email us
Hedge Fund Technology
July/August 2002
VIEWPOINT - MATTHEW SKINNER - XENOMORPH
 |
Taming the bond market
Despite shrinking margins due to an influx of activity in bond trading, Xenomorph's Matthew Skinner believes that using the right kind of disciplined methodology, hedge funds can still achieve good returns with convertible bonds.
THE CONVERTIBLE BOND market has been growing strongly for many years and has attracted many market participants along the way. It has risen from being an esoteric corporate financing instrument, tailored to appeal to certain types of fixed income fund, to a liquid and actively traded mainstream market that routinely practices some of the most sophisticated strategies observed in traded markets.
Indeed, it is has been largely responsible for many of the financial innovations of the past decade, including the growth of the credit default swap market, the maturing of the securities lending industry, and of course the increasing sophistication of the bond itself.
This maturity now presents the market with the risk of being a victim of its own success as traders jostle daily with each other in the same markets and the same instruments chasing ever-smaller margins. The volatile results of some of the funds last year adds grist to the mill for those arguing that supply in the bond market now exceeds the underlying cash (equity) market sufficiently to remove forever the kinds of profits seen in the strongest funds of recent years.
Some do not subscribe to this argument. As with all mature markets, profits exist by necessity, and trading success will be transferred to the most mature players: those that practice rigorous and systematic methods. Large markets offer liquidity on a wide variety of instruments, giving rise to sophisticated hedging tools and the potential for portfolio insurance and prudent risk management; all of which are critical for the long-term success of prudent funds.
 |
Convertible Bond Arbitrage
The one common observation of all successful funds and banks is that there is a huge appetite for information. Hardware and software has evolved exponentially over the last few years, and it is now easily possible to store the terms and conditions for every bond traded globally on a small laptop computer, including market prices, analytics and theoretical calculations. It is to this area that we now focus to show what is done and why.
The criteria for true convertible bond arbitrage (the creation, through trading, of a riskless profit) rarely exist anymore, and all players in the market occupy most of their time with spread trading. They will determine, using analysis and experience, that there exists a mispricing in some given parameter. There are many parameters that are studied, but volatility and credit spread are most commonly analysed, with interest rates, dividend forecasts, company financial data (share price), borrow costs, sector analysis to name but a few runners up.
The trader then devises a set of trades (a strategy) to take advantage of this mispricing for the duration of the mispricing, or for when it comes to an end, or both. Determining and timing such values, setting stop-loss limits, and managing the overall risk of the portfolio all demand a great deal of market data. It is for this reason that funds have been early adopters of new technology - a systematic method cannot be implemented without high quality data and analytics, and at the heart of this is a database.
Complex Data Requirements
Accurate pricing requires all the relevant terms and conditions of each bond, plus their underlying stocks, company information (two if it's an exchangeable), plus supporting data such as yield curves, exchange rate data, dividends and so on. A comprehensive system will need not only to be able to capture and store all this data automatically each night, but also keep it updated regularly with dividends, stock splits and new issues. with the minimum of manual maintenance. Perhaps the most troublesome data to collect, especially on an historic basis, is implied volatilities, credit spreads, borrow costs and forecast dividends.
The determination of volatility is fundamental to delta hedging, and forms the basis for the calculation of credit spreads and risk management. It is of course a very subjective measure, but good historic information is vital. Ideally, one should scan the whole market and compare like with like - study information on similar instruments, if available. If the underlying equity enjoys a liquid singlestock options market then gather as much historic data on these as possible in order to assess: a) how their implied volatilities have shifted through time; and b) where they are trading now. A good system will be able to capture and maintain such inform ation directly and automatically from data vendors and produce historic implied volatility surfaces tailored to your particular needs and models.
The underlying stock price volatility is also measured, as is its evolution through time, to derive a picture of how it can change and how variable it can be (the so called 'vol of vol'). It is very important that all stock splits and dividends have been adjusted out when carrying out this measurement.
Good quality historic credit spread data is diff icult to find, which can cause problems, as bond prices are so very sensitive to them. More recently, as the bond asset swap and credit default swap markets have developed, standardised sources of pricing for liquid issuers have become more available. One such provider, Credit Market Analysis, now produces constant maturity interpolated 'fair value' spreads for bonds of all major issuers of straight bonds (see below).
 |
If no credit spread can be found, then many practitioners will 'back out' the value by implying the credit spread from the market price of the bond using a suitable pricing model. To ensure this value is accurate, all other parameters must have already been determined, and on an historic basis. This includes the bond's implied volatility as discussed above, as well as dividends and assumptions about borrow costs for the underlying equity.
Historic Data and Theoretical Pricing
Traders vary in their preferences for historic data, but more is preferable to less and most like to have at least three or four years for any issue being studied or traded, unless of course it is a recent or new issue. If so, then other problems may need to be overcome, concerning proxy data and hedging, which are beyond the scope of this article. The acid test of any such database is that the theoretical values of any instrument, on any date in history, agree closely with the market price on that day. Once achieved, analysis can proceed with confidence.
