The best way to describe the Flyer Forex Fund’s birth into the currency world is with a horse analogy. At the beginning of the fall 2009 semester, there was much student interest in building the fund, but no knowledge base to leverage. The professor was stuck with simultaneously teaching the fundamental and technical trading of the currency market and building the fund. The horse ran wild in many directions and required a lot of effort to corral and tame. Then we found that the cowboys liked certain tasks (mostly the trading) but lacked the discipline to perform consistently the middle and back office functions. Again, the lack of knowledge about middle and back office functions came into play, and the professor found that assuming a knowledge base often meant a valuable waste of time. Christmas break came and the horse stopped running and just stood around in the field.
The spring semester began; the student body managing the fund totally changed, with only three students carrying over from the previous semester. Of the three carry-over students, each student tried to actively lead a portion of the fund. However, now it seemed the horse had trouble being led forward and there was significant resistance on the harness. While much improvement had been made to the back and middle office functions at the end of the previous semester, it seemed in January we started from scratch again. By February, after resolving many operational issues (both old and new), our horse was off his feet and plopped down in the pasture. We looked around and knew we needed a defibrillator to revive it.
The student management of the model trade produced a loss of $4,280 vs. a gain of $7,980, if the model was run purely over the time horizon. “Not how it was planned,” right, illustrates the difference between the pure model and our students’ execution of that model; in other words, it is the cost of the learning curve of the students running an FX algorithmic model. The object of the simulation of the fall semester was to get to trading real money in the model during the spring semester. That did not happen. Of further note, this fund was purely a student self-motivated experience and no grade was given. It was not for any academic credit and its purpose was solely to increase the students’ practical knowledge base.
The desired method
The students opened a demo account at FXCM of $50,000. They were to trade one lot ($100,000) as directed by the model based on the four-hour chart (see “Learning to Trade,” October 2009). The only currency the fund traded was the euro. We began trading on Sept. 6, 2009 and finished for the purposes of this analysis on March 4, 2010. All documents were kept on a Google Group site. We also ran a weekly update on a blog spot http://www.flyerforex.blogspot.com. Each team of students, regardless of semester, rotated though six trading teams and one management team weekly. As a result, every student had experience in the front, middle and back office of the fund. To discuss this experiential learning event, we break down our analysis of the cost of the learning curve into front, middle and back office issues and management buckets, which occurred over both semesters unless indicated.
(Click on image above for full-size chart.)
Front office issues
Operationally, the “trader” had the following duties to the Flyer Forex Fund: read the market data at the correct four-hour candlestick, immediately update the data within the model as the close of that candlestick, run Girth calculations to determine if a market action was needed, and execute the trade in a timely manner in the correct pre-determined lot size as necessary. If any error occurred in this process, the trader was required to update the error report and estimate the dollar cost of the error. If there was any market action, the trader was to print out the FXCM profit/loss report from the system and at 3 p.m. each day the trader ran the same report regardless of market action.
The fund faced issues across the entire trader process. Traders were consistently unable to update the model, execute in a timely fashion, report errors, or find trader replacements if they were in class at the execution time and could not reach a computer. Trading teams could not communicate consistently, despite the fact that they were in close geographical proximity. All data was located on a Google Group site, but many traders lacked the discipline to monitor the site regularly. Operational issues entailed entering a trade with the incorrect lot size (for example, $90,000 instead of the pre-determined $100,000), entering the incorrect data in the model either from the wrong candlestick (not the close of the four-hour candlestick) or even the wrong time period candlestick, using the wrong version of Excel (version 3.0 instead of 7.0), which caused model calculation issues and prompted incorrect market action, incorrectly managing the model, which in one case drove the trader to an erroneous early close of a position and thus a re-entry of the desired trade, and finally, leaving the trading platform open with a live trade, which provoked a passing student to close some of the fund’s trades and execute positions in the pound, the yen and the yuan. Amazingly, those trades were profitable!
The biggest drawback of running the fund with new students each semester was the lack of continuity. While the fall students stated that they would oversee and train the new spring students, only three of the fall students fulfilled their mission. However, the majority of the fall students did not participate in the spring semester.
