Want to improve your trading results? Here’s a guide to fine-tune your adaptive trend -following strategies for changing market conditions. This article explains how to classify market regimes, set clear rules for entry and exit, and manage risk effectively.

Key highlights include:

• Classify Market Regimes: Use tools like ADX, ATR, and Bollinger Bands to identify trending, ranging, or predicting volatility in various markets.
• Baseline Strategy Rules: Define consistent entry, exit, and position sizing rules to avoid emotional trading.
• Walk-Forward Optimization: Test and adjust your strategy across rolling data windows to ensure performance stays consistent.
• Risk Management Overlays: Protect your capital with maximum drawdown limits and volatility-adjusted stops.
• Regular Reoptimisation: Periodically update your strategy to align with current market trends.

sbb-itb-466c9b0

Define Your Market Regime Framework

Before diving into parameter adjustments, it’s crucial to identify your market regime. Markets typically operate within four main regimes: Trending (clear directional movement), Mean-Reverting (oscillating or ranging), Random/Choppy (indeterminate), and Crisis/Extreme Volatility. Each regime calls for different strategies. For instance, applying a trend-following strategy in a ranging market often leads to unnecessary losses. This framework is the backbone for accurately recognising and responding to market conditions.

The consequences of ignoring market regimes can be severe. Take the case of Marcus Chen, a trader with a S$128,000 account, as an example. During the “Golden Run” in September and October 2023, his trend-following strategy thrived, achieving a 75% win rate and earning him S$18,420 in profit as the SPY exhibited strong trends with an ADX between 28 and 35. However, when the market shifted to a ranging regime from November 2023 to January 2024 (ADX dropped to 17), his unchanged strategy backfired, resulting in a staggering S$43,280 loss – a 29.4% drawdown – as his win rate plummeted to 27%. He managed to recover S$28,640 by implementing a regime detection system using tools like ADX, ATR, and Bollinger Bandwidth to filter trades.

“Identifying the current regime is the prerequisite for selecting appropriate tools and setting realistic expectations.”
– Ken Nobak, Market Analyst, FractalCycles

Classify Market Conditions

To classify market regimes, rely on specific indicators to quantify each condition:

• Trending markets: Consistent directional movement, typically with an ADX above 25 and a Hurst exponent exceeding 0.55.
• Ranging markets: Price oscillates around a stable mean, with ADX below 20 and a Hurst exponent under 0.45.
• High volatility or crisis conditions: ATR spikes beyond 2 standard deviations above its mean.
• Compression or choppy markets: Tight price ranges, indicated by Bollinger Band width below 4%.

Develop a Regime Dashboard that tracks four key metrics: ADX (14-period) for trend strength, ATR ratio for volatility, Bollinger Band width for range expansion, and volume relative to its 20-day moving average. Use a multi-indicator scoring system to classify a market as “Trending” only when at least three signals align. For Singapore-specific instruments like the STI or SGX-listed stocks, adjust thresholds based on typical volatility patterns. For instance, blue-chip stocks may require different ATR ratios compared to mid-cap stocks.

Document Regime Transition Rules

After defining regimes, establish clear transition rules to avoid false signals. Implement a 3-bar confirmation rule where a new regime signal must persist for three consecutive bars before switching strategies. For example, transitioning from a ranging to a trending regime might require ADX to stay above 25 or the Hurst exponent to remain above 0.55 for three days.

Set specific thresholds for each regime shift. A trending regime might require:

• ADX > 25
• Bollinger Band width > 3% of the middle band
• Moving average (MA) separation > 2%

For ranging conditions, criteria could include:

• ADX < 20
• Prices oscillating within 1.5 ATR of the moving average

Document these rules using a finite-state machine model with clear triggers, such as crossovers or threshold breaches. This ensures consistency across all market conditions and reinforces the systematic approach outlined earlier in the article.

Establish Your Baseline Strategy Rules

To build a strong trading approach, start by defining a baseline strategy with consistent and testable rules. This process begins with identifying market regimes and then crafting a core trading strategy that specifies entry, exit, and risk management rules. These rules act as the backbone of your trading system. Without them, any adjustments risk becoming guesswork, making it hard to separate real improvements from random fluctuations.

A solid baseline strategy typically starts with a primary trend filter. For example, align trades with a 200-period Exponential Moving Average (EMA) and confirm momentum using the RSI (Relative Strength Index). Only consider long trades when the RSI is above 50 and the price is above the EMA. To avoid getting caught in erratic, choppy markets, incorporate a volatility filter. Use the Average True Range (ATR) as a guide – only enter trades when the 14-period ATR is above its own moving average. This combination provides a clear and logical framework that’s easy to monitor and even automate.