Some traders employ techniques similar to the technical analysis seen in equity trading. A popular strategy is to study the relationship between a price and parity through time, as shown here:
 |
The trader fits a line to the data that best represents the baseline value that the bond rarely trades under. On any given day, if the bond plots significantly below this, the trader buys the bond with the expectation that it will rise above it in the near future. Deriving the line, and deciding the levels at which to put on and take off the strategy, are empirical issues that require the trader's experience and large amounts of historic data.
Gamma trading is one of the most widely practiced and successful strategies currently and is re a lly a variant on the play above. A bond is purchased at an implied volatility that is hopefully below that at which it can be delta hedged by the underlying stock (normally estimated using historic volatility). If successful then, as time passes, more money is made on rehedging with the stock than is lost on the bond.
A good example of such a strategy can be found with the Siemens Nederland NV 1 per cent 2005 bond (see below). It was issued in August 2000 with a delta of around 60 per cent, but the stock has fallen dramatically since then from over EUR70 to under EUR20. Players who shorted sufficient stock to hedge it correctly at issue would have enjoyed an almost unbroken run as it traded down as shown in the example:
 |
What makes this bond so interesting is that it is exchangeable into stock of another company, Infineon Technologies AG, not into Siemens stock. The credit of Siemens has remained high (AA and trading at around 50 bps in the default swap market) throughout Infineon's fall, and hence the bond hit the bond floor but did not slip any further, so limiting any losses incurred on the bond. This has provided traders with an added bonus from the move.
A trader's due diligence for such strategies requires a great deal of data; he must make sure that stock borrow is readily attainable at a predictable cost with little threat of it being called, and the liquidity of the stock must be sufficient for the rehedge volumes to be absorbed into the market. The strategy must be immunised to movements in other parameters by way of credit default swaps, interest rate swaps and often positions in forward foreign exchange. The expected costs of these need to be factored into the pre-trade analysis and then the whole monitored as a package of trades, or strategy, for the purposes of risk management.
 |
Maturing Markets
A direct consequence of increased market activity is the narrowing gap between the implied volatility of the bond and the volatility of the stock. It is forcing traders not only to calculate their rehedge more carefully, but also to reconsider the trade in the first place.
Pressure on this gap is coming both f rom the primary market where the traditional discount on new issues (as a sweetener to ensure the issue is taken up) are steadily being eroded (for example the recent issue of the Orix 2022 Zero - Coupon bond), and from the secondary market where the increasing number of players is driving up the implied volatility of the bonds, and beginning to drive down the volatility of the stock. The latter effect is a consequence of the amount of rehedging being carried out relative to the average traded volume for the stock, resulting in a damping effect on stock price moves. For example there has been some interesting market discussion recently on the amount of issues by Vinci; oversupply shows signs of crushing the volatility of the stock and has not only suppressed demand for the recent issue, but also stymied the performance of their existing bonds.
Technical architecture for convertible bond risk |
One may argue that these effects are merely symptomatic of maturing markets, but the problem is compounded here through an asymmetry whereby there are no short gamma players. It is very difficult to short convertible bonds, so the only short players are the issuers themselves, and they are normally trading from a position of 100 per cent cover, and hence will not delta hedge. This means there are few players taking opposing positions to the hedge fund traders, which in turn has a large liquidity implication for the market. This will exacerbate until short selling is more widespread.
Going deeper
A consequence of this is that rehedging techniques developed in illiquid stock options markets are increasingly being used, including more analysis of the depth of the market; the gamma 'hot-spots' for stocks - those stock prices where bond issues will need to be rehedged the most - intraday volatility; and simulations using historic data, all intended to wring out the last few basis points. This in turn requires more and more market data and better financial transparency from all of the companies involved.
All such tasks are very data intensive and require good histories of prices, traded volumes, shares outstanding, free float, cross holdings and soon. Often sector analysis will also be performed to determine where the company and its securities are trading relative to their sector in terms of credit yield. These tasks can easily become very manual and time consuming if they are not carefully planned and executed. Consider a one-off analysis carried out by a trader in a spreadsheet before a trade is made. This analysis has to be automated and repeated on a nightly basis, almost in its entirety, by the risk manager in order to monitor the position, and ensure that the risk profile does not change significantly. Using spreadsheets alone to manage and automate this process given the quantity and complexity of data involved soon becomes unwieldy, open to human or operational error and is ultimately impractical.
Increasing accuracy
 |
In order to perform rigorous and robust risk management, rehedging and trade opportunity analysis, the trading and risk systems used by convertible bond practitioners must be able to move these parameters freely before repetitively repricing the portfolio (see image chart above for required systems architecture). A point not to be overlooked is that the system must be capable of using these parameters in the calculations in the first place, and as the markets grow and they become steadily more sophisticated, the number of factors involved rises, the number of instruments analysed grows, and the accuracy and complexity of the analysis increase too.
Given the continuing expansion and the acknowledged complexity of the convertible bond market, clients and software vendors alike must hope that the exponential growth in computer performance continues for a good while yet.
Hedge Fund Technology is the only provider of dedicated news, information and advice on technology for hedge funds. From the publishers of Hedge Funds Review, Hedge Fund Technology provides crucial information on which hedge fund management technologies are available to help funds run more effectively and stay ahead of the competition. For subscription information please call 020 7306 7122 or e-mail info@incisive.co.uk.