Despite disclaimers warning against it, past performance is exactly how most investors, whether institutional or individual, determine who to invest with. There are obviously other factors; institutional investors want to make sure a trading advisor has a solid business structure and hopefully both professional and retail investors look at the risk adjusted return of a manager. But those are ways to qualify the past returns.
In general an investor, either institutional or retail, will be attracted by positive returns. They then will look at other qualitative measures: is it a solid business structure, what risk is taken to achieve those returns, are those returns repeatable, will the manager run into capacity issues, does he have a disaster recovery program, and how does the strategy correlate with other investments?
They perform due diligence to confirm the returns were not a fluke, the program is not overleveraged, the manager has no outstanding bunko warrants and the business has a solid structure.
That is quite a bit different than looking at a couple of hundred solid commodity trading advisors (CTA) and determining which 10 to 20 are due for an exceptional year and which ones will underperform. That is what Van Kar Trading claims to do with its VK-Alpha CTA Selection model.
Van Kar has developed a proprietary system to select when to allocate to CTAs. They are allocating client funds based on the method and are creating a fund-of-funds base on the model. The system first qualifies a large group of managers, then separates them in sub-groups and analyzes their performance cycles and outside market conditions that can predict those cycles.
Emil van Essen, president of Van Kar, began his research into CTA performance in the early 1990s when he was director of managed futures for the Bank of Montreal. “It seemed to me that what everyone was doing; Sharpe ratios, correlations, all the typical ways people pick a group of CTAs, didn’t seem to work. Nobody ever tested it to see if this way to pick CTAs had any validity,” he says.
That led van Essen to ask, “If the way allocators select CTAs doesn’t work, what does? It certainly isn’t random.” So he started looking for factors that would predict when a CTA would do well.
He looked at markets traded, market conditions and volatility and began creating a software model that would take all the managers in a database and create a fingerprint for each manager.
“Then it would go back in history and look for people that matched that fingerprint. Not just their performance characteristics but also the market conditions, and once it finds a match in history, it [shows] how these matches do going forward,” van Essen says.
Institutional allocators typically require a three-year track record and $10 million under management before making an allocation. Many allocators will wait for five years of performance and $50 million under management. They then would employ due diligence, analyzing a manager through a series of quantitative and qualitative measures.
Ernest Jaffarian, managing partner of Efficient Capital Management, says, “due diligence cannot qualify a manager, it can only disqualify him.”
Efficient Capital offers a multimanager fund and uses many standard measures for identifying quality managers and some unique ones. “Our first cut is always based on mathematical analysis based on daily return data. A trader needs to meet specific mathematical criteria based on return and the character of the return profile,” Jaffarian says.
“What makes us unique is that we start on a mathematical basis and we work with only daily return history. What makes that unique is that we are working with higher frequency data and that we do that mathematical profile in a vacuum. We don’t start by doing due diligence, we start by doing mathematical analysis,” he adds.
The higher frequency data allows Efficient Capital to analyze managers in more market conditions and see how they handle stressful markets.
Growth in the popularity of alternative investment strategies has turned manager selection from a buyer’s market to a seller’s market. RJO Futures is raising money for emerging managers who have lower minimum investments and shorter histories. “I pretty much look at the retail CTA. I need a year of data; most people want two years,” says Rick Gallwas, president at RJO Futures.
“Ironically, I want to see a drawdown in that year rather than just good results because that makes people more nervous. They want to see how they handle a drawdown,” Gallwas says.
Jaffarian also acknowledges that popularity in managed futures has created more competition for allocators to find a manager earlier in the game. He says Efficient’s selection criteria allows them to make a more qualified analysis in a shorter period of time, but adds, “For this trading style there is no substitute for experience, so we continue to place a high value on experience.”
Both RJO Futures and Efficient Capital have perfected manager selection models. “You’re always looking to find the diamond in the rough,” Gallwas says. “I have taken managers with less of a track record than a similar manager trading the same products because their philosophy in the long run is going to perform better in the trading environment that we are in rather than the trading environment we were in four years ago.”
