Is it better to adopt a more advanced approach or stick to a simple idea when developing a quantitative trading strategy?
An age-old question in the quantification community is whether system traders should stick to using simple quantification strategies, or should they strive to implement more advanced methods.
It is commonly thought that retail algorithmic traders use only simpler strategies, whereas quantitative hedge funds use highly sophisticated and mathematically complex methods. However, this has recently changed.
Retail algorithmic traders are now able to perform complex analytics with relatively cheap cloud computing, alternative data providers that offer affordable and easy-to-use datasets, and an open source research framework.
In this article, we will discuss whether retail quantitative analysts should invest time in executing these advanced strategies, or stick to a simpler idea.
Investor preferences
Before listing a series of advantages and disadvantages between simple and complex strategies, it is necessary to outline how we will judge the relative advantages and disadvantages of each approach.
One of the key issues is that each investor has their own specific preferences, so they have a set of nested target functions nested for the goals they are trying to achieve through system trading.
For example, an investor may have a large capital base, but may need to regularly extract any trading income from that capital. It is important for such an investor to save money to minimize losses.
Another investor may have a relatively small capital base and is only interested in increasing total wealth. The volatility of the overall net profit/loss curve may be less of a concern if greater returns are obtained.
Some quantitative traders place more emphasis on the intellectual stimulus of developing effective systemic trading strategies. They may actually get positive returns as a nice side effect of the scam they love.
Clearly, investors have many different preferences. These aspects help to build a discussion framework for retail quantitative analysts who may be deciding whether to adopt a more advanced approach to simple versus complex systemic trading strategies.
Simple strategies are easier to research and deploy to market. They require less complex data and infrastructure. Even if the signals themselves are automatically generated, some can even be executed manually.
On the other hand, advanced strategies are more intellectually rewarding, and they tend to have more favorable Sharpe ratios. That is, they provide better expected returns per unit of volatility. Sharpe ratios will be an important indicator to consider for investors concerned with how to minimize losses and volatility.
In this article, we will explore in detail whether simple yoga is better than complicated yoga. We will keep in mind the above motivations as well as other advantages and disadvantages.
Simple trading strategies
Whether a trading strategy is considered to be a straightforward one depends largely on the investor's educational background and technical ability. A person with a PhD in Random Calculation may have a very different definition of a straightforward one than a self-taught retail quantitative analyst.
For the purposes of this article, we will roughly define a trading strategy as one that uses simple tools with basic mathematical or statistical complexity in a large known asset class, if applied to developed markets.
Examples of such strategies include technical analysis of indicator bubble signals, no clear portfolio structure or risk management components, and high liquidity markets such as US stocks, ETFs or foreign exchange.
The advantages of a simpler strategy include:
- The data- All systemic trading strategies require data. Simple strategies typically use ready-made price/volume data from tools that are well-tradable in mature asset classes. The cost of obtaining such data is very low, even free.
- Research- There are a large number of retesting environments for testing strategies in the hashing style, from commercial products (such as TradeStation or MetaTrader 5) to open source libraries (such as QSTrader, Backtrader and Zipline), and even libraries such as Pandas.
- Cost of transactions- It is relatively easy to estimate transaction costs because simple tools are used in developed, highly liquid markets. This in turn makes it simpler to determine whether a strategy is likely to be profitable in the sample.
- Infrastructure- Low-frequency execution of technical analysis-type strategies can be automated through a relatively simple infrastructure. Depending on the required level of robustness, cron jobs can be set to generate the desired list of transactions, while also being executed manually.
- Capacity- Likewise, the use of simple tools in highly liquid markets makes it less likely that capacity limitation issues will arise.
However, there are disadvantages to using a simpler strategy:
- Alpha- Technical analysis of leverage indicator leverage strategies are very well known and common in financial markets. It is not yet clear whether the simplest strategy is more valuable than the basic buy and hold or the dynamic-based tactical asset allocation. That is, the strategy may not itself generate leverage alpha leverage, but rather leverage beta leverage from the market itself or other known academic risk factors.
- Profitability- Because of the universality of this approach, it can be challenging to sustain profitability in the off-sample once the realistic transaction costs are taken into account. This is why it is essential to estimate transaction costs as effectively as possible in any retrospective review.
- Statistical tests- While not a problem of simple trading strategies, there is often little or no robust statistical analysis of simple strategies. Therefore, many such strategies that show high performance in retrospective testing may simply be due to over-fitting of data in the sample.
- The Right to Free Will- Simple strategies executed manually may lead to discretionary elements being applied to the process. For example, delaying entry of a trade due to a busy opening time or using an intuitive lever to overturn a trade. This makes it challenging to determine the true performance of the strategy.
