Trade time measures in high speed trading markets
Dan Bernhardt's presentation will encompass three papers, with primary focus on, “Market microstructure and the cross-section of systematic volatility” (with Yashar Barardehi (Ohio University) and Tom Ruchti (Carnegie-Mellon University ). An abstract for this paper is as follows:
“We investigate the relationships between systematic risk factors and trade-time return volatilities in high frequency markets, and how they vary with market conditions. We first document that risk factor loadings and firm characteristics---historically identified to explain the cross-section of first moments of stock returns at low frequencies---also systematically explain the cross-section of high frequency normalized trade-time return volatilities, and that factor loadings and characteristics tend to be more positively associated with normalized trade-time return volatilities in more active markets.
We use the normalized trade-time return volatilities to construct a test of the market-microstructure invariance hypothesis that does not require observations of primitive trading decisions. We uncover systematic violations of invariance that rise when the activity levels of market conditions being compared differ by more. We close the loop of our analysis by showing that the magnitudes of deviations from invariance are well-explained by systematic risk."
Attached also are two background papers, "Decomposing the dynamics of intraday trading activity and trading outcomes" (with Yashar Barardehi),
Abstract: "Separating trading activity from trading volume, we find that although calendar-time return volatility follows a U-shaped pattern over the trading day, the return volatility associated with trading stock-specific fixed-dollar positions falls: higher calendar-time volatilities near close solely reflect greater trading volumes. Controlling for time-of-day, trade-time return volatility and price impacts are inversely related to trading activity. Trading activity, not trading location or order type, underlies these patterns. Unexpected trading activity drives patterns early in the day, while expected activity matters later. Our findings support predictions of models featuring strategic inventory-rebalancing traders, and hint at the importance of imperfectly-competitive liquidity provision."
and "Trade-time Based measures of Liquidity" (with Yashar Barardehi (Ohio University) and Ryan Davies (Babson College))
Abstract : "Traditional measures of stock liquidity have become noisier following dramatic changes in liquidity provision post decimalization. We develop stock-specific liquidity measures that control for short-term variations in liquidity supply and trading activity. Our trade-time based measures capture per-dollar price impacts of fixed-dollar positions. Our measures outperform standard measures including bid-ask spreads and Amihud's (2002) measure, especially in recent years. Post-decimalization, expected trading costs still explain the cross-section of expected returns for NYSE-listed stocks: we obtain monthly liquidity premium estimates of 5.3bp for expected returns and 2.7bp for risk-adjusted returns. Moreover, estimated liquidity premia rise after the implementation of Reg-NMS."