Dynamic Trend and Multi-Kernel Regression with Staged Exits

Summary

This strategy combines an ATR and SMA based Dynamic Reactor with a weighted regression using Gaussian and Epanechnikov kernels. The reactor forms adaptive support and resistance bands and changes trend state when price crosses the prior band. The kernel estimate is compared with the reactor to produce directional signals, with a 200-period moving average and consolidation filter used to screen entries.

Trade management uses three profit targets at 1.5%, 3.0%, and 4.5%, dividing the position into 33%, 33%, and 34% portions, alongside a 1% stop. The document supplies parameter defaults and describes a backtest setup on SOL/USDT using hourly bars from February to August 2024, but gives no performance statistics. It warns that lag, overfitting, ranging or highly volatile markets, and incomplete order fills may impair results. The proposed refinements, such as volatility-based parameters and position sizing, are suggestions rather than tested improvements.

Key ideas

• ATR and SMA bands define a dynamic trend state and act as support or resistance.
• A weighted blend of Gaussian and Epanechnikov kernel regressions contributes to directional signals.
• A 200-period moving average and a consolidation filter restrict entries.
• The trade plan scales out at three profit levels and uses a shared percentage stop.
• The document reports no backtest results and flags lag, overfitting, market regime, and execution risks.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.