The Gateway to Algorithmic and Automated Trading

Auto ADRs

Published in Automated Trader Magazine Issue 26 Q3 2012

Certain mandate restrictions can hugely increase frictional costs for asset managers. A classic example is where US managers can only hold ADRs as opposed to directly holding foreign stocks, and have historically been hit hard by instrument conversion costs and FX dealing spreads. Joseph Wald, Managing Director at Knight Capital Group, shows how multi asset algorithmic execution can massively reduce these costs.

Joseph Wald

Joseph Wald

The Scenario: It is the afternoon of Tuesday April 10, 2012. The portfolio manager of a New York based asset management firm has been tracking the performance of Nokia with increasing concern since taking a 10 million share position in December 2011. The position was acquired for an average price of $4.59 (close to Nokia's low for 2011) by taking advantage of the stock's decline on the back of bearish brokers notes and a sympathetic slide with Research In Motion's posting of a sharp profit drop. As a result, while the position is currently trading at a profit, the manager feels that much is wrong with the company and that it has been strategically drifting for too long.

In his view, Nokia's woes started with its epic wrong footing over smartphones, which effectively confined it to the budget space as a result of its failure to match leading contenders such as Apple and Samsung as well as others, such as HTC. According to IDC figures, Nokia's total market share fell 7.9 percent from 2010 to 2011, but its smartphone figures were dire - a fall of 22.8 percent over the same period (one in which Samsung posted annual smartphone sales growth of 310.5 percent and Apple of 96.2 percent). The marriage with Microsoft has also effectively constrained Nokia's products, because it can't yet offer support for dual-core processors and higher resolution screens on smartphones.

The scale of the damage done by Apple is little short of epic. Nokia might once have been the globe's leading maker of smartphones, but it has shed $80bn (80 percent) of its market value since the 2007 launch of the iPhone. The comparative underlying numbers for the two companies are similarly stark. Since the iPhone's introduction, Nokia's operating income per unit sale has collapsed from $23 to $3, while over the same period Apple's has leapt from $74 to $241 a unit. The CEO, Stephen Elop, also seems to have little influence on market sentiment, with the shares down around 50 percent since his arrival in 2010.

The manager is also increasingly pessimistic about Nokia's future. One specific concern is that the smartphone market is becoming glutted as various manufacturers see it as the ideal way to boost their volumes and margins. The problem is that this hugely increases the risk of a price war, similar to that which hit the LCD TV market. One obvious consequence of this that will affect all manufacturers is a squeeze on margins. Another, which will specifically harm Nokia, is that the prices charged for low-end smartphones will become so low as to start competing directly with "feature phones". This market is already stagnating, but it is a segment where Nokia has dominated, particularly in terms of many emerging markets where most consumers have historically been unable to afford smartphones.

After also taking into account other factors - such as the way that Nokia's restructuring process is now burning the corporate cash pile at an alarming rate - the manager decides that the time has come to close the position. He therefore instructs his sales desk to sell the 10 million shares over the next three days at a limit of $4.60.

The Asset: Nokia

The Challenge: To sell 10 million shares in Nokia during the next three trading sessions with a limit of $4.60. Even over three sessions, the volume alone is demanding because Nokia's average daily volume during US trading hours over the past year has been around 32 million shares. The price limit may prove similarly demanding, because while the price running into the close on April 10 is fluctuation around $5, underlying sentiment is fickle and a further piece of bad news from the company could trigger a sharp decline.

However, the biggest challenge by far is that the manager is only mandated to hold European stocks in ADR form. In view of the standard conversion charge levied by most brokers of five cents per share, plus the further costs attributable to the spread on FX rates offered by brokers, the trader will have some serious work to do in order to complete the trade. Logically, to have the best chance of achieving this, the asset manager's trader will want to be able to access the market for Nokia for as long as possible, which will make it essential that both European and US trading hours are available.

