How Algorithmic Autotrading Strategies Work

In the prior article, I explained what Autotrading is at a high level.  Now, let’s take a peak underneath the hood of the iQ (InvestiQuant) Autotrading engine and learn how autotrading strategies work. 

Data Capture 

The first step in the algorithmic trading strategies process is historical data capture.  Algorithmic autotrading strategies require large amounts of clean, robust market data for optimal evaluation and signal generation. Sophisticated software and 3rd party data feeds are utilized to capture and cleanse historical market and price data for a diverse array of contextual market inputs: 

• Market Events (e.g. FOMC Fed Announcement, Options Expiration,  Futures Rollover, etc.),

• Economic Reports (Employment, CPI, PPI, earnings, etc.), 

• Price Patterns (new market highs/lows, overbought/oversold levels, unfilled price gaps, etc.), 

• Volatility (VIX, ATR, etc.), 

• Calendar Considerations (day of week, day of month, month, quarter, etc.), 

As the saying goes, “garbage in = garbage out” and InvestiQuant expends considerable resources to capture, cleanse, organize, store, update, and monitor nearly a thousand different data points (most of which update daily) that help its algorithms evaluate the real-time state of the market. When determining the optimal signals and risk parameters for a given trading day, context is king.

Strategy Tracking

But market data alone is not enough to generate a trading signal.  Every signal is created by a well-constructed strategy with very strict trading parameters (long or short, target, stop) for a given trading instrument (futures contract, option, or share), market (S&P 500, AAPL, etc.) and environment (trend, volatility, etc.). Every strategy should be thoroughly researched, analyzed, tested, traded, and refined before entering production. 

Once a strategy is trading live in production, both its historical and ongoing live trading results should be stored and tracked in databases for real-time evaluation each day.  Maintaining a comprehensive database of strategies and their daily performance across all market environments is critical for enabling them to adapt to changing conditions, as well as their own performance. With the help of advanced financial machine learning techniques, the most sophisticated trading algorithms (like many of those utilized by InvestiQuant’s autotrading programs) will automatically deactivate until their performance improves, or a more conducive environment is present. 

Note: One of the most common reasons algorithmic trading strategies fail is because they were over-optimized originally (typically by a novice developer), or their algorithmic rules were not adaptive. Meaning, the rules are too simple and rigid, and not engineered to evolve with the ever-changing markets. The adaptive nature of InvestiQuant’s strategies and machine learning techniques is what has allowed them to stand the test of time and perform across a wide range of market environments (e.g. raging bull markets, volatile bear markets, etc.)

Signal Identification

Algorithms come in many shapes and sizes and utilize a series of processes, calculations, and other problem-solving operations to achieve a desired outcome. Even tying your shoelaces is a form of an algorithm! Of course, trading algorithms are a bit more involved, to say the least. 

The 3rd step of the algorithmic trading strategies process is algorithmic signal identification.  The goal of this step is to identify which trading strategies’ requirements (market conditions, price patterns, volatility levels, events, calendar considerations, etc.) are met and should be considered for execution. In other words, for which strategies are market conditions favorable for trading today? 

Risk Management and Signal Selection

Once all of the potential strategies for a given day have been identified, they are then segmented into the autotrading programs that each supports. (InvestiQuant offers different size autotrading programs powered by algorithmic trading strategies to accommodate the wide range of desired investment levels of traders and investors. The larger the program the more strategies it includes.)

Each iQ Autotrading program operates with a predetermined maximum daily risk budget (varies by program, typically 3-5%) and a suggested funding level ($30k - $1M).  For example, the iQ Trader Program consists of about 15 different algorithmic strategies, and its max daily risk budget for a given day is $2.5k.  If the trade criteria for four of its strategies are met and their collective required risk budget (i.e. the amount of risk needed to allow them to trade optimally) is $3k total, then the signal selection algo will only have enough risk budget to select three of the strategies’ signals for execution that day. 

Signal Execution 

After the filtering process of risk management and signal selection, the signals are then ready for electronic distribution.  That day’s trading signals are electronically sent within milliseconds of selection by our signal server (co-located with the CME exchange in Chicago, Illinois) to our executing brokers’ order management software which processes and converts the signals into electronic orders for execution at the exchanges on behalf of InvestiQuant’s clients. 

The brokers utilize sophisticated APS (average price sold) algorithms to ensure that all clients earn the same results (to the penny) that I get in my account for each signal executed. 

Algorithmic Autotrading Strategies - Powered by IQ

Since 2008, InvestiQuant has armed traders and self-directed investors with institutional quality algorithmic trading strategies based on statistical edges. InvestiQuant’s AI-driven automated solutions help clients better protect and grow their wealth, hands-free—regardless of the direction of the broader stock market. Request autotrading information today to learn more!