“A fixed-period indicator that works well in a strong uptrend may fail miserably in a choppy, sideways market.”
– PyQuantLab

This foundational structure sets the stage for detailed entry, exit, and risk management rules.

Set Entry and Exit Criteria

Entry signals should be clear and repeatable. One effective method is a price action breakout confirmed by trend and momentum filters. For instance, you could use a 50- and 200-period moving average crossover when trading Singapore-listed stocks.

Exit rules should follow multi-layered logic to protect your capital while locking in profits. You can define three types of exits:

• Trend reversal: Exit when the price drops below the 50-period moving average.
• Fixed risk boundary: Set a stop-loss, such as 20% below the entry price.
• Profit target: Use a take-profit level, like 40% above the entry price.

Some traders add a time-delayed exit, where the crossover signal must persist for several bars before closing the position. This helps avoid exiting prematurely during minor pullbacks in a strong trend.

Determine Position Sizing Rules

Position sizing is critical for managing risk. A common approach is to allocate a percentage of your total account equity – 10% is a typical benchmark. Instead of using fixed lot sizes, this method ensures your positions adjust proportionally as your account grows or shrinks.

Adding a volatility-adjusted sizing layer can further protect your capital. In volatile markets, you might widen stop-loss levels (e.g., from 2x to 3x ATR) while reducing position sizes to maintain consistent risk. For example, with a S$100,000 account and a 10% allocation, you’d allocate S$10,000 per trade. If your entry price is S$50 and your stop-loss is S$45 (2x ATR), your risk per share is S$5. To limit your risk to 2% of equity (S$2,000), you would buy 400 shares (S$2,000 ÷ S$5).

To safeguard against extreme losses, consider setting a maximum drawdown threshold, such as 30%. This “circuit breaker” would automatically close all positions if your strategy’s overall retracement exceeds this limit, helping to protect your remaining capital during unexpected market events.

Select Indicators and Parameter Ranges

When crafting an adaptive trading strategy, it’s crucial to choose the right indicators and set appropriate parameter ranges. A good strategy typically uses three types of indicators: trend, momentum, and volatility.

For tracking trends, tools like Exponential Moving Averages (EMA) and Donchian Channels are effective. A common length for Donchian Channels is 20 periods. To gauge the strength of a trend and filter out sideways markets, the Average Directional Index (ADX) is a reliable choice. For momentum, the Relative Strength Index (RSI) works well; it signals bullish conditions when above 50 and bearish conditions when below 50. Lastly, volatility is best measured using the Average True Range (ATR), which helps in setting dynamic trailing stops and adjusting reversal thresholds based on market conditions.

“Traditional moving averages suffer from a fundamental dilemma: short periods respond quickly but generate false signals, while long periods lag behind genuine trends.”
– Nicolae Filip Stanciu, Quantitative Researcher

Choose Relevant Indicators

To refine your strategy, select indicators that complement your baseline rules.

• Trend indicators like EMA alignment provide insights into the broader market direction.
• Momentum filters such as RSI or Stochastic confirm the strength of market moves. For example, Stochastic might use a K value below 30 to signal buying opportunities or above 70 for selling, ensuring trades align with pullbacks.
• Volatility measures like ATR dynamically adjust stop-loss levels. Depending on your trading timeframe, ATR periods can range from 5 to 250 days.
• Strength indicators like ADX are essential for distinguishing trending markets from ranging ones, helping you avoid choppy price action.

Keep it simple – use one indicator per function to prevent unnecessary complexity. Once you’ve chosen your indicators, set clear parameter ranges to ensure your strategy remains robust.

Define Parameter Ranges

Setting parameter ranges involves balancing thoroughness with computational efficiency. For dual moving average setups, it’s critical to maintain logical relationships: the slow-moving average must always exceed the fast-moving average to preserve crossover logic.

An example of this in action comes from a study conducted in August 2025 by Andres Romo and Ricardo Soto at the Pontificia Universidad Católica de Valparaíso. They developed a hybrid trading system for BTCUSDT on Binance Futures, combining a dual Simple Moving Average (SMA) strategy with a Learning-Based Linear Balancer (LB2) optimiser. In their system, the fast SMA ranged from 2 to 1,344 intervals, while the slow SMA ranged from 10 to 1,344 intervals. This approach achieved a maximum ROI of 17.2% during the best-case test period, which included 34 training/test splits with overlapping 60-day windows.