The Trader: The trader heads the asset manager's in-house execution desk in New York and in view of the trade's complexity and demanding nature, he elects to handle it himself. The trader has noted Nokia's slide of nearly 10 percent since the start of the month and has similar concerns as the manager about the resilience of the share price should further bad news emerge. As a result, he takes the view that the trade should be completed as quickly as possible and therefore recommends to the manager that the order should be sent out to the chosen broker tonight rather than waiting for the US open tomorrow (April 11).

Fortunately, one of the brokers the firm uses is capable of providing completely transparent multi-asset algorithmic execution for ADRs in real time. The automated currency translation element embedded in this executes against real time FX prices rather than end of day. Apart from any frictional cost savings, this is particularly important because although the position has slightly stabilised during the past two trading sessions EURUSD has fallen by almost 2.3 percent since the start of the month.

It is therefore agreed that the trade will be immediately routed to this broker in order to able to trade the stock from 3am EST (10am Helsinki time) the following day in order to be able to maximise the number of trading hours and thereby gain access to the maximum amount of available liquidity. Using this multi asset algorithmic execution workflow, the trader is also able to specify the execution algorithm to be used for stock purchases. In this case he selects a participation algorithm and in consultation with the manager chooses a 15 percent participation rate.

The Algo: The algorithm used for the trade is perhaps more accurately described as a sophisticated trade workflow process that combines stock execution using any one of the broker's algorithms with simultaneous FX translation using the broker's own FX liquidity pool. It works as follows:

• When a client submits an order, the algorithm evaluates the ADR program and the underlying ordinary security ("ORD")

• If the ADR program is in good standing, the algorithm will automatically determine the appropriate ADR/ORD conversion rate as well as the FX-adjusted limit price and will start trading

• The stock execution algorithm will complete the order as per the client's instructions while simultaneously hedging the FX exposure and also entering a cross-book order to convert the ADR to the underlying ORD or vice versa (in this example ORD to ADR)

• The client then receives an execution in the designated line of the security and the local currency of the original order

The process is continuous, and the client has visibility on executions as they occur at all times, together with data on the FX rate used to arrive at the finished ORD/ADR translation. It also delivers enhanced market access in that users are able to trade US-listed ADRs (denominated in USD), while the US equity market is closed. Two other market access benefits are the opportunity to trade thinly traded ADRs via local market liquidity and the way in which the algorithm will tap whichever is the more liquid market during the European/US market crossover period. The use of a transparent and highly liquid FX pool (average daily turnover during the last year of nearly $60bn) substantially reduces frictional costs. A further benefit is that the process automatically taps the secondary market for ORD/ADR broker cross books in order to obtain price improvement. Therefore, the client doesn't automatically end up paying the standard five cents per share conversion charge, but benefits from any savings the broker achieves there.

Therefore, in addition to the transparency benefits of the process, the client can also achieve some major savings in comparison with the cost of normal ORD/ADR translation - in many cases savings of approximately 90 percent are achievable.

Trade Schedule:

3.00am - (all times are EST) - Nokia opens in Helsinki €0.02 lower than the previous close at €3.8. 561,396 shares trade in the first minute and the participation algorithm manages to sell 83,300.

3.04am - Significant volume in the market in the last couple of minutes, with which with the algorithm interacts to dispose of 55,450 shares.

3.46am - Nokia has been trading around the dollar-equivalent of $5 for most of the past 40 minutes but has now softened slightly. Volume is currently light with shares traded per minute only hitting 100,000 twice in the past 40 minutes. Nevertheless, the algorithm has managed to sell a further 344,455 shares at an average price of $4.993.

3.51am - Liquidity has picked up, while the stock price has weakened fractionally to just below $5. The algorithm has sold another 58,325 shares in the past five minutes.

4.11am - Occasional bursts of volume in the last 20 minutes (with a total of just under 1.8mn shares traded), have allowed the algorithm to unload a further 267,465 shares, making a total so far of 808,995.

Nokia Helsinki: Accumulated intraday volume April 11th 2012

Figure 1.

Nokia Helsinki: Accumulated intraday volume April 11th 2012

Figure 2.

4.17am - Nokia has been bumping along on indifferent volume, but the algorithm has nevertheless managed to sell a further 35,055 shares.

