EVO, The Evolutionary Market Trading System
What is EVO?
Based on quantitative analysis, EVO is the underlying algorithm that generates buy, sell, and short signals, which powers the
EVO 1 , EVO 2, and EVO 3 strategies.
Conceptually, EVO’s investment process relies on three main ideas:
Improvement through Constant Evolution,
Clarity through Aggregation, and
Risk Management through reliance on statistics and probabilities.
Always Evolving.
“Evolutionary: a gradual process in which something changes into a different, and usually more complex and different form.”
We chose the name EVO for our strategy to reflect the evolutionary process that the system development has undergone over the years resulting in its enhanced performance.
Clarity through Aggregation.
Starting in 2002, Rich Paul analyzed over 60 independent trading systems to identify the eight best performing systems to be used in the composite. These trading systems, when combined together in various ways, created the original EVO signals. Today, we’ve analyzed over 120 systems, and EVO has grown to over 30 trading systems, leading us to capture many more opportunities (Time invested increased from 39.9% to 55%, as of 12.31.2020).
Each of the systems is back tested separately and in various combinations with each other to optimize the risk and return performance. Systems and their combinations that give the best returns are not necessarily included in the final composite system. Those chosen have to meet strict requirements for risk and reward and have to provide a significant number of trades in the sample. The system combinations selected are further aggregated together to generate composite system buy, sell or shorting signals when a number of unique combination of systems point one way, i.e., when the signal is strong. This builds on famed hedge fund manager and author of “Beat the Dealer,” Edward Thorp’s finding in his Journal of Portfolio Management Article “when many trading systems are aggregated and combined together, the noise and error trends to diversify away - while the signal remains stronger.”
Predictions based on Probabilities.
Each EVO signal accepted into the composite drills down to a probability of success. We look at things like % winning trades, total gain, maximum drawdown, the ulcer index, etc., so that we know the exact risk associated with a signal. In our 17 years actual performance, 62% of our trades are winners (EVO 1).
Creating the Algorithm
EVO’s 30+ trading systems use a myriad of quantitative indicators based on historic financial and raw stock market technical data including major and secondary market indexes, breadth, volume and volatility measures, interest rate sensitive data, sentiment, and seasonality statistics. Each system has its own unique formula, sometimes using the same data to create a variety of trend following systems, which include price, market breadth and volatility indicators, and moving average crossover systems. Countertrend indicators identifying significant overbought/oversold levels are also used.
The rigorous process was designed around the concept of “filter” systems and “trigger” systems. Filter systems are generally intermediate to long-term systems that may be standalone market timing systems if used on their own. These systems function to screen trades generated by the trigger systems. Trigger systems are designed to capture shorter- term market inefficiencies or repetitive patterns that translate into very high rates of return while invested. All the individual systems in EVO have exhibited superior risk adjusted returns than the market in backtests measured from April 1987. The identification of these systems is the backbone of risk management and the excess return generated by EVO in bear markets as well as in volatile periods in bull markets.
To reduce and/or eliminate the bias effects of over-optimization during back testing, measurement of consistency of performance over rolling time periods are employed in addition to out-of-sample testing and forward performance testing, which are used to confirm the results during the in-sample data period. Sensitivity analysis is also performed to mitigate the bias by varying the parameters of potential timing systems incrementally and eliminating those parameter choices that do not lead to a smooth performance profile. That is - eliminate spike trades that distort the result.
Potomac Advisors is always looking for new trading systems to analyze which may improve the performance of the algorithm. Since 30+ trading systems create so many combinations - over 150,000 - new combinations are reviewed in real time, daily, for inclusion, allowing our signals to stay relevant in all market environments.
Consideration of Risk
While buy and sell decisions generated by EVO may have been successful in the past, the algorithm is constantly evolving and may be ineffective when applied to future market environments. EVO is subject to unique and varying risks in addition to the traditional market risks of equity investing. These risks, described below, are attributed to the mechanical nature of the EVO strategy.
1. Patterns of market data and technical indicators that are used in the formulas that have correlated with stock market direction, and were used to identify stock market buy and sell timing decisions in the past, may temporarily or permanently not correlate with the stock market direction in the future. This could lead to losses when using these formulas. To mitigate this risk, the EVO strategy is constantly evaluated to determine whether certain formulas should be changed or omitted when such formulas are no longer providing value.
2. A mechanical trading system may generate a series of consecutive losing signals resulting in the compounding of losses more than the stock market over the same time period.
3. Material market conditions such as unpredictable sudden financial, economic, or political news events that affect the market and reverse the direction of the market with respect to the direction forecast by the mechanical formula may have a short term negative impact leading to losses.
To limit the impact of such losses, EVO 1 & 3 (the mechanical models) currently employ a 7 ½% stop loss on each trade, measured from the closing value of the S&P 500 stock index from the date of entry. Once this threshold is met or surpassed, we exit the fund on the next trading day, unless a new EVO composite system buy signal occurs on the same day as the stop. Once stopped out, a new EVO buy signal is required for reentry. However, there is no assurance that the net trade loss would be 7 ½% or less. if stopped out. While stop loss trades have been infrequent in the past, there is no assurance that the frequency of such trades will not increase in the future.
These strategies may be considered aggressive because of the use of leverage and/or the elevated frequency of trading. However, EVO 1 and EVO 2 have actually demonstrated in over 12 years of actual performance that their level of risk has been significantly lower overall than that of the S&P 500 in terms of maximum drawdown, with slightly lower volatility as measured by the monthly standard deviation. The lower level of volatility is a result of having avoided severe market declines and spending half the time in cash. However, there is no guarantee that this risk performance will continue, and therefore, investors need to be aware that the significant leverage employed and the possibility of whipsaw trades resulting from a sequence of consecutive losing trades could result in accentuated losses greater than the S&P 500 or NASDAQ 100 indexes.
Overall, investing in securities involves a risk of loss that you should be prepared to bear.