For stop-loss and take-profit boundaries, ATR multiples are highly effective. For example, a stop-loss set at 3x ATR is robust in high-volatility environments like 1-minute crypto markets. Take-profit levels can range from 5.5x ATR (TP1) to 19x ATR (TP3) for tiered profit locking.

Testing the sensitivity of these parameters is crucial. If small changes in values lead to significant performance drops, the strategy is likely over-fitted.

“If the net profit dropped off sharply at values other than [the optimal], then you would look for other look-back lengths around which the net profit was more stable.”
– Mike Bryant, Founder, Adaptrade Software

Lastly, consider adding a “volatility spike” filter to block trades when price movements exceed 3x ATR. This can help protect against erratic market conditions.

Conduct Walk-Forward Optimization

Walk-forward optimisation (WFO) is a method that fine-tunes your strategy to adapt to changing market conditions. Unlike traditional backtesting, which applies fixed parameters to the entire dataset, WFO reoptimises parameters for each new segment of data. This process involves dividing your data into sequential segments, with an In-Sample (IS) period for optimisation and an Out-of-Sample (OOS) period for validation. As the testing progresses, both windows are rolled forward by the length of the OOS period until the dataset is fully covered. By stitching together the results from the OOS periods, you create an equity curve that closely mirrors real-world trading scenarios.

“Walk-forward optimisation simulates how you could work with the strategy during real trading – you can optimise it on some historical data and then trade it with the optimal values.”
– Mark Fric, StrategyQuant

Set Up Training and Testing Windows

To implement WFO effectively, it’s essential to define the training (IS) and testing (OOS) windows clearly. A common setup uses 70% of the data for the IS period and 30% for the OOS period. The IS window should be long enough to capture major market cycles but still sensitive to recent price movements. For instance, in February 2025, Ajay Pawar from QuantInsti showcased this method by using a rolling five-year IS window to optimise portfolio weights, followed by a one-year OOS window for validation. He began with training on data from 2011–2015, tested on 2016, and continued rolling forward through 2025.

When setting up your windows, don’t forget to include transaction costs, commissions, and slippage in your calculations. Ignoring these elements could lead to overly optimistic profitability estimates. Once your windows are defined, the next step is to assess the stability of your strategy’s performance.

Evaluate Results for Stability

To ensure your strategy isn’t overfitted, compare its performance in the IS and OOS periods. A significant drop in performance between the two suggests overfitting. Instead of focusing solely on total profit, prioritise risk-adjusted metrics like the Sharpe Ratio. Consistency across multiple OOS segments is critical. For example, if a strategy’s IS Sharpe Ratio drops from 2.31 to -0.25 in an OOS segment, it’s a clear red flag for overfitting.

A practical rule of thumb is to require at least four out of six walk-forward runs to be profitable before considering a strategy reliable. Additionally, avoid relying on a single successful configuration after testing dozens of variations. A strategy that performs well across various configurations is far more likely to withstand real-world market conditions.

Evaluate Strategy Performance Across Regimes

After completing your walk-forward optimisation, the next step is to evaluate how your strategy performs across different market regimes. This helps confirm whether adjustments made for market conditions lead to meaningful, risk-adjusted improvements. Break down the performance by regime labels – such as trending, ranging, high volatility, and low volatility – and examine key metrics within each regime to uncover hidden vulnerabilities.

When assessing performance, prioritise risk-adjusted returns over raw profits. Metrics like the Sharpe ratio (which measures return per unit of risk) and the Sortino ratio (which focuses on downside risk) provide deeper insights. Additionally, monitor Maximum Drawdown – the largest equity decline from peak to trough – and the time it takes to recover, ensuring alignment with your risk controls. Professional traders also rely on the Profit Factor (gross profit divided by gross loss), targeting values between 1.3 and 1.6 after factoring in transaction costs. Another key metric is the Correlation Coefficient of the equity curve, which should ideally stay above 0.95. This indicates smooth, linear growth rather than erratic or unpredictable movements.

