Always treat others with respect and kindness, even if you disagree with them. Avoid making personal attacks or insulting others, and try to maintain a civil and constructive tone in your discussions.
- Manlobbi
Halls of Shrewd'm / US Policy
No. of Recommendations: 2
What is the difference between rrs(regression relative strength) and rsi(Relative Strength Index)?
Which one does the simple relative strength calculation:
"price(today) / price(1 year ago)"
No. of Recommendations: 10
What is the difference between rrs(regression relative strength) and rsi(Relative Strength Index)?
RRS uses all data points between (today) and (1 year ago), and builds a least-squares regression line from them. It's an open-ended number that's higher when the trend is a gain.
RSI is a kind of exponential-average comparison of gains and loss days. As such, it uses only (today) and its own accumulators. It's a range-bound number (typically 0 to 100) that is nearer the top as gains exceed losses.
Which one does the simple relative strength calculation:
"price(today) / price(1 year ago)"
Nether. RSS is at least a different measure of the same general idea.
- Jamie
No. of Recommendations: 2
: trp(1,252) top 20 #Example
No. of Recommendations: 5
No. of Recommendations: 5
I don't use GTR1 as a screener tool. And maybe nobody can right now? as the database seemed to stop being updated on 11/18/2025.
I've only used GTR1 as a backtesting tool, preferably using parameters that can be easily gotten from various data sources available on the internet.
rrs and rsi are not freely available, they seem to be GTR1-only.
With a bit of work, tr (total return = return including dividends) can be found.
Relative strength (price today vs. price NN days/weeks ago) is readily available, especially for common lookbacks of 1 year, half year, quarter, month. Other periods, such as 10 months, can be computed with a little bit of work.
So for using it to backtest with parameter values I can easily get, it seems like the best rule to use would be "ratio(aprc(1,1),aprc(1,252)) top 10"
Actually, and interestingly, 1 year lookback backtests for Nasdaq100 using rrs, rsi, trm, and price-only (aprc) all show similar backtest results.
No. of Recommendations: 7
So for using it to backtest with parameter values I can easily get, it seems like the best rule to use would be "ratio(aprc(1,1),aprc(1,252)) top 10"
If you use unadjusted pricing for your selections, aprc is the correct factor. If you use adjusted pricing (dividends, splits, etc.), you should use gprc for backtesting.
Aussi
No. of Recommendations: 5
Rayvt:” Actually, and interestingly, 1 year lookback backtests for Nasdaq100 using rrs, rsi, trm, and price-only (aprc) all show similar backtest results.”
One year lookback on a test with only 100 stocks isn’t a very robust evaluation. The noise in using just two single total returns at the start and end of a period is significant.
I just ran a test with selecting 1 to 50 stocks out of >7000 stocks over 26+ years and:
rrs over all the tests had an average of 22% better CAGR with a 15% lower annual turnover than trp.
But it obviously requires more computations.
