Saturday, August 19, 2017

More on algorithm aversion - it can be eliminated if given some choice

In spite of the strong growth of systematic trading, the problem of "algorithm aversion" is real and must be addressed. In many cases and in many disciplines, models do better than humans, but humans feel anxious with models and do not want to place decision-making in the hands of a machine. Even if a model does better at forecasting, individuals may still prefer the discretionary or non-alto approach if there is the perception that the model is imperfect. It seems as though individuals will like to have the optionality to choose an approach (model or discretion) as opposed to being locked into one choice. See our posts: "Algorithm aversion" and managed futures and Algorithm aversion or just a desire for low cost optionality

The same researchers who identified algorithm aversion have published new work which suggests simple ways to avoid the problem. (See Overcoming Algorithm Aversion: People Will Use Imperfect Algorithms If They Can (Even Slightly) Modify Them by Dietvorst, Simmons, and MasseyThe authors find that if some modest amount of control is given to the decision-maker, he will choose the algo over his own decision-making. Allow the individual to adjust the model output by 10% and they are happy. Allow the individual to reject 10% of the output from the model and they are happy. Aversion falls when the human has some control over the model. You could say that the way to combat one form of behavioral bias - algorithm aversion can be through another bias, the illusion of control. Fight the bias with a bias. Or, allow the decision-maker to have some optionality to override and aversion goes away.

I have seen this happen in the systematic model space. Investors seem to be always fearful that the manager who always follows the model will at some point drive it into a brick wall. If the decision-maker has, under extreme circumstances, the chance to turn-off the model, anxiety falls. The optionality does not have to be used. It just has to be available. This can be tested a through real life survey. If an investor is shown two systematic managers, one with no discretion and the other with some modest discretion, the question can be asked as to whether one will be preferred to the other. If there is a preference, an additional question can be asked as to the amount of excess return necessary to switch. Investors do face this type of problem.

Algorithm anxiety affects investment decisions, but research suggests that it can be overcome so that the best choices are made in spite of biases. This is especially important given the backlash against systematic managers that is starting to creep into news stories.

Thursday, August 17, 2017

Kuhn and systematic research - Solving one crisis at a time

Thomas Kuhn, the science historian, developed big ideas like the paradigm shift in science, but his ideas can also work on the "smaller" ideas of finance research. The Kuhn cycle, which has been applied to the evolution of science, is a durable model for how real world research is conducted. It is an effective way to look at one critical part of the systematic research process. Quantitative research for many firms is broken into two parts: 
1. A search for new models and strategies that are either uncorrelated with existing models or a new variation on an existing strategy theme, or 
2. Maintenance of existing models through improvement and enhancements of existing parameters and frameworks.

This is a simplification because research can be broken into asset classes, trading, risk management or a number of other topics; nevertheless, the two-part breakdown is an effective generalization. The Kuhn cycle can be employed as a framework for discussion on how research moves from maintenance to a shift to new models. Under normal conditions, a model will be said to work until there is a significant drift in performance. This drift can come in the form of a performance drawdown or forecast degradation. If there is a drift in performance from expected norms, a model may be thought of as in crisis or broken. Parameter adjustments may not be enough to save the model because it is misspecified. 
When parameter adjustments are not enough and the specifications for what could be driving performance are wrong, research should lead to a potential model revolution. If the new model proves to be better, the quant strategy will have a paradigm shift in decision-making. 

The evolution of model building can be consistent with the evolution of science in finance. On a very micro level, the search for risk factors and risk premium is following a Kuhn cycle one factor at a time. Tests are done, model performance on a go-forward basis drifts, there is a model failure, a new hypothesis is generated, and there is then a shift to a new factor. Academics pursue normal science while practitioners test under live market conditions with profit and loss measuring drift. Poor profits create a model crisis and a search for new alternatives. The speed of new models is dependent on the speed to a P/L crisis. This is an adaptive process.

Global Macro - Managed futures as an alternative to corporate bonds

Corporate bond risk is rising. Of course, with improvement in the overall economy and continued bond flows many will not believe it, but the statistical data suggest that spread moves are no longer symmetrical. There is more potential for spread widening versus continued tightening. 

For the post Financial Crisis period, BBB corporate spreads are now at some of the tightest levels on record and significantly tighter than average levels by over 100 bps. For high yield spreads the same story is also true with an even wider gap between current and average levels. 

Default levels have fallen since the upheaval in the oil market so there would need to be further economic and company improvement to push spreads through all-time lows. There might be further tightening if there is a reduction in supply from corporations deleveraging, better earnings, or reductions of equity buy-backs programs. We have already see cuts in buybacks. On the other hand, it is possible that corporations believe there are better long-term investment prospects and there is actually an increase in borrowing to build new plant and equipment. Nevertheless, overhanging these macro drivers is the threat of a liquidity event if there is a reversal in buying. 

The tight spread levels make choosing an alternative much easier. For example, the LQD ETF is currently at a spread of 150 bps over Treasuries with the same duration. This stand-alone hurdle rate for an alternative investment over Treasuries is significantly lower than a year ago. It is easier to make a switch and earn a comparable return.

So what can be a good alternative to corporate bonds? We think it needs three things:
  • A stand alone return of Treasury yields plus the corporate spreads.
  • An uncorrelated return to corporate spread widening.
  • A risk that is symmetric and not skewed to higher spread duration. 

Managed futures or global macro could be a good alternative. First, managed futures could be done as an overlay on Treasury bonds as collateral through a separate account. This overlay will allow the trade to generate the Treasury returns and the excess return hurdle will be the spread on corporates. In this case, the managed futures investment has to generate at least a 1.5% rate of return to match the yield spread on corporates. While many funds are currently in a drawdown, there are also many managers that have generated high single digit annualized returns over the last three years with a symmetric risk profile. These managed futures returns should be uncorrelated with corporate spreads. An additional benefit is that if there is a dislocation or divergent event that will drive corporate spreads wider, it is likely that this will be good for long/short managed futures managers who generally do better when there is a market dislocation.

Switching to managed futures from corporate bods achieves three goals: 1. a reduction in credit risk; 2. a switch into a strategy that may have more upside potential; and 3. an increase in the diversification within the overall portfolio.

Monday, August 14, 2017

What do CIO's care about? Only four things

"Four things CIOs care about: asset allocation, portfolio construction, manager selection, and risk management." Kip McDaniel is the Chief Content Officer and Editorial Director at Institutional Investor, from capital allocators podcast.

The comment from Kip McDaniel provides a roadmap for what any hedge fund needs to address when marketing to a pension or any client. It is not about you, the manager, but the investor. 
1. How does this investment fit within the asset allocation framework of the pension? Why does it matter?
2. How should this investment be delivered to the client? How does it fit within the overall portfolio construction and use capital efficiently?
3. What is your edge versus other managers and how can you generate confidence that this edge can be achieved?
4. What will be done by your fund to protect the money allocated to you? How will your investment help protect the overall portfolio? 

The questions are relatively simple, but the answers require a lot of thought if the manager wants to truly be a top service provider.