Portfolio transitions are very different from conventional asset management, but the judicious use of algorithms can still be hugely beneficial. Chris Jackson, Director, Execution Sales, Merrill Lynch International explains the nuances of using algorithms as part of transition management.
Director, Execution Sales, Merrill Lynch International
Unlike active managers, transition managers are not paid to take risk, but to minimise it. Whether acting as an advisor, executing agent or Fiduciary, a transition manager's primary purpose is to manage the risks and costs associated with any large-scale restructuring of a fund's assets. At Merrill Lynch, we have spent a lot of time working with our transition clients on refining algorithmic trading techniques to reduce the risks and costs associated with their business.
Low Risk Trading
At the execution stage of a transition, the transition manager needs to design a trading strategy around the assets of the fund and the needs of the underlying client; how long can the transition take, how much risk is the fund willing and able to take on, how liquid and volatile are the assets, can the fund use hedging instruments? These are just some of the questions that dictate strategy. The trading strategy will normally include a clearly specified execution benchmark, invariably either implementation shortfall (arrival price), market average e.g. VWAP) or closing benchmark. It is our experience at Merrill Lynch that in the appropriate situation, an algorithm tends to be far better at trading towards this type of benchmark than a manual trading strategy. The algorithm will constantly measure market metrics such as volumes, momentum and volatility and adjust its trading strategy to stay as close to the target as possible. It is often simply not possible for an individual to make that many real-time calculations across a portfolio of names.
With algorithmic trading, not only does performance vs. the benchmark tend to improve in absolute terms but the volatility or variance of performance around that benchmark also reduces significantly. An active manager may encourage a trader to take a strong view and accept performance that is +/- 50 bps around the benchmark, trusting that there will be more up days than down. However, being minus 50bps on the NAV of a fund because "the market felt good" but wasn't, is not a conversation most CIOs, trustees and clients want to have with their transition manager. Algorithms use a clear quantitatively driven process, which can deliver consistent high performance, making them a natural fit for this segment of the market.
One of the most exciting developments in transitions in recent times at Merrill Lynch has been portfolio algorithms. Here, the trading strategy is generated by a risk-based optimisation process such that portfolio VaR (Value at Risk) is reduced more quickly and efficiently than via alternative algorithmic or manual trading strategies: risky assets are liquidated more quickly than less risky assets within specified market impact constraints. This type of portfolio-based strategy is popular with many types of dealer but is particularly useful for Merrill Lynch's transition clients given that, in order to reduce market impact, trading will tend to take place over extended periods leading to potentially higher risk. By reducing portfolio risk more efficiently, portfolio algorithms allow a dealer to extend the trading period and so adopt a more passive low market impact trading style, rarely crossing the spread and maximising crossing opportunities. An example of a risk-efficient, trading schedule for the theoretical transition of a fund out of FTSE 100 and into a Eurostoxx 50 fund is shown in Figure 1.
Figure 1Fund Transition trading Schedule.
Source: Merrill Lynch Global Equity Analytics
Transition clients should be paid for providing liquidity
The point about an ideal transition trade being essentially passive is an important one. Typically these trades are not executed to capture fast alpha but rather as part of a long term restructuring of a fund, potentially decided on and executed over many months. We believe a transition should therefore be seen as provider to (rather than a taker of) liquidity from the market, rarely paying a premium for liquidity by crossing the spread. In addition to developing portfolio based algorithmic strategies, we have also customised the behaviour of the transition algorithm at stock level on the order book to reflect the passive trading style of a typical transition. Where available, our 'I Would If I Could' algorithm will simultaneously work an order in the market whilst polling alternative crossing opportunities at ECNs and dark pools in search of additional liquidity. These algorithms effectively allow transition dealers to leverage their passive trading style and get paid - in the form of better pricing -for providing liquidity to the market. This benefit accrues directly to the underlying client.
How much will this cost?
There's nothing worse than a job that ends up costing twice as much as you thought. In transitions, accurate pre-trade cost estimation is often difficult. A market impact model's trading assumptions e.g. strategy, participation rate, aggression levels are rarely understood and delivered at the point of execution, often leading to significant differences between the expected and realised costs of a transition. The assumptions in the Merrill Lynch cost model become parameters in the algorithmic engine, thereby dovetailing the predicted trading style far more closely with reality.
Keep it quiet
At Merrill Lynch, we understand that clients in transition need absolute confidentiality around their process and order flow. However, transitions tend to be very large and
"There is no doubt that the introduction and continued
evolution of algorithmic trading techniques has revolutionised the
trading strategies employed by transition managers. Whilst it is
important to analyse and filter the deal flow that can be committed
to an algorithm, the range of strategies available means that
algorithms now offer a quick and efficient platform for trading a
large volume of stock in a risk controlled manner. These
efficiencies are reflected in superior execution outcomes so that
ultimately the main beneficiary is the client."
Peter Walker, Head of EMEA Transition Management, BlackRock
can often be the result of a very public event - merger of asset managers, award of a new mandate, restructuring of an investment trust. Algorithmic trading tools offer one of the most discreet access points to the market - no traders, no information leakage, no IOIs or blocks being marketed, just an algorithm, working hundreds of smaller less obvious orders passively in the market.
The right tools for the job
Finally, one point needs to be emphasised, Algorithmic trading
strategies offer excellent opportunities to improve trading
performance and reduce risk in transitions. However, it would be
a mistake to claim that algorithms offer a one-stop solution for
all stocks in all markets. The skill of the transition team is in
screening the portfolio and deciding which names do and don't
lend themselves to trading in an algorithm. To support this, all
trades routed to Merrill Lynch's algorithms are automatically
pre-screened for suitability. If we don't think the stock is
suitable, we'll recommend an alternative venue e.g. through a
sales trader. A critical part of transition trading is being able
to accurately estimate and then effectively control the risks and
costs associated with a transition. By careful use of algorithmic
trading tools, transition dealers are able to better control
risk, access more liquidity and better maintain anonymity to the
of the underlying client.