“Robust backtesting is the difference between a plan that survives changing regimes and one that collapses on first contact with live markets.”
– Daniel Cheng, Financial Analyst, Singapore Forex Club

Benchmark Against Buy-and-Hold

To validate the effectiveness of your adaptive strategy, compare its performance against simpler benchmarks like Buy-and-Hold or a non-optimised version of your own strategy. If the adaptive strategy fails to consistently outperform these benchmarks, its added complexity might not be worth it. For example, in August 2025, researchers from the Pontificia Universidad Católica de Valparaíso tested an adaptive 2-SMA system on BTCUSDT futures. Using 34 overlapping 60-day windows from August 2020 to January 2022, the optimised strategy achieved an average ROI of 7.9% in unseen test periods, with a maximum ROI of 17.2% in its best period. This demonstrated statistically significant outperformance over both Buy-and-Hold and non-optimised versions.

To ensure the results are not due to chance, use statistical methods like the Wilcoxon Signed-Rank Test. Additionally, apply a “realism surcharge” by adding conservative buffers to spreads and slippage estimates. This ensures your strategy’s edge holds up under real-world trading conditions. These comparisons strengthen the case for your adaptive strategy and pave the way for continuous performance tracking.

Monitor Rolling Performance

Regularly evaluate your strategy using 12-month rolling windows to observe its performance over time. This approach reveals whether your strategy remains consistent or suffers from periods of underperformance. For instance, a momentum strategy might achieve a Sharpe ratio of 1.82 during trending, low-volatility regimes but drop to -0.41 in choppy, high-volatility conditions. Recognising these patterns allows you to make timely adjustments, such as modifying position sizes or tightening risk controls during challenging regimes.

Pay close attention to regime-specific attribution to identify which market conditions enhance or hinder your strategy’s edge. If your strategy struggles during regime transitions, consider adding a confirmation mechanism – for example, requiring a new regime state to persist for three bars before making parameter adjustments. This can help avoid false signals and reduce the impact of whipsaws. By monitoring these details, you can catch performance issues early and maintain a more resilient strategy.

Implement Risk Management Overlays

Unexpected market shocks can disrupt even the most well-planned strategies. Risk management overlays act as a safeguard, automatically limiting losses during extreme conditions. By using dynamic controls that adjust to real-time market behaviour, you can better protect your capital. For instance, tracking peak equity and calculating current drawdown as a percentage of that peak can help you respond swiftly. If equity drops by 30% from its peak, an automated kill-switch can close all positions immediately, stopping further losses. Additionally, using volatility-adjusted stops, such as those based on the Average True Range (ATR), allows for flexible exit points – widening during volatile periods and tightening when markets stabilise.

“Adaptive algorithms… can self-adjust which enables them to reduce exposure to risk during turbulent market phases.”
– Bookmap

Below are two key methods to strengthen your risk management framework: setting maximum drawdown thresholds and defining rules for volatility spikes.

Set Maximum Drawdown Thresholds

Before deploying any strategy, it’s crucial to establish a clear maximum drawdown limit. A common benchmark among institutions is 30%, meaning positions are closed automatically if equity declines by 30% from its peak. You can calculate this in real time using the formula:
(Peak Equity – Current Equity) / Peak Equity × 100
For example, if your peak equity was S$100,000 and it drops to S$72,000, your drawdown is 28% – nearing the limit.

In addition to a kill-switch, consider using trailing stop-loss multipliers that adjust dynamically based on favourable price movements. Research from Pontificia Universidad Católica de Valparaíso highlights the effectiveness of this approach. Between August 2020 and January 2022, a hybrid 2-SMA system trading BTCUSDT futures applied trailing stop multipliers (e.g. 0.95× the peak price for long positions) alongside fixed stop-win thresholds. This setup achieved an average ROI of 7.9% during test periods, while also protecting the strategy against market reversals during high-volatility phases.

Establish Volatility Spike Limits

Markets can become chaotic in an instant, making it essential to have rules for handling sudden volatility spikes. Tightening stop-loss settings during these periods can help limit damage. For instance, you might adjust stop-loss distances from a standard level (e.g. 3× ATR) to a more conservative level (e.g. 2× ATR) when volatility surges.

Keep an eye on order book dynamics and trading volume, as these can provide early warnings. A sharp drop in volume often signals a potential trend reversal, while sudden spikes in sell orders – visible through heatmap visualisations – may indicate an impending flash crash. During heightened volatility, adaptive algorithms can also shorten moving average periods, such as switching from a 50-day MA to a 20-day MA. This enables quicker responses to price changes and reduces risk exposure. These adjustments ensure your strategy remains flexible and avoids unnecessary risks when market conditions shift.

