Deep Landscape Forecasting in Multi-Slot Real-Time Bidding.

Real-Time Bidding (RTB) has shown remarkable success in display advertising and has been employed in other advertising scenarios, e.g., sponsored search advertising with multiple ad slots. Many current RTB techniques built for single-slot display advertising are thus no longer applicable, especially in the bid landscape forecasting. Landscape forecasting predicts market competition, including the highest bid price and winning probability, which is preliminary and crucial for the subsequent bidding strategy design. In the multislot advertising, predicting the winning prices for each position requires a more precise differentiation of bids among top advertisers. Furthermore, defining the winning probability and addressing censorship issues are not as straightforward as in the case of a single slot. In view of these challenges, how to forecast the bidding landscape in the multi-slot environment remains open. In this work, we are the first to study the landscape forecasting problem in multi-slot RTB, considering the correlation between ad slots in the same pageview. Specifically, we formulate the research topic into two subproblems: predicting the distribution of the winning price and predicting the winning probability of the bid price for each position. Based on the observation from the production data and survival analysis techniques, we propose a deep recurrent model to predict the distribution of the winning price as well as the winning probability for each position. A comprehensive loss function is proposed to learn from the censoring data. Experiments on two public semi-synthetic datasets and one private industrial dataset demonstrate the effectiveness of our method.