Algorithmic trading ... - page 3

This paper proposes a modified version of the widely used price and moving average cross-over trading strategies. The suggested approach (presented in its 'long only' version) is a combination of cross-over 'buy' signals and a dynamic threshold value which acts as a dynamic trailing stop. The trading behavior and performance from this modified strategy is different from the standard approach with results showing that, on average, the proposed modification increases the cumulative return and the Sharpe ratio of the investor while exhibiting smaller maximum drawdown and smaller drawdown duration than the standard strategy.

We provide a practical and technical overview of volatility trading strategies:
1) The insight for the design and back-testing of systematic volatility strategies
2) Understanding of risk-reward trade-off and potential pitfalls of volatility strategies

We focus on systematic and rule-based trading strategies that can be marketed as an investable index or a proprietary strategy:
1) Delta-hedged strategies for capturing the volatility and skew risk-premiums
2) Without delta-hedge: CBOE and customized options buy-write indices

We overview important implementation aspects:
1) Measuring the historic realized volatility
2) Forecasting the expected realized volatility
3) Measuring and forecasting implied and realized skew
4) Computing option delta consistently with empirical dynamics
5) Analysis of transaction costs
6) Managing the tail-risk of short volatility strategies

This paper studies the impact of algorithmic trading (AT) on asset prices. We find that the heterogeneity of algorithmic traders across stocks generates predictable patterns in stock returns. A trading strategy that exploits the AT return predictability generates a monthly risk-adjusted performance between 50-130 basis points for the period 1999 to 2012. We find that stocks with lower AT have higher returns, after controlling for standard market-, size-, book-to-market-, momentum, and liquidity risk factors. This effect survives the inclusion of many cross-sectional return predictors and is statistically and economically significant. Return predictability is stronger among stocks with higher impediments to trade and higher predatory/opportunistic algorithmic traders. Our paper is the first to study and establish a strong link between algorithmic trading and asset prices.

We consider a multi-player situation in an illiquid market in which one player tries to liquidate a large portfolio in a short time span, while some competitors know of the seller's intention and try to make a profit by trading in this market over a longer time horizon. We show that the liquidity characteristics, the number of competitors in the market and their trading time horizons determine the optimal strategy for the competitors: they either provide liquidity to the seller, or they prey on her by simultaneous selling. Depending on the expected competitor behavior, it might be sensible for the seller to pre-announce a trading intention (sunshine trading) or to keep it secret (stealth trading).

Based on the law of large numbers, several options researchers have proposed using (different) weighted averages of the implied standard deviations, ISD, calculated from numerous options with the same expiry to obtain a single best ISD measure. However, most commercial providers use an average (and often an equally weighted average) of just a few at-the-money ISDs. We find that the practitioners' restricted averages forecast slightly better than the broader weighted averages from the academic literature but that neither group forecasts actual volatility very well.

We suggest an adjustment to the extant models which improves their forecasting ability considerably. We also suggest a new weighting scheme which yields better estimates on an out-of-sample basis than any of the existing models (adjusted or unadjusted). However, we also find that because there is little independent noise in individual options ISDs (at least in the S&P 500 options market), there is little gain to averaging. Consequently, individual option ISDs and averages of just a few ISDs forecast almost as well as weighted averages of many ISDs and the weighting scheme choice is relatively unimportant.

We consider a Nash equilibrium between two high-frequency traders in a simple market impact model with transient price impact and additional quadratic transaction costs. Extending a result by Schöneborn (2008), we prove existence and uniqueness of the Nash equilibrium and show that for small transaction costs the high-frequency traders engage in a "hot-potato game", in which the same asset position is sold back and forth. We then identify a critical value for the size of the transaction costs above which all oscillations disappear and strategies become buy-only or sell-only. Numerical simulations show that for both traders the expected costs can be lower with transaction costs than without. Moreover, the costs can increase with the trading frequency when there are no transaction costs, but decrease with the trading frequency when transaction costs are sufficiently high. We argue that these effects occur due to the need of protection against predatory trading in the regime of low transaction costs.

This paper proposes a modified version of the widely used price and moving average cross-over trading strategies. The suggested approach (presented in its 'long only' version) is a combination of cross-over 'buy' signals and a dynamic threshold value which acts as a dynamic trailing stop. The trading behavior and performance from this modified strategy is different from the standard approach with results showing that, on average, the proposed modification increases the cumulative return and the Sharpe ratio of the investor while exhibiting smaller maximum drawdown and smaller drawdown duration than the standard strategy.

We examine the role of algorithmic traders (AT) in liquidity supply and demand in the 30 DAX stocks on the Deutsche Boerse in January 2008. AT represent 52% of market order volume and 64% of nonmarketable limit order volume. AT more actively monitor market liquidity than human traders. AT consume liquidity when it is cheap, i.e., when the bid-ask quotes are narrow, and supply liquidity when it is expensive. When spreads are narrow AT are less likely to submit new orders, less likely to cancel their orders, and more likely to initiate trades. AT react more quickly to events and even more so when spreads are wide.

Research shows that evidence-based algorithms more accurately predict the future than do human forecasters. Yet, when forecasters are deciding whether to use a human forecaster or a statistical algorithm, they often choose the human forecaster. This phenomenon, which we call algorithm aversion, is costly, and it is important to understand its causes. We show that people are especially averse to algorithmic forecasters after seeing them perform, even when they see them outperform a human forecaster. This is because people more quickly lose confidence in algorithmic than human forecasters after seeing them make the same mistake. In five studies, participants either saw an algorithm make forecasts, a human make forecasts, both, or neither. They then decided whether to tie their incentives to the future predictions of the algorithm or the human. Participants who saw the algorithm perform were less confident in it, and less likely to choose it over an inferior human forecaster. This was true even among those who saw the algorithm outperform the human.

This paper tests the accuracy of the commonly used cutoffs for determining the statistical significance of autocorrelations in time series. Monte Carlo simulations with 50,000 replicates were used to generate 95% confidence limits by varying sample size from 21 to 252 using both normally distributed and t-distributed data. The simulations show that the confidence limits derived from the commonly used formulas are biased at sample sizes of less than several hundred and should not be used.