Position Trading

Algorithmic Precision: Engineering Positional Trading Systems in AmiBroker

Systematic positional trading is the art of identifying a major market shift and maintaining commitment until the structural narrative changes. Unlike the high-frequency churn of intraday trading, positional systems thrive on low turnover, reduced transaction costs, and the power of compounding. AmiBroker, with its robust AmiBroker Formula Language (AFL), provides the ideal environment for traders to translate long-term macro-theses into high-performance, backtested algorithms.

In the world of AFL, a positional strategy is not merely a set of buy and sell arrows. It is a comprehensive mathematical model that accounts for trend confirmation, volatility-adjusted exits, and intelligent capital allocation. This exploration details the technical requirements for constructing an institutional-grade positional trading engine within AmiBroker.

The Positional Philosophy in AFL

Position trading requires a different temporal perspective. In AmiBroker, this means shifting the focus from 5-minute charts to Daily and Weekly intervals. The primary objective is to capture the "fat tail" of the market distribution—those rare but significant moves that account for the majority of a portfolio's annual return.

The Statistical Edge of the Long View Positional trading significantly reduces the impact of "noise" and "slippage." In AFL, we prioritize signals that exhibit "persistence." A signal that persists for multiple days on a daily chart is far more statistically significant than a fleeting crossover on a sub-hourly timeframe. Our goal is to minimize the "Churn Rate" while maximizing the "Profit Factor."

The power of AFL lies in its array-processing nature. Instead of looping through data points like traditional languages, AFL treats price history as a continuous stream of data, allowing for lightning-fast backtests across decades of market history.

Core AFL Functions for Trend Detection

To build a positional system, one must master the specific AFL functions designed for trend identification and persistence. These functions form the building blocks of our logical triggers.

The Cross() Function The fundamental trigger for any trend system. It identifies the exact moment an indicator pierces through a threshold or another average.
HHV() and LLV() Highest High Value and Lowest Low Value. These are essential for creating "Donchian Channels" or breakout systems that trigger only after multi-month highs.
Ref() and ValueWhen() These allow the system to "look back" at previous signals to determine if a current move is a continuation or a divergence from historical norms.

Building the Dual-Indicator Framework

A professional positional system rarely relies on a single indicator. Instead, we use a "Confluence of Evidence" approach. A classic institutional setup involves a trend filter (like an Exponential Moving Average) and a volatility filter (like the Average True Range).

The Signal Logic

In our example strategy, we seek to enter a position only when the shorter-term trend is accelerating above the long-term trend, provided volatility is within an acceptable range to allow for a wide trailing stop.

AFL Logic Wrapper
// Parameters for Optimization
FastPeriod = Param("Fast EMA", 50, 20, 100, 1);
SlowPeriod = Param("Slow EMA", 200, 100, 300, 1);

// Core Logic
FastEMA = EMA(Close, FastPeriod);
SlowEMA = EMA(Close, SlowPeriod);

// Entry Trigger
Buy = Cross(FastEMA, SlowEMA) AND Close > SlowEMA;
Sell = Cross(SlowEMA, FastEMA);

// Plotting for Visual Confirmation
Plot(FastEMA, "Fast", colorBlue, styleLine);
Plot(SlowEMA, "Slow", colorRed, styleLine);

This basic logic ensures that we are always trading in the direction of the primary trend. However, a positional trader knows that the "Sell" signal is often the most difficult to time. Relying on a crossover for an exit can often lead to giving back a large portion of the gains.

Engineering Dynamic Position Sizing

In positional trading, the distance between your entry and your stop loss is often much larger than in swing trading. If you trade the same number of shares for every ticker, you are exposing your account to uneven risk. AFL allows for "Volatility Adjusted Sizing."

By using the ATR() function, we can calculate the current "shake" of the market. If we want to risk exactly 1% of our total capital on a single trade, we divide that 1% by the current ATR (multiplied by a multiplier like 2.5). This ensures that we buy fewer shares of volatile assets and more shares of stable assets, equalizing the dollar risk across the portfolio.

The Optimization and Curve Fitting Trap

AmiBroker’s optimization engine is incredibly fast, but this is a double-edged sword. It is easy to find the "perfect" periods for a moving average that would have made millions in the past. This is known as curve fitting.

For a positional system to be robust, the parameters should work across a wide range of values. If your strategy makes money with a 200-day EMA but loses everything with a 190-day EMA, the strategy is fragile and will likely fail in live markets.

Metric Positional Target Red Flag Value
Profit Factor 1.8 - 2.5 Below 1.2
Max Drawdown Under 25% Over 40%
Recovery Factor Above 4.0 Below 2.0
Exposure 30% - 60% Over 90% (Too much heat)

Walk-Forward Analysis for System Robustness

To combat curve fitting, professional AFL developers use Walk-Forward Analysis (WFA). This process involves optimizing the system on a slice of past data (In-Sample) and then immediately testing it on the following slice of data (Out-of-Sample) which the system has never "seen."

If the performance holds up during the Out-of-Sample periods, you have a system that possesses genuine predictive power. AmiBroker automates this entire cycle, providing a "Walk-Forward Efficiency" (WFE) score. A WFE score above 60% is generally considered the threshold for a tradable positional system.

Scaling Scans with the AmiBroker Explorer

A positional trader doesn't need to look at charts all day. They use the AmiBroker Explorer to scan thousands of stocks at the end of the day or week. By writing a simple "Filter" variable in AFL, the software can provide a list of stocks that have just triggered a breakout or a trend-following signal.

The Exploration Code Structure To turn your strategy into a scanner, simply add the following to your AFL:
Filter = Buy OR Sell;
AddColumn(Close, "Price", 1.2);
AddColumn(ATR(14), "Volatility", 1.2);
AddColumn(EMA(C, 200), "Trend Line", 1.2);

Summary of Professional Best Practices

Building a positional trading system in AmiBroker is a journey of refinement. The most successful systems are often the simplest ones, as they are less prone to breaking when market regimes shift. As you develop your AFL scripts, keep the following institutional principles in mind:

Avoid Over-Indicatorization More indicators do not equal more profit. Each additional parameter increases the risk of curve fitting. Stick to one trend filter and one volatility filter.
Prioritize Exit Logic Position traders often enter correctly but exit poorly. Use a "Chandelier Exit" or an "N-Bar Trail" to ensure you stay in the trend as long as possible while protecting capital.
Maintain a Global View Use AmiBroker's "Watchlists" to segregate high-quality stocks. A positional system is only as good as the universe of stocks it trades within.

Ultimately, AFL is a language of logic. It removes the emotional burden of the market and replaces it with mathematical certainty. By mastering the functions, the backtesting engine, and the optimization protocols of AmiBroker, you empower your capital to work with the relentless force of the market's primary trends.

Final Expert Note: Always remember that a backtest is a simulation of the past. Real-world execution involves "slippage" and "execution delay." Always subtract at least 0.1% per trade from your backtest results to account for these real-world frictions.

Related Posts

Scroll to Top