Trading Metabar Doji Signals with Linear Regression Trend Filters

Summary

The article develops a strategy around Doji patterns detected in metabars, which combine consecutive ordinary bars over a variable width. Because a Doji is treated as a possible reversal rather than a standalone entry signal, the method first checks for a preceding trend. It estimates that trend with linear regression on ordinary bars, excludes the bars making up the Doji metabar, and filters signals by trend direction, minimum slope, and channel width relative to the Doji’s height. The resulting indicator can provide signals for an automated strategy.

The article compares conventional-bar and metabar implementations using historical training and forward tests. It reports that metabar-based systems generated more trades, while forward-test performance was substantially weaker than training results across the strategies. The author concludes that more frequent signals can amplify either gains or losses and that Doji trading needs better system design and parameter selection. The evidence is tied to the tested settings, timeframes, currency pairs, and historical ranges; it does not show that increased trade count alone improves durable profitability.

Key ideas

• A metabar combines consecutive bars over a variable width, allowing pattern detection beyond fixed single-bar candles.
• The strategy requires a preceding trend before treating a Doji as a potential reversal signal.
• Linear regression estimates trend direction and slope, while channel width helps compare trend movement with the Doji’s range.
• The metabar versions produced more trades in the reported tests, but forward results were weaker than training results.
• More frequent trades can magnify losses as well as gains, so signal frequency does not establish profitability.

Tags

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