No. of Recommendations: 6
I am actually surprised that GTR1 doesn't seem to have EMA.GTR1 does have an EMA capability: sgema(...)
Here's an example, using the 50-day EMA compared to the 200-day EMA for SPY as a simple timing method. I chose this because both the SMA and EMA measures for these two lookback periods have been discussed for many years.
gtr1.net: GTR1 BacktesterIf you change the first step to "EMA_Diff > -999", results will show SPY with no timing for that period (1993-2025). The 50/200 EMA timing shows a tiny drop in CAGR and improvement in measures that include "risk".
The 50/200 parameters can be changed to improve performance, at least for SPY for this period. For example, using 100 and 350 improve both CAGR and risk measures, and decrease annual turnover. I'm sure searching over the longer period allowed by using Robbie's ^S5T would find different "better" parameters. And doing the same for QQQ/N1T or other indexes would have their own best measures.
