6 Points Of Refinement for Transforming Manual Trading Ideas to Automated Trading Strategies
This is a continuation of our 10 part series for Automating Trading Algorithms. In this part, we examine the critical points of definition of manual trading strategies that must be explicitly defined in order to automate.
Entry and Exit Conditions Dependent Upon Current Portfolio
Strict entry and exit conditions for trades must be adopted by anyone looking to automate trading strategies. There is no manual guesswork when implementing exact figures — trading software, including IBKR and QuantConnect, will require exact figures when implementing entries, exits, liquidations, and position sizing.
Most individuals tend to size their positions dependent upon the net cash on hand of their portfolio — ratios which must be reflected in their sizing algorithms. It is also important to precisely measure this, or risk having orders rejected by your broker. Margin maintenance and margin calls can also be inadvertently triggered by imprecise order creation.
Conditions in terms of trading signals are often used in this situation. However, what most people do not realize is that when their simple conditions are triggered, they are often triggered multiple times within a short timeframe, and the reverse conditions could also be triggered simultaneously. Therefore, it is important to smoothen signals for entries, implement techniques such as cool-down periods and false alarm detection, and set parameters for max position sizes as a percentage of cash on hand.
Leverage
Oftentimes, especially in trading futures and forex pairs, one has access to leverage. It is critically important that one executes trades that do not put them in a situation where margin calls or margin maintenance is imminent. Futures often allow 10–20x leverage, while forex could go as high as 100x leverage on collateral. With imprecise adjustments on risk, one can lose more than invested in the markets very quickly. Be sure to have proper risk mitigation in place, preventing leverage from resulting in margin calls.
Trading Indicator Definitions
This should be easy, right? Indicators such as moving averages and RSI are already defined and pre-built in numerous languages. But, one must be very careful on their execution, and what they are truly calculating. For starters, the parameters must be precise, the moving average over 100 and 200 days are completely different statistics that can result in different performance and execution of your ideas. As well, people overlook the timeframe that indicators are computed on. 1 minute bars result in much different statistics than rolling 4 hour bars.
As one combines and creates more complex indicators from a base set pre-provided by brokers and software systems, these issues compound, resulting in potentially undesirable results. This happens often when using support-resistance trading algorithms, as support and resistance lines calculated on higher timeframes can still change often enough to divest from trader calculated lines.
News Events, Volatility Spikes, and Spread
Oftentimes, when impending news events are near, the spread on forex pairs, futures, options, and even stocks tend to widen. Afterwards, volatility tends to increase in the case of surprise results to both the upside and downside. It is important to take this phenomenon into consideration when constructing automated algorithms, as volatility spikes can result in numerous losing trades in a row with quick entries and exits by the bots. It is often recommended to turn off bots and close all positions before and after news events for a specified period of time, and/or until volatility spikes come down.
Data Integration from Alternative Data Sources
Many quantitative traders use alternative data sources to guide trading decisions. It is critical that the ideas and indicators based on alternative data sources, including those provided by the broker, are validated in practice and checked for consistency against the sources used in manual trading.
In some cases, this would require exact checks on data formats and values of commonly published data, such as federal funds rates, put call ratios, short interest, economic metrics, and fundamental data on the security.
In many cases, one must find other estimators to compensate for the inexact matches in available integrations. Oftentimes, this happens when using sentiment scores, institutional money flow / smart money concept estimators, and even options greeks (can vary by method).
Backtesting Against Manual Trades Made
This leads us to our final recommendation — backtest (either by yourself or by a professional) your newly created automated trading algorithm against dates where you manually traded your strategy. Check for divergence of ideas, increased or decreased sharpe ratios, probability of backtests overfitting, and low drawdowns, just to name a few. Oftentimes, brokers allow for backtesting using recorded live data, and allow paper trading. It is key for backtesting performance to both be acceptable and reasonable — 10000% returns per year combined with Sharpe Ratios of greater than 10 do not qualify or meet this criteria.
Conclusion
We highlight six key points of refinement for transforming manual trading ideas into automated trading strategies. These include defining strict entry and exit conditions, taking into account portfolio considerations, leveraging, properly defining trading indicators, considering the impact of news events and volatility spikes, and integrating data from alternative sources. It is important to note that automating trading strategies requires precision and attention to detail, as imprecise adjustments can result in margin calls and other negative consequences. Furthermore, it is important to validate and test ideas and indicators based on alternative data sources for consistency.
