A popular misconception when execution algorithms first appeared was that they would completely dumb down the execution process. Institutional traders would have nothing to do, apart from sit and act as 'algo minders' awaiting some apocalyptic event that might require their intervention.
Reality has proved very different: for most institutional traders, execution algorithms have proven a valuable productivity tool. Even in relatively illiquid names, algorithms can steadily accumulate stock, thereby significantly reducing the amount of click trading the human trader has to do. This reduction in routine trading activity leaves traders free to spend more time deploying their expertise to add alpha to the execution process. In some cases, this may consist of finessing execution in instruments too illiquid/difficult for even the most sophisticated algorithm to trade. However, in many others it will consist of deploying their trading skills to fine tune the way in which an execution algorithm (or algorithms) functions.
Some traders will base this tweaking of the algo's parameters on the instinctive 'feel' they have developed through experience, while others will base it upon the output of a few favourite technical indicators. However, while some execution algorithms incorporate quantitative alpha capture modelling, it appears that relatively few institutional traders use such quantitative modelling as a guide to either adjusting an execution algo's parameters or manually trading around the algo.
Nevertheless - and obviously depending upon its predictive quality - using a quantitative alpha model for either of these activities represents an opportunity to introduce both increased alpha and rigour to the trading process. The objective here is obviously not alpha capture as a prop trader would understand the term, but instead to maximise execution alpha.
The downside to this approach is that institutional traders do not usually have the time to develop and test their own quantitative modelling; it's typically not part of their job description. Fortunately there are a few high quality third party alternatives available, such as the suite of ATM (Advanced Trading Methods) quantitative models developed by Capital Market Research and available on the Bloomberg Professional® service.
These provide a rigorous reference framework for overall market trend conditions, together with a variety of shorter term market timing tools, both of which can be applied across multiple time frames (see Figure 1). Using ATM as a guide, a trader can fine tune aggression level, change tactics, and spot opportunities during the trade to seek better average prices.
The Trend Bars1 study colours the price bars green or red to indicate whether the market is in an up or down trend; uncoloured price bars indicate that the market currently has no distinct trend. A complementary indicator of trend direction is the Trend Line2, with blue indicating an up trend and red a down trend. The strength of any directional trend detected is denoted by the colour of the Trend Strength Bars3, which are the horizontal bars in the lower pane below the price chart. Bright green or red indicate the strongest trend, while dark green or orange indicate that the up/down trend is just starting. If all three of these studies are aligned in the same direction, there is a greater certainty of a price trend in that direction.
The lower pane also shows the ATM Fast Trigger; if it turns Green (up) or Red (down) a security is likely to continue to move in the direction projected until it reaches the opposite extreme, which is typically in approximately eight price bars (this applies across all time frames). Again, the probable accuracy of any signal is increased if it concurs with the ATM Trend Lines, ATM Trend Bars and ATM Trend Strength Bars.
This combination of quantitative techniques has been extensively tested both historically and in real time. Across a portfolio of the OEX 100 stocks where the various ATM components concur as to market direction (as outlined above) it has delivered an overall prediction accuracy over an average investment cycle of eight days of close to 70%.
Putting the edge to work
Many profitable alpha capture models have a significantly lower prediction accuracy than this, so how can institutional traders harness this superior performance to deliver execution alpha? There are a number of possibilities that work well in conjunction with Bloomberg Tradebook algorithms and tools.
For example, Tradebook's B-SmartSM algorithm has a range of aggression settings. These include a Passive setting that will post intelligently, a Normal setting that will post intelligently and nibble at the spread, and an Aggressive setting that will sweep the market. The output of the ATM models can be used by the trader to select the most appropriate setting for the current market conditions, as well as providing early warning of when the situation is likely to change and further adjustment of the B-SmartSM aggression setting may be necessary.
