Real Time Bidding Optimization In Computational Advertising

How to make the best match between advertisers and customer under budgetary constraint is an eternal topic for business. Computational advertising improved the algorithm design and became a new battlefield of marketing. Real-Time-Bidding based display advertising is the most advanced tool in next 20 years, a market as huge as $ 9 billion per year with over 100 billion auctions traded every day need to be further analyzed. As a game with incomplete information, most existing papers regarding RTB auction is based on Ad exchanger\u27s view how to maximize seller\u27s revenue. This research is aim to formulate deep reinforcement learning in a second priced auction, use macroeconomic models such as auction theory and game theory to make prediction on the market price from the historical data, in order to optimize the sponsors\u27 utility