A trading system in R, based on technical (chart) analysis. The LONG (buy) or SHORT (sell) decision was based on a combination of the Simple Moving Average… · More and the Relative Strength Index. The model was optimized using trend signals and stopp-loss rules.
The project's goal was to get familiar with R and demonstrate what can be done in R with financial data.
We could not realize any profits with our model.
- technical trading indicators
- writing code in R
Incentive users to park shared cars at certain spots
Product Manager · Learnings:
- interdisciplinary team work - communicating through mock-ups and wireframes - writing a business… · More plan - communicating with external stakeholders
EffLocate is a white-label mobile application for car-sharing providers. Besides common features such as finding and reserving a car, EffLocate incentivizes users to park cars at certain spots within a city, where the likelihood for a new ride to take place is highest. These spots are identified through an algorithm that takes into account various data sources (see image).
Customers are operators of carsharing fleets (such as BMW DriveNow, car2go, etc.).
EffLocate helps customers make more money by (1) minimizing the need to manually relocate cars and (2) increasing accessibility for customers thus reducing idle time
Lead Researcher · This project (Master's thesis) set out to answer the question which user characteristics on microblogs relate to the quality of… · More investment advice.
Twitter has become the latest wire of Wall Street! Thousands of users tweet about stocks. Many of them give (implicit) advice whether to buy, hold, or sell a specific stock. Like in the real world, some users may be better at giving advice than others.
The results of this work help individuals and (financial) institutions to identify higher quality sources of information, ultimately improving their return on investment.
Steps: 1. Development of hypotheses such as 'users with more followers give better investment advice', or 'users who tweet about many different industries give worse investment advice'
2. Data collection: I used the Twitter (Streaming) API to collect 16M stock-related tweets over a 3-month period and Thomson Reuters' database for collecting stock data
3. I developed a Naive Bayes classifier to assess the sentiment of each tweet (Buy, Hold, Sell)
4. I calculated each user's investment advice quality over time and tested my hypotheses