By James Hilton, Co-Head of AES Sales, EMEA, Credit Suisse
In economics, a perfectly competitive market must satisfy a number of conditions. There can be no information asymmetries, participants must not be able to unduly influence prices, the cost of entry and exit must be zero, and all transactions must be voluntary and mutually beneficial to both parties. As this economic nirvana remains out of our grasp, buy-side and sellside trading desks have a job to fulfill. We operate in an imperfect market, where short-term price volatility is commonplace and the cost of trading can be substantial. The use of smart trading methods is ever more important, especially where the trading goals of different participants vary so considerably.
Transaction cost analysis has helped many diligent investment firms adapt and improve their trading style to suit different market environments. Both buy- and sell-sides have built models to predict market impact, often adjusting their trading participation based on their findings. For example, we have seen a marked shift into more passive trading on our orders, presumably to reduce market impact. However, it is often difficult to pinpoint exactly what causes impact on an order book, and therefore how to adjust one’s trading style accordingly. Impact should be about not only what happens after a trade, but also not trading at the wrong price in the first place.
In imperfect markets, traders take advantage when they perceive a price to be cheap, reacting to – and correcting –short-term mispricing. Indeed, many brokers now offer tools to adjust trading participation based on the “fair value” of a stock when compared to either its own recent trading patterns or an index benchmark. These work well on an intraday basis and help buy-sides avoid (or take advantage of) obvious mispricing. However, we also suspect that some participants/strategies in the market actively seek to create short-term mispricing and use it to their advantage. This has been the subject of our most recent research.
The debate over high frequency trading (HFT) continues unabated. In our view, HFT isn’t in itself a bad thing; however, certain strategies may be detrimental to the market. We have taken a practical approach to the different types of HFT. We measure certain behaviours, determine the best way to detect them in a very short time-frame and then set out to adjust our strategies so that they trade in the most optimal manner.
One of the most visually impactful examples of detrimental shortterm trading is quote stuffing, whereby an order book is flooded with huge numbers of orders and cancellations in rapid succession (see Exhibit 1). Our AES Quant team uses techniques adapted from signal processing – including real time burst detection and pattern recognition – to detect this sort of behaviour. For the STOXX600 universe in Q3 2012, we found that stocks experienced 18.6 incidents a day on average, with more than 42% of stocks averaging 10+ events per day. Though we can’t always identify the exact strategy behind these patterns, we would want to adjust our algorithms in response.
“Layering” provides another practical example. This occurs when a number of orders are placed on an order book – or across alternative trading venues – with no intention that those orders ever execute. Many traders will have tried to lift an offer from the consolidated order book, only to find that suddenly liquidity has “faded”. Exhibit 2 shows the likelihood of orders fading immediately after a partial or full “take”. There are definite steps that can be taken to counteract this activity, and Credit Suisse, as well as other firms, have developed a range of options to achieve better fill rates.
The ability to detect these sorts of behaviours is just the first step. Being able to engineer trading strategies to be cognisant of them and adapt accordingly will begin to set broker algorithms apart. Ultimately, only those algorithms that take into account all factors of market impact will succeed. Imperfect markets create a challenge for traders; having the right tools is essential.
In economics, a perfectly competitive market must satisfy a number of conditions. There can be no information asymmetries, participants must not be able to unduly influence prices, the cost of entry and exit must be zero, and all transactions must be voluntary and mutually beneficial to both parties. As this economic nirvana remains out of our grasp, buy-side and sellside trading desks have a job to fulfill. We operate in an imperfect market, where short-term price volatility is commonplace and the cost of trading can be substantial. The use of smart trading methods is ever more important, especially where the trading goals of different participants vary so considerably.Transaction cost analysis has helped many diligent investment firms adapt and improve their trading style to suit different market environments. Both buy- and sell-sides have built models to predict market impact, often adjusting their trading participation based on their findings. For example, we have seen a marked shift into more passive trading on our orders, presumably to reduce market impact. However, it is often difficult to pinpoint exactly what causes impact on an order book, and therefore how to adjust one’s trading style accordingly. Impact should be about not only what happens after a trade, but also not trading at the wrong price in the first place.
In imperfect markets, traders take advantage when they perceive a price to be cheap, reacting to – and correcting –short-term mispricing. Indeed, many brokers now offer tools to adjust trading participation based on the “fair value” of a stock when compared to either its own recent trading patterns or an index benchmark. These work well on an intraday basis and help buy-sides avoid (or take advantage of) obvious mispricing. However, we also suspect that some participants/strategies in the market actively seek to create short-term mispricing and use it to their advantage. This has been the subject of our most recent research.
The debate over high frequency trading (HFT) continues unabated. In our view, HFT isn’t in itself a bad thing; however, certain strategies may be detrimental to the market. We have taken a practical approach to the different types of HFT. We measure certain behaviours, determine the best way to detect them in a very short time-frame and then set out to adjust our strategies so that they trade in the most optimal manner.
One of the most visually impactful examples of detrimental shortterm trading is quote stuffing, whereby an order book is flooded with huge numbers of orders and cancellations in rapid succession (see Exhibit 1). Our AES Quant team uses techniques adapted from signal processing – including real time burst detection and pattern recognition – to detect this sort of behaviour. For the STOXX600 universe in Q3 2012, we found that stocks experienced 18.6 incidents a day on average, with more than 42% of stocks averaging 10+ events per day. Though we can’t always identify the exact strategy behind these patterns, we would want to adjust our algorithms in response.
“Layering” provides another practical example. This occurs when a number of orders are placed on an order book – or across alternative trading venues – with no intention that those orders ever execute. Many traders will have tried to lift an offer from the consolidated order book, only to find that suddenly liquidity has “faded”. Exhibit 2 shows the likelihood of orders fading immediately after a partial or full “take”. There are definite steps that can be taken to counteract this activity, and Credit Suisse, as well as other firms, have developed a range of options to achieve better fill rates.
The ability to detect these sorts of behaviours is just the first step. Being able to engineer trading strategies to be cognisant of them and adapt accordingly will begin to set broker algorithms apart. Ultimately, only those algorithms that take into account all factors of market impact will succeed. Imperfect markets create a challenge for traders; having the right tools is essential.
