Using/Delaying McGinley Dynamic seeding to chart parabolicsc.mql5.com
In summary:
McGinley believed moving averages are not supposed to be used as a trading signal, and instead identify the main trend. The formula is designed to go slower when price is trending up, and faster when price is trending down, to mimic how investors react to market movements.
This does not work when an asset starts at a low price then rises astronomically. Unless Gann and friends are right about markets having full cycles, I don't believe bitcoin will be under $100 any time soon. Therefore x bars are ignored to give a usable McGinley Dynamic.
As a trend identification tool, the McGinley Dynamic will trail far behind during forceful uptrends. The utility created here is based on the assumptions that:
Less the extremely parabolic assets, an uptrend will retrace eventually to the McG, and the trader using McG understands that the moving average trails far away to keep a trader secured in position by slowing down for movement upwards
The trading strategy employed is not based solely on moving averages
Like all moving averages, McG will always suffer in terms of giving clear trading signals when the trend is too strong, too weak, too volatile, etc. vs a coincidental/leading indicator
Length chosen should be 60% of your chosen MA to account for lag. In this case, a McG of 15 is equal to a 25 EMA (as TradingView uses EMA as the base MA for McG).
Bars ignored should be before an uptrend, and only ignored up to where the MA is usable for the particular ticker.
Something I have noticed is to be cautious when publishing ideas or sharing charts with indicators that weren't intended for sharing, especially for those who do not have the ability to publish invite-only scripts.
While scripts are not meant to be copied straight off a chart, in rare cases there are ways to grab an indicator for self-use when the author did not intend such. As a first step preventive measure and self-incentive to keep scripts updated, it would be ideal to apply a 'expiration date' using the timenow function to published and nonpublished indicators that are accessible to the public. Although indicators are mostly derivatives of each other (except my stuff, I'm a real snowflake), there are many such cases in which seemingly 1:1 copies pop up, and there is no real way to identify who is in the right, if anyone even is. It is possible (and not uncommon) for indicators to be 'replicated' just by coincidence, so that should not be ruled out either.
Example of McGinley Dynamic outperforming an EMA in the whipsaw department:
Formula as described by Investopedia:
MD = MD-1 + (Index – MD-1) / (N * (Index / MD-1 )^4)
where MD-1 is the previous period's moving average, N is length, and index is price source. Adjustments can be made to the formula to optimize, but at what point do you scrap it and move on to other, more appealing methods? The wide berth that McG gives is not very useful over candlestick analysis for near instant reversal identification.