4.47am - The stock has been around the equivalent of $5 for most of the past 40 minutes and volume has also picked up a little. Another 135,360 shares sold, making a total of 979,410 so far.

5.36am - A combination of further patches of decent liquidity and little movement in the stock have allowed the sale of a further 497,565 shares at an average price of $4.985.

7.59am - Nokia's price has firmed up and has been trading in the equivalent of a $5 to $5.04 price range since 5.36am. Volume has also been reasonable and the algorithm has unloaded a further 762,525 shares since then. With 2,239,500 shares sold so far at an average price of $4.961 the order is nearly a quarter complete.

8.00am - Mayhem - as the manager and trader's concerns about Nokia are suddenly realised. The company announces an operating loss for its mobile phone division and lowers its first-quarter profit forecast. It states that its operating margin in the first quarter for devices and services (which comprise some 60 percent of its sales) would be a loss of three percent, in contrast with its previous predictions of a margin of plus or minus two percent. The company's longer term outlook is also not encouraging, as it predicts that second quarter devices and services operating margins will be similar to or lower than the first quarter. Various reasons are given for the weak outlook including:

• Strong competition hitting the company's Smart Devices and Mobile Phones business units

• The timing, ramp-up and consumer demand relating to new products

• The macroeconomic environment

The market reaction is unequivocal. In the 60 seconds following the announcement the stock price in Helsinki drops from €3.81 to €3.464 and more than six million shares change hands. The trader immediately calls the manager and requests that under the circumstances the average and limit prices for the trade be adjusted downwards significantly and the participation rate increased. The manager quickly agrees to a new average price of $4.57, a limit of $4.25 and a participation rate of 20 percent, which the trader immediately enters into the broker's EMS.

8.04am - The delay caused by having to agree and implement these new prices actually works to the trade's advantage as the stock price has rallied slightly to above $4.60. During that time nearly 8 million shares have traded and the algorithm has managed to sell 1,951,415 shares at an average price of $4.645.

Figure 3

Figure 3

8.20am - Nearly 30 million shares have traded since 8.04am, but the price has also continued to fall to around the $4.34 level, also exacerbated by marginal weakening in EURUSD over the same period. The algorithm sells a further 4,530,980 shares during this period at an average price of $4.439. At this stage, the trade is nearly 90 percent complete with 8,721,895 shares sold at an average price of $4.619.

8.25am - The stock price is continuing to weaken and the trader makes the call that it is likely to stabilise in the latter stages of the US premarket. He therefore, with the manager's consent, temporarily suspends the algorithm.

9.00am - US premarket activity strengthens, with Nokia at $4.23. Despite volume of 287,263, US liquidity is still well behind Helsinki.

9.22am - After a worrying 20 minutes, the trader's decision to suspend the algorithm is beginning to look vindicated, as Nokia has rallied strongly. The trader reactivates the algorithm.

9.27am - Though still below Helsinki, US trading volume has continued to build significantly. In the last five minutes a total of 6,871,769 shares have traded on both markets, with the algorithm accounting for 523,765 of them at an average price of $4.341.

9.28am - US volume exceeds Helsinki volume in Nokia for the first time today.

9.33am - Strong liquidity on both markets (including 5.5 million shares trading in one minute in the US) have seen a total of more than 15 million shares change hands in the past five minutes. This coincides with a rally in the share price to $4.39, allowing the algorithm to complete the trade by selling the remaining 754,340 shares at an average price of 4.379. The average sale price achieved for the complete trade was $4.587 (see Figure 3).

The ability to trade and auto-convert from ADRs during Helsinki trading hours made a very significant difference to the average price achieved. Before the bad news broke at 8.00am EST, the trade was already nearly a quarter complete at an average price of $4.961. Even in the immediate aftermath of the news, the algorithm was able to sell nearly two million shares at an average price nearly 20 cents better than would have been possible at any time after the US market opened.

Furthermore, the average price of $4.587 for the trade would not have been achievable using a conventional ORD/ADR conversion mechanism, which for a trade of this size would probably have resulted in an average sale price approximately four cents lower ($4.547) - which on a position of this size represents a performance difference of $400,000.