While Gallwas is looking for managers who will perform in the future, his model is not trying to predict it.
What is truly unique about Van Kar is that it is doing more than just qualifying quality managers. That is the first step. It then looks more closely at the managers’ performance to gain an understanding of which market environments it does well in.
Van Kar requires managers to have a two-year track record and their overall performance to be positive. Once they eliminate managers without a long enough track record and those with negative performance, they are left with around 300 managers to analyze when the system kicks in. It separates those programs into 13 groups.
“It will find out what sort of market conditions is best suited for the next year for individual subgroups and then it will look at the individual CTAs within those subgroup,” van Essen says. The system also is trying to get a read on the markets in terms of what markets are good for certain kinds of CTAs. It runs cross-correlations among mangers to place them in certain buckets and tests for style drift. Using all these factors, every CTA every month will get a percentile rank from 0 to 100. The average hold time for a manager, once selected, is 18 months.
“The model will look at the markets they’re trading, the subgroup they are in, the correlations, their performance, then give it a ranking. It uses all the historical data to decide whether the CTA will do well in the next 12 months. After it produces all the rankings then we will take the top ranked ones and we will start screening out ones that have some issues,” van Essen says.
The final ranking is a combination of both the cycles of the subgroups managers fall into and individual managers. Van Essen says if a CTA looks like it is going to be in an up cycle but the market conditions for its particular group are pointing down, the model will neutralize the two. “If the market conditions say it is going to be an up and the cycles in its performance are indicating it is going to be an up, then it is a double bonus and it is going to be a high ranked CTA.”
Van Essen says the model did a good job in predicting the recent poor performance period for long-term trend followers but notes that period could be ending. “For us to see good performance in long-term trend followers, one of the indicators is volatility in the S&P 500. When volatility comes into the S&P, it seems to reverberate through to the long-term trend followers.”
The model learns how certain CTAs perform following certain market conditions and rates managers based on the current market conditions and on how they performed following similar periods. “What we are doing is looking at a number of factors and we create market condition zones. We know that certain zones are very good for certain CTAs and these zones would include things to do with interest rates, with stock indexes, with the movement of commodity prices and the way that they are all interacting together,” van Essen says. While van Essen notes volatility is a factor, he says the model is not trying to predict volatility. “It is trying to isolate conditions that will lead to profits in certain CTAs in the future,” he says, adding, “Right now what is going to effect the rankings is this bull market in the commodity indexes, and if the stock market starts rolling over and goes kaboom.”
They don’t want to create too many buckets because that could lead to curve fitting. “We create three key markets conditions and we create three zones for each market condition. We have found three things that have predictive value, so that gives us 27 potential zones that the market can be in.”
The important thing to remember with the model is that current performance doesn’t have much to do with selection; it is not trying to pick the hot manager. It is categorizing types of managers and types of market conditions, and based on its analysis predicts which types of managers perform well in periods following these market conditions. So a manager’s recent performance has little to do with his ranking. A hot manager might end up on the bottom of the list, van Essen says. “He has had his upswing in his performance, he has had his upswing in market conditions and he is ready to implode.”
The model has proven it can add value throughout a large group of qualified CTAs. And by moving in and out of allocations it has the potential to profit from allocating to more CTAs. Most managers will perform well throughout certain periods but many fewer of them will consistently perform well. If the model can capture the hot streaks of middle of the road CTAs and discard them before they are due for a fall, it will have a much larger universe to choose from. Jaffarian acknowledges this potential but has doubts. “If you can do that it would be phenomenally valuable. I have done a lot of research in that area and I have serious questions about the [viability] of that.”
Van Essen compares what he does to the difference between hedge funds and mutual funds. Mutual funds would invest in the recognized safe equities to hit certain marks where hedge funds attempt to add value. “That is what our current crop of managed futures funds are like. They try and do things nice and safe but they are really not adding value, in fact they are reducing value because of the double fee structure. That is what we want to do on the managed futures side. Something that is not being done now.”
Van Essen says they are not picking CTAs but managing a portfolio.
“We are trying to turn it into more of a science than a guessing game.”