Empirical Evaluation of Trend Filters and Exit Configurations in a Renko-Based Algorithmic Trading System

Technical Summary: Empirical Evaluation of Trend Filters and Exit Configurations in a Renko-Based Algorithmic Trading System

Problem Statement
Algorithmic trading systems typically integrate signal generation with trend confirmation and exit rules. While technical indicators (e.g., moving averages, MACD, ADX) are widely studied, the specific interaction between a Renko-based price representation, a chosen trend filter, and an exit mechanism remains under-explored in controlled settings. A critical methodological challenge arises when using non-standard chart types, such as Heikin Ashi or internally generated Renko bricks, which TradingView notes may produce synthetic price representations differing from real-market execution. This study addresses the question: How does the choice of trend filter and exit configuration affect the simulated behavior of a Renko-based trend breakout strategy under a common historical backtesting environment? The research deliberately avoids predictive claims, focusing instead on documenting observed differences in performance metrics across controlled variations.

Methodology
The study employed a controlled empirical design using the Renko Trend Breakout Strategy 3.6.2, implemented in Pine Script v5 on the TradingView Strategy Tester.

• Data & Environment: The experiments utilized approximately one year of historical data (October 1, 2025 – October 2, 2026) for BITSTAMP:BTCUSD on a 1-minute Heikin Ashi chart. Crucially, the Renko representation was generated internally by the strategy (using an ATR-based brick size with length 14 and multiplier 1.0) rather than using TradingView's native Renko chart type.
• Strategy Parameters: Fixed position size (1 unit), initial capital of $100,000, pyramiding limited to one, and a commission of 0.01% per transaction. Both long and short positions were permitted.
• Experimental Factors: The study created twelve controlled experiments by crossing six trend filters with two exit configurations:
  1. Trend Filters:
     • Renko Higher-High/Higher-Low (HH-HL)
     • EMA200 (Internal Renko close vs. 200-period EMA)
     • MA Crossover (50-period vs. 200-period EMA)
     • MACD Trend (12, 26, 9 parameters)
     • Supertrend (Period 10, Multiplier 3.0)
     • ADX Trend (Length 14, Threshold 20)
  2. Exit Configurations:
     • Base / Reverse-Logic Exit: Relies on the strategy's intrinsic reversal signal to close positions.
     • Explicit Risk-Management Exit: Adds fixed parameters: 400-point Take-Profit (TP), 100-point Stop-Loss (SL), and a Trailing Stop (activating at 100 points with a 100-point trail distance).
• Evaluation Metrics: Performance was assessed via Net Profit, Return, Total Trades, Win Rate, Profit Factor (PF), Expectancy, and Maximum Drawdown.

Key Contributions
The paper's primary contribution is the documentation of a stark, consistent divergence in simulated strategy behavior based solely on the exit architecture. By holding the strategy logic and historical period constant, the study isolates the impact of exit mechanisms on:

1. Trade Frequency: Explicit risk management significantly increased trade counts compared to the base configuration.
2. Profitability: The shift from a reversal-based exit to a fixed TP/SL structure inverted the profitability of the strategy across all tested trend filters.
3. Metric Interdependence: The study highlights that win rate alone is an insufficient metric for strategy evaluation, demonstrating cases where low win rates yielded positive expectancy and high win rates yielded negative expectancy depending on the exit logic.

Results
The results revealed a binary outcome based on the exit configuration:

• Base / Reverse-Logic Configuration: All six trend filters produced positive simulated net profit. Net profits ranged from $38,442 (MA Crossover) to $354,575 (ADX Trend). Profit factors were consistently above 1.0 (ranging from 1.081 to 1.324). Notably, the EMA200 filter achieved a positive profit factor (1.284) despite a low win rate of 22.07%.
• Explicit Risk-Management Configuration: All six trend filters produced negative simulated net profit. Net losses ranged from -$223,595 (MA Crossover) to -$400,658 (Renko HH-HL). Profit factors fell below 1.0 (ranging from 0.694 to 0.759), and expectancy became negative for all configurations.
• Trade Frequency: The explicit configuration generated substantially more trades than the base configuration (e.g., Renko HH-HL increased from 18,165 to 31,625 trades), leading to higher accumulated commission costs and more frequent premature exits.

Significance and Limitations
The authors assert that the principal finding is the consistency of the configuration-level difference: within this specific simulation, the exit architecture materially altered the system's behavior, turning profitable strategies into losing ones when fixed risk parameters were applied.

However, the paper maintains a modest stance regarding the broader implications:

• No Universal Superiority: The results do not prove that reverse-logic exits are universally superior or that fixed stop-losses are inherently detrimental. The failure of the explicit configuration is attributed to the specific parameterization (TP 400/SL 100) interacting with the dynamic Renko brick size and volatility regimes.
• Simulation Constraints: The authors explicitly state that results are limited to the TradingView simulation environment using synthetic Heikin Ashi and internal Renko data. They caution that these results should not be interpreted as evidence of live-trading profitability.
• Data Snooping: The study acknowledges the potential for data-snooping bias due to testing multiple configurations on a single historical sample without applying formal multiple-comparison corrections (e.g., White's Reality Check).

In conclusion, the study serves as an exploratory empirical record demonstrating how exit mechanics can dominate strategy performance in a controlled backtesting environment, urging caution against generalizing these specific parameter results to live market conditions.