COVID-19 outbreak brought unprecedented changes to people?s lives and made significant impact on the US and world economy. It wrought havoc on livelihood, businesses and ultimately the economy. Understanding how the sentiment on economy is changing and main factors that drives the change will help the public to make sense of the impact and generating relief measures. In this paper we present a novel Na?ve Bayes model using a word-based training approach to perform the analysis and determine the sentiment of Twitter posts. The novelty of this methodology is that we use labelled set of words to classify the tweets to perform sentimental analysis as opposed to the more expensive methods of manually classifying the tweets. We then perform analysis on the resulting labelled tweets to observe the trend of economy from February 2020 to July 2020 and determine how COVID-19 impacted the economy based on what people posted on Twitter. We found our data was largely inclined towards negative sentiment indicating that the economy had been largely negatively impacted as a result of COVID-19. Further, we correlate the sentiment with the stock market index aka Dow Jones Industrial Average (DJIA) because stock market movement closely mirrors the economic sentiment and is shown as one of the main factors influencing people's attitude change from our sentimental analysis. We found strong correlation between the two, indicating stock market change is one of the driving factors behind people's opinion change about economy during pandemic. This work proposed and tested a generic lower-cost text-based model to analysis generic public?s opinion about an event which can be adopted to analyze other topics.