattleFin.
\r\n\r\n", "isPublished": false, "hasIsPublished": false, "postTitle": "", "hasPostTitle": false, "mimeType": "", "hasMimeType": false } ] }