For example, if working a buy order while most of the ATM trend indicators are pointing down, it makes little sense to pay the spread. As the market declines, the B-SmartSM aggression setting can be left at Passive so others will cross the spread to interact with you. On the other hand (and particularly if there is a measure of urgency about trade completion) if the Trend Strength Bar then disappears and the Fast Trigger turns up, it may be advisable to switch the B-SmartSM aggression setting to Aggressive and accumulate as much stock as possible before the market starts to move higher.
A common challenge for traders is knowing when to do nothing and leave well alone. The ATM models can help here too. For instance, if the various ATM trend indicators are neutral, then the oscillations of the Fast Trigger are less likely to develop into an extended price move. Based upon this information, the trader might see no need to intervene and just leave the B-SmartSM aggression setting at Normal and nibble at the market until a clear trend/direction emerges.
While the technique outlined above for adjusting the B-SmartSM aggression setting definitely adds value, it may not always be practical to use it for workflow reasons. For instance, if a trader is having to manage multiple trades simultaneously, it may not be physically feasible for them to be continually hitting the 'bump buttons' to change B-SmartSM aggression settings for a large number of stocks.
However, an alternative technique for this type of multi trade situation that is increasingly popular with Tradebook users is to use the Decrement functionality. Particularly when employed in conjunction with quantitative models (such ATM) that have proven predictive value, this can generate significant execution alpha, while still keeping workflow manageable.
The Decrement functionality4 in Tradebook is turned on by checking a box in an order ticket (see Figure 2). (If an algorithm is already running on the stock in the ticket, the box will be checked by default.) If Decrement is applied, an automatic link is created in the background between any manual orders in a stock and any algorithms currently working the same stock. So if an algorithm is handling an order for 20,000 shares and the trader enters a manual order ticket for 5,000 shares that is then filled, the Decrement functionality automatically reduces the order in the algorithm to 15,000. (If multiple algorithms are running in the stock, Decrement will reduce the size in the algorithm with the largest order.)
This functionality minimises the risk of double fills, while also sparing the trader from having to manually stop the algorithm, adjust the volume and restart the algorithm. It therefore also avoids the potential issue of the algorithm losing queue position in the trading venue's order book.
The ATM Fast Trigger5 Turning Point study displays the price level (see Figure 3) that if the market closes above/below will result in the Fast Trigger study changing direction.
The Fast Trigger Turning Point study is generated and displayed for the current price bar immediately the preceding bar closes. Therefore, if a trader creates an audible/visual alert in Bloomberg that fires if the market comes within a specified distance of the Fast Trigger Turning Point level, they will have advance warning of any situations where a change in Fast Trigger direction is likely. To focus on only high probability situations, the trader might also specify in the alert that the potential Fast Trigger turn must be in the same direction as the three ATM trend indicators.
Link this with the Decrement functionality and the B-SmartSM algorithm, and the trader has an opportunity to add serious execution alpha and substantially beat their benchmark, even across multiple simultaneous stock orders. The trader could run B-SmartSM in the background at the Passive aggression setting across all the stock orders they are currently working (if appropriate), but then supplement this by using the ATM studies to assist in adding larger block captures of stock at advantageous price points.
For example, if one of the Bloomberg alerts on the ATM Fast Trigger Turning Point that the trader has established fires, the trader has time to decide a course of action. If the price has already broken the Fast Trigger Turning Point (but the bar has not yet closed) they may opt to simply place a substantial limit order or a B-SmartSM aggressive limit order close to the Fast Trigger Turning Point price level. If it fills before the market takes off, they have a high probability of capturing a significant basis point advantage that will benefit the average execution price for the order.
Tradebook's execution consultants are experts that help traders with the ATM studies. They educate and explain the indicators to help traders spot execution opportunities and, most importantly, consult with traders on appropriate algorithms and settings to optimise and seek to lower implicit costs of execution.
While the ATM studies already have an established reputation among prop traders from all types of institution, they also offer tangible execution alpha opportunities for institutional traders. These opportunities can be further augmented when the studies are used in conjunction with algorithms such as B-SmartSM and functionality such as Decrement.
The net result of this combination is a win/win for the trader enabling them to seek faster order completion and improved execution price.