Hello,
what can you do against repainting of indiors? Whenever you do a visual style backtest of an empty EA and you drop the indior on the chart, you really see how it performs (repainting or not),
opposed to simply placing it onto a live chart (too slow to evaluate repainting). If you have noticed that it will counter, what can you do against that?
Can there be a general rule that you have to follow? Or do you need to go through the code, step by step?
As an example I've attached a Hodrick-Prescott filter, an econometric procedure to split timeseries in a fashion part and a cycle component. I enjoy it a lot,
but only it repaints... How do I clean this up code, therefore I get a realistic perspective of where it will have been around the past bars?
I cant imagine what 'nobs' number you're using but you have to really like it
you would need to add a delay on the entire thing, so it is repainted/calculated/reported in a predetermined interval.
Kenz987,
thank you for your code. Strange thing you only begin to see the difference when doing the item: you don't find any difference when you have dropped them on the chart. Any pointers on how best to see that immediately without having to do the EA backtest that is empty? (this should be how all indiors are shown in the first place I presume, to avoid the many false Eureka-moments. .)
Kd3726,
nobs isn't that's important, it is the lambda that's the smoothing parameter. The main reason why I really like it's that Hodrick-Prescott is the best design of a timeseries, the reason I hate is it isn't a causal filter: it utilizes data to the right of the current pub to compute the current pub HP-value, therefore, not actually usefull in real-time trading, unless you can find some kind of change...
Reply With Quote