- Portfolio building- Simple strategies usually avoid the use of any robust portfolio building or risk management techniques. Although hedging is often used, it is rare to see volatility targets, equity volatility weightings (also known as hedge risk parity hedges) or cross-market diversification as potential mechanisms for improving risk-adjusted returns.
- The return of intelligence- Simple strategies usually do not use any complex math or advanced analysis. If the investor's goal is an intellectual return, then a simple strategy is unlikely to achieve this goal.
It can be seen that while simpler trading strategies are easier to implement, test and trade, this simplicity may come at the expense of statistical robustness and long-term profitability.
High-level trading strategies
Advanced strategies include those based on statistical hypothesis testing, broad asset class domain knowledge, rigorous portfolio construction methods, and strategies for less liquid, niche asset classes or instruments, such as emerging markets, commodities, and derivatives.
These strategies are typically the domain of institutional quantitative hedge funds, but due to the availability of data and the popularity of better simulation tools, these strategies are now becoming increasingly common in retail quantitative trading.
The advantages of complex strategies include:
- Relevance- By design, advanced strategies tend to be less relevant by design to the overall market and to any existing portfolio composed of other trading strategies. This often results in a higher Sharpe ratio of the overall portfolio.
- Profitability- With advanced field knowledge, transaction costs can be reasonably estimated. This means that it is usually easier to determine whether a strategy is likely to be profitable outside the sample. Therefore, many unprofitable retesting ideas can be rejected before the real-time testing period.
- Statistical tests- Strict trading strategy statistical analysis is usually accompanied by more advanced methods. This means that statistical analysis is usually accompanied by more advanced methods. This means that deployed strategies perform less poorly outside the sample compared to simple strategies that may over-fit within the sample.
- Alpha- The potential for alpha spikes in such strategies is greater because of the use of niche tools in underdeveloped markets. This alpha tends to decline more slowly because of the slow spread of strategic knowledge across markets.
- Portfolio building- Portfolio building and risk management are complemented by more advanced approaches; this helps to align investor objectives with strategic performance.
- The return of intelligence- Advanced strategies require more sophisticated analysis, more mature mathematical knowledge and more extensive software development. For some amateur investors, this is more like a goal than creating wealth. They are therefore often attracted to more sophisticated systematic trading methods.
As with simple strategies, advanced strategies have some disadvantages:
- Mathematical complexity- Some of the more advanced systems trading methods usually require a background in statistical analysis, time series analysis, randomization or machine learning. Although these knowledge can be self-taught, it is much easier to acquire relevant knowledge through undergraduate, MFE and/or PhD degrees.
- Expertise- Even with multiple postgraduate degrees, it is still necessary to have a reasonable field knowledge of the underlying asset class or instrument type in order to continuously generate alpha from any advanced systemic trading technology. These expertise are usually gained through years of work experience, working at a specific counter counter counter of a bank or fund.
- The data- In general, data costs vary with sampling frequency, scope width, historical length, data quality and asset class/tool specificity. Advanced strategies rely on niche markets to generate alpha. Therefore, data can be very expensive. These costs must be considered in order for the strategy to be profitable.
- Research- If the strategy is used to trade more complex tools, then a dedicated retrieval environment is required. This usually means developing completely custom code from scratch. This is a huge investment of time. It also requires extensive software engineering skills to avoid introducing errors.
- Infrastructure- Even if a robust backtesting framework has been established to study advanced strategies, it also requires a complex infrastructure to transact. It may require full automation. It requires complex deployment, testing and monitoring.
- Capacity- Some advanced strategies are not effective because they are limited in capacity. Large funds cannot trade these strategies because the time invested is not worth the absolute return they can generate. This means that there is a cap on the amount of capital that can be applied to advanced methods.
It can be seen that while advanced trading strategies offer more alpha opportunities and potentially high profits, this requires more sophisticated mathematical knowledge, the necessary expertise and a more sophisticated automated trading infrastructure.
Summary
All in all, it is clear that simple trading strategies can be pushed to market much faster; they require much less expertise and can be executed manually even if the signals are automatically generated; however, they are more likely to be over-configured and lower profitability compared to advanced methods.
Complex strategies provide irrelevant alphas, reasonable profitability, and intellectual returns. However, this comes at the cost of higher data costs, more time spent on developing research and transactional infrastructure, and a deeper educational background.
The original link:https://www.quantstart.com/articles/simple-versus-advanced-systematic-trading-strategies-which-is-better/