Monitor and Reoptimise Regularly

Once you’ve set up an optimisation framework, it’s important to revisit and fine-tune it regularly. Markets don’t stay the same, and periodic reoptimisation ensures your strategy keeps up with these changes. But timing is everything – reoptimise too often, and you may face unstable parameters and higher trading costs. Wait too long, and your strategy could become outdated. Research from Talyxion Research highlights that monthly rebalancing works best for volatile markets like cryptocurrencies, improving the Sharpe ratio by 1.07 in a 36-month study compared to static parameters. On the other hand, weekly rebalancing resulted in “excessive parameter instability” and increased turnover costs.

“The monthly rebalancing cycle balances adaptivity with stability. More frequent rebalancing (e.g., weekly) leads to excessive parameter instability and higher turnover costs, while quarterly rebalancing fails to adapt to the rapid regime shifts characteristic of crypto markets.”
– Duc Bui and Thanh Nguyen, Talyxion Research

The next step is to figure out how often you should reoptimise for your specific market.

Define Reoptimisation Frequency

The frequency of reoptimisation should match the pace of the market. For fast-moving cryptocurrency markets, a monthly update is recommended. Slower-moving markets, like traditional equities, may allow for quarterly reviews.

Keep an eye on turnover rates, as frequent adjustments can eat into profits through fees and slippage. To manage this, you can apply filters that prioritise parameters with a rolling Sharpe ratio above 1.3. Additionally, tools like volatility multipliers (e.g., ATR) can help stop-loss and take-profit levels adjust automatically between reoptimisation cycles.

Track Parameter Changes Over Time

Once you’ve settled on a reoptimisation schedule, it’s crucial to monitor and document how parameters change over time. This helps ensure your strategy stays in tune with market conditions and actually improves performance. Use earlier walk-forward testing results to check for stability.

It’s also helpful to track metrics like the Calmar ratio for risk-adjusted performance. Pay attention to how your parameters perform in different market conditions, such as trending versus choppy periods, and adjust dynamically rather than sticking to a fixed setup. Lastly, always include transaction fees and slippage in your simulations to confirm that your adjustments remain profitable.

Conclusion

This checklist brings together regime classification, dynamic optimisation, and adaptive risk controls into a well-rounded system. The goal? To refine adaptive trend strategies using a structured, data-driven framework that removes emotional biases. As Bookmap highlights, adaptive algorithms “do not experience fear or greed, nor do they require rest”. This method replaces gut feelings with precise, adaptive execution.

One key issue it addresses is the limitation of traditional moving averages, which often force traders to choose between fast responses prone to false signals and slower responses that lag behind trends. By classifying market regimes, leveraging walk-forward optimisation, and applying adaptive risk overlays, this system adjusts dynamically to trending, ranging, or uncertain markets.

Backed by empirical research, this approach emphasises continuous refinement to stay effective. For instance, a study on BTCUSDT futures showed that a 2-SMA strategy, optimised using a learning-based metaheuristic, achieved an average ROI of 7.9% during unseen test periods, with a peak return of 17.2% in its best-performing window.

To ensure stability, walk-forward analysis can validate performance across rolling windows. Metrics like the Calmar ratio help assess risk-adjusted returns, while position sizes can be scaled based on regime confidence – up to 95% of capital in strong trending markets, but limited to about 10% in uncertain conditions.

For more insights, visit the Collin Seow Trading Academy website (https://collinseow.com).

FAQs

How do I know my strategy is overfitted?

When a strategy delivers outstanding results on historical data but struggles with new, unseen data, it may be overfitted. Red flags include returns that seem too good to be true and minimal drawdowns, which often suggest the strategy has latched onto random noise rather than actual market behaviours. Overfitting happens when insignificant fluctuations are misinterpreted as meaningful trends, resulting in disappointing performance during live trading. To minimise this risk, steer clear of excessive fine-tuning of parameters.

How often should I reoptimise my parameters?

When using trend-following strategies that adjust to market conditions, it’s a good idea to revisit and fine-tune your parameters every six months. This regular adjustment helps you stay aligned with shifting market dynamics and keeps your strategy performing as intended.

What’s the simplest way to detect market regime changes?

The easiest way to spot shifts in market regimes is by relying on real-time indicators such as ADX, ATR, and Bollinger Band width. Here’s how they work:

• High ADX signals a strong trending market, while low ADX points to a range-bound or sideways market.
• ATR and Bollinger Band width are excellent tools for assessing market volatility and identifying consolidation phases.

By combining these indicators into a multi-factor approach, traders can better identify changes in market conditions and adjust their strategies accordingly. This method ensures a more responsive and informed trading process.