It seems, however, that seamless communication between PMs and traders is lacking here. The RMS and the OMS aren’t really linked in any way. The portfolio construction process is comprised of a series of inputs and assumptions that generate investment decisions. The execution process is comprised of a completely separate set of inputs and assumptions that support the implementation decision. Without using the same data, it is likely that trading strategies are not always an accurate extension of investment strategies. They may, in fact, be at odds.
Take the example of how PMs and traders might use the transactions costs data in a different way. While spreads, average daily volume, volatility and risk factors might be inputs to the portfolio construction models, they are a subset of a much larger series of data points that go into the choice of securities. Fundamentals and quantitative metrics such as historical price ranges, price and earnings ratios, growth and margin trends, cost and revenue comparisons, corporate announcement dates and pending corporate actions are just a small set of inputs that the PM may take into consideration. Transaction cost data can be integrated with these input variables to develop timing, portfolio weighting or even stock substitution decisions. The data would be utilized to evaluate the optimal risk vs. return or opportunity vs. cost ratio as well as the relative size and weighting of individual securities within the portfolio.
The trader, on the other hand, could take TCA data and apply it in a completely disassociated manner. Say a trader is using a benchmark such as arrival price or VWAP, and he receives two baskets of orders from two different PMs at the start of the trading day. The order sizes as a percentage of average daily volume and real-time pricing and volatility data might lead that trader to conclude that the same equity algorithm would be equally effective for a list of buy orders in the small cap growth portfolio and another basket for the small-cap value portfolio. Using historical trade data and share sizes, the two portfolios of securities may appear to have a similar profile, and, in a quiet market, a participation strategy might appear to be the appropriate algorithm for both.
Advancements in portfolio trading algorithms support the ability to maintain the integrity of the overall basket during the transaction. There is functionality to protect individual stock executions from proceeding to quickly or too slowly relative to the rest of the securities in the basket. Portfolio-level algos can prevent individual orders from being completed while others are only partially executed, for example. Or they can alert to the fact that executions in stocks in a particular sector or industry are executing faster than others. These capabilities allow the buy-side trader to avoid inadvertent tilt or risk when prices and liquidity behave in an uneven fashion. But they don’t necessarily produce the optimal trading strategy to align perfectly with the underlying stock selection process.
Clearly the growth portfolio and the value portfolio were constructed based upon different expectations of stock price behavior. They are comprised of securities with very different alpha characteristics, covariance and risk factors. If they are traded through the same algorithm, some securities may very well be executed sub-optimally relative to portfolio construction expectations. Value stock executions may be crossing the spread too frequently while growth stocks with momentum may not be traded aggressively enough. The result is a significantly different market impact and opportunity cost for both baskets than would have been expected given the TCA data and alpha expectations built into the portfolio construction models. The right choice of algorithms with the right set of order parameters would take into consideration differences in the expected behavior of securities beyond similar percentages of trading volume and execution results vs. the benchmark. And would do it in real-time.
Even within each of these portfolios, every individual small-cap stock has its unique characteristics with reference to liquidity, expected alpha, volatility and risk. And when combined together into an investment portfolio, their individual stock characteristics have blended to achieve a unique investment profile. If a trader then takes this basket of blended behaviors and executes them using one algorithm, one set of order parameters and one execution time frame, it can be expected that execution results would not optimize the investment decision. The result would likely be differential alpha preservation and loss within that basket trade due to a lack of differentiated execution. While these differences were an integral component of the investment decision process, they are not now an integral part of the implementation strategy.
Next generation tools will allow for a higher degree of integration and coordination between the investment mandate and the trading mandate. If the PM has utilized particular trade cost data in his portfolio investment process, those assumptions become part of the analytic used by the trader to put investment analytics and trading analytics in sync with one another. Best execution, in this instance, will be the ability to establish trade parameters that coincide precisely with the assumptions used in portfolio construction as both the investment model and the trading strategy are optimized in real-time. Today there is no effective means for the PM to communicate his trade cost input variables or his risk / return ratios to the trader, and there is no practical means for the trader to apply them in his execution strategy.
The role of the buy-side trader is to implement his PM’s investment decisions in an optimal fashion so as to minimize transactions costs and preserve alpha. The best way to do that would be to have tools at his disposal that would extend those parameters from the investment decision right into the trading room. Bringing the PM and the trader closer in concert would engender execution strategies that are truly Best Ex. No more bubble gum and scotch tape.
