Adaptive MAs such as FRAMA, Kauffman's AMA, Laguerre and VIDYA do a good job of smoothening out the noise without introducing lag. The AMA have a distinct STEEP slope (almost vertical) during trends and are FLAT (horizontal) during trading ranges.

How do i generate a Buy/Sell signals using the slope change?
i.e. BUY when AMA Turns up and is STEEP (don't buy when it turns up but is flat)
SHORT when AMA Turns down and is STEEP

Please suggest sample code or an overlay indicator. Note that -

1) I couldn't use ROC to detect steepness because it's as a % of underlying price, so the formula can't be used across a data-set of symbols at different price levels.

2) I tried crossover and crossover of Bollinger Band of AMA (or moving avg of the AMA), but the crossovers don't distinguish well between steep and flat regions

3) I looked at the wiki but none of the documentation or links on the Web suggested good Buy/Sell signals for AMAs.

thanks
Kiran

QUOTE:
1) I couldn't use ROC to detect steepness because it's as a % of underlying price,

ROC accepts any DataSeries you pass to it: FRAMA, AMA, etc.

QUOTE:
3) I looked at the wiki but none of the documentation or links on the Web suggested good Buy/Sell signals for AMAs.

How about LinearRegSlope?

ROC and LinearRegSlope are good proxies of slope - but their values are absolute, not normalized across symbols and time periods (e.g. it ranges from 2.5- -2.5 for AUD, but when i change the timeframe or symbol, the range is different)

To normalize this rule across all symbols, I was thinking of using a large period SMA (e.g. SMA30) as a cross-over for ROC - e.g.
BUY when ROC(FRAMA, 3), 30) Crosses Over SMA(ROC(FRAMA,3), 30)

-> Would you recommend any other Normalizing indicator that statistically provides a better signal to show sharp slopes?

thanks
Kiran

In version 2019.09 of Quandl provider, we've improved fundamental data update speed by ignoring requests for the unsupported symbols like Forex, international stocks if they exist across your DataSets.