Back office issues
Traders were notorious for not managing the profit and loss of the model consistently and correctly. In fact, to produce the above graph, “Missed opportunity,” it was necessary to hand check the results of the student P/L output stored in the spreadsheet against the paper copy printed from the report. While all students loved the trading aspect of the simulation, few could barely tolerate what they termed ‘grunt work:’ maintaining the historical P/L records, producing daily and monthly P/L reports correctly, consistently and timely without prompting from the professor or teammates. This problem did not improve over the life of the fund.
Middle office issues
It was the intent of the professor to include the use of the value at risk (VaR) model in the Girth model. Several students were given the task of producing the VaR calculation for the trading model. These students were not as quantitatively adept as was expected, and the professor had to produce the model and instruct the students on the concept and use of the VaR model. It was a frustrating experience educationally, although it ultimately resulted in the full understanding of VaR within the trading model. However, no attempt was made at VaR implementation as full understanding of the VaR model occurred during the end of the fall semester. Again, it was a lapse of continuity. Spring students never got to the VaR understanding.
Management
One of the biggest drawbacks to the management team was the initial lack of knowledge: what was the euro, what drove it, how the model worked, how to produce reports, how to actually report P/L, how to write market descriptions and how to drive a team of traders. The management team eventually learned that the buck stopped with them and drove themselves to produce results, some more timely than others. The fall semester team was much more effective in their methods than the spring semester team. The spring semester team actually was driven by three students from the fall team and they became the de facto management team. The hindrance of the spring team came in the initial knowledge transfer from the fall team, which was rocky. The professor spent much more time with the fall team than the spring team; she relied on the fall team to transfer the management knowledge base to the spring team and that transfer fell to three students. Further, the professor wanted to take a hands-off approach to the building of the fund; it was imperative that the students learned through experience. This approach probably produced more operational errors than necessary, but it broadened the knowledge base.
Surprisingly, the management team reports to the Futures magazine Web site for weekly updates was very sporadic. The students as a whole seemed to be unable to recognize the update as a valuable marketing tool. The students had similar issues with their blog; they simply could not seem to update it on the same day every week, and often missed a week.
Perhaps the biggest debacle of the total management of the Flyer Forex Fund was not trading the month of December against the recommendation of the professor (see “Missed opportunity,” above). The professor illustrated the high volatility trends of the euro of the past several years and indicated that, while the euro was range bound during November, historically it had significant December movement. The loudest cry for closing the fund over the holiday break came from a student who had already obtained a full time trading position after graduation.
The defibrillator
The biggest problem with moving toward real money funding in the Flyer Forex fund is a firm knowledge base and a truly committed core team. Therefore, the spring team will be producing a set of policies and procedures for the fall 2010 team. Only three students will roll into the fall team. The core spring team will then interview for students to form a less than 20-member team to manage the fund throughout the next academic year. While there will be a continued problem of teaching commodity trading, while starting the fund in the fall, it is hoped that the new team of students will practice on the FXCM trading platform over the summer and read ahead for the fall 2010 class. Thus, current students will interview for future students, knowledge will be passed, and a committed team should have increased accountability. The Girth model itself has been rewritten by a current student to simplify the model management issues, the error reporting issues and the introduction of VaR.
Like most endeavors, success in trading is often contingent on executing consistently the more mundane aspects of that endeavor not necessarily the sexier aspects. These are lessons every trader/manager must learn, preferably before placing his own or other people’s money on the line.
While it is true that the student driven model did not perform as well as the pure Girth model (real trading never does), the exercise was extremely valuable to discover operational and human fragilities in the system. Regarding professorial input, it is an exhausting process when coupled with other academic and service duties. However, the cost of the learning curve has been mostly paid for and it is worth the effort to move forward and take another stab at attempting a smooth transition toward real money management in this fund.
Leslie K. McNew is a clinical professor of finance and visiting director of the Hanley Group Derivatives Trading Center, Department of Economics and Finance at the University of Dayton Business School. E-mail her at lesmcnew@gmail.com.
Charles B. Keller graduated with a bachelor’s degree in finance from the University of Dayton Business School in May 2010. E-mail him at charles.b.keller@gmail.com
Kevin L. Woeste has an MBA concentration in finance and will graduate with a master’s degree in
financial mathematics from the University of Dayton Business School in May 2011. E-mail him at woeste.kevin@gmail.com.
John Rapp, Ph.D., is the chair and professor of economics at the Department of Economics and Finance at the University of Dayton Business School. E-mail him at rapp@udayton.edu
