Does informed trading amplify liquidity shocks? We study a trading model with informed traders who have superior information on fundamental value as well as on a noise trading process. If the noise trading process is persistent, the informed trader amplifies liquidity shocks in order to better camouflage his trading activities in the future. We find that the pattern of liquidity forms an inverted U-shape over the trading period because the informed trader reverses his trading pattern near announcements. We also provide some implications on (i) dual capacity trading and (ii) isolation of proprietary trading from market-making desks. We test our model predictions using estimated noise trading processes based on a machine learning technique which captures market sentiment expressed in textual data.
“Perfect” calibration occurs when, at any given date, a surface of model derived option prices perfectly matches the surface of market option prices, model parameters are thus "perfect" with respect to market conditions on that date. I test whether stochastic volatility models can be consistently “perfectly” calibrated using historical market data. To this end I calibrate, using a particle swarm optimization algorithm, four well know models: Heston, Bates, Barndorff-Nielsen-Shephard and the Normal Inverse Gaussian – Cox-Ingersoll-Ross to a large set of S&P 500 option chain, dividend yield and interest rate data. Under the condition of “perfect” calibration, I examine which models produce the most accurate deltas i.e. I analyze how reliable individual model predicted probabilities of being in the money are by comparing ex-ante deltas with ex-post price realizations.