SNTMNT (sentiment) launches a Trading Indicator API that gives stock price predictions based on (financial) Twitter sentiment for all the S&P 500 funds. Our stock price predictions are based upon machine learning algorithms and have an accuracy as high as 60 percent, with an average of 54 percent. The predictions take away extra noise in asset pricing, and can be used by investors as an additional trading indicator on top of fundamental analysis and technical analysis. As far as we know, SNTMNT is the first in the world to provide such a service.
We are inspired by the work of professor Johan Bollen, who found correlations between the Dow Jones Industrial Average and Twitter sentiment and formed the basis for Derwent Capital. There is a big difference though between his models and ours. Where Bollen focuses on a very macro level of Twitter sentiment (general mood), we believe that more value can be extracted by focussing on a more micro level. The SNTMNT predictions are based on very specific fund related Twitter sentiment through stock tickers, in the way StockTwits does, and brand related sentiment (name, products, CEO).
We are offering the following tools:
Trading Indicator API
The API gives hourly and/or daily buy & sell signals for all S&P 500 stocks based on online sentiment on Twitter. Our stock price predictions are based upon various machine learning algorithms and have an accuracy as high as 60 percent with an average of 54 percent. Each signal is accompanied by a confidence interval.
Financial Sentiment API
Sentiment is hard to classify in financially themed Twitter messages, mainly because financial markets speak their own language of fear and greed. Our Financial Sentiment API is specifically trained to deal with unstructured context and financial jargon, and has an accuracy of 84.3% on a binary scale and an accuracy of 76.8% on a three point scale. This is an outperformance of about 10 percent over the best generic models that we know exist.
Brand Sentiment API
Classify sentiment in social media messages surrounding brands. Our algorithm is specifically tailored to brands, products & company information. It has an accuracy of 84.7% on a binary scale and an accuracy of 86.9% on a three point scale.
SNTMNT is a technology company founded at Startup Weekend, specializing in natural language processing (NLP) and text analysis for social media. We are developing online sentiment analysis and prediction tools that help investors make better investment decisions.
SNTMNT is founded by Vincent van Leeuwen, Kees van Nunen, Durk Kingma and Tim Harbers. Vincent and Kees have a background in behavioral finance and have done work for ABN AMRO, and Durk and Tim have a background in artificial intelligence and have done peer-reviewed research at NYU and Cornell.