Execution Consulting (
Comments | Post a Comment
9 Comments to "Putting the Pieces Together: Portfolio Managers and Traders Unite!":
GCCWolff
21 July 2011
Great point. This role might be filled by strat teams if the integration is really 360 degrees?
lberke
22 July 2011
Yes, agreed, if the firm has the resources, which most buy-side managers likely don't at this point in the cycle. Only the very largest buy-side desks have a strategy group, while the majority depend on their desktop trading technology, their TCA, and their sell-side coverage. None of which is really tied into the investment decision making process. It won't be long before the technology will evolve to the point where the data associated with the porfolio construction process will be seamlessly integrated with the data that is utilized in the transaction implementation process. This will undoubtedly enhance the trader's ability to minimize costs and preserve alpha in a much more precise manner than he can today.
Comments (129)
tonygau
22 July 2011
Technology always evolves. It enhances cooperation among different people, like social network. As new platforms are emerging to facilitate the link between portfolio managers and traders, now it is up to organizations to take a top-down approach and search for the treasury (e.g., missed alpha) hidden inside. Before portfolio managers dedicatedly hunt for signals, and traders speedily chase after liquidities, executives should give them higher initiatives to work together.
Comments (4)
cameronhight
26 July 2011
Interesting article. This is written with a more quantitative shop in mind, but we have built a solution to bridge the gap between the portfolio manager and the trader (ie. between the RMS and OMS). We capture the analysts's expectations for Risk/Reward and use them to construct a model portfolio which we compare to the existing portfolio to point out the largest discrepancies. www.AlphaTheory.com.
Comments (2)
lberke
27 July 2011
To your comment, Cameron, I would also include fundamental shops in this equation. I think that with the research-driven fundamental investment process, the challenge is even greater to efficiently translate investment decision inputs into trading decision inputs. I'm curious. Does your model portfolio include any assumptions about transactions costs or liquidity factors?
Comments (129)
cameronhight
27 July 2011
I couldn't agree more. The method for translating research into a portfolio is almost purely heuristics for most fundamental shops. This causes huge inefficiency in position sizing. Our model really isn't a model at all. We just give the portfolio manager, trader, and analysts a way a tool kit they can use to take their research, analyze it, and point out where their position sizing doesn't match their research. The question I always as a fundamental manager to explain what Alpha Theory does is to ask, "What is your 6th best idea?" The answer is almost always, "I don't know." If that is the case, you can rest assured that their 6th best idea isn't their 6th largest position.
Comments (2)
lberke
28 July 2011
tonygau, I'm interested in your comment about social networks. Do you see technology like Twitter or IM as the foundation for better integration of information between the PM and the Trader? I was envisioning more along the lines of a realt-ime data exchange that would deliver order and portfolio characteristics derived from the investment decision process to the trading desk along with the orders or portfolio trades themselves. And then there would be a mechanism whereby a trader could communicate back to the PM that the estimated pre-trade transactions costs or liquidity profiles can be expected to have a certain potential impact on net alpha.
Comments (129)
tonygau
28 July 2011
Iberke, I will get back to your question in the next message. But, here is some thought on the alpha and liquidity. Investors have learned a lot from 2007 liquidity crisis and then 2010 flash crash. Your best hands are not unique. All of sudden, people realized everybody is on the same side and running to the same exit. The liquidity matters. One question people have to ask is if I should bet on an idea that lacks liquidity. Could I be able to get into the position but can’t get out? Does the liquidity cost eat up my return? More often than not, big alpha lies in tough trades.
Comments (4)
tonygau
28 July 2011
lberke, social network has brought in an revolution on how people communicate and work together. The fundamental change is a single platform to exchange and store information. Traders have long been using IM (e.g., Bloomberg) to exchange information with market participants and PM. However, current practice involves sequential steps and the PM and trader may need to communicate back-and-forth several times until the expected net alpha and the estimated market impact cost are well satisfied. It is important for PM and trader to work on a single platform and contribute/upload their own information/data, for instance, alpha/risk model from PM and pre-trade TCA from traders. PM and traders can then work as one team to construct the final portfolio that meets risk/return profile, while concurrently building a trade alongside trading strategy to better catch the alpha with a minimum transaction cost. It will then be a true investment optimization. www.tgoptima.com.
Comments (4)