Force Study to Not Repaint?

[JoshF]

Is there a way that we can force a study to not repaint once a signal fires? We all know that repainting ****s and especially with MTF studies it makes false signals. What I am trying to figure out and backtest is if I forced it to not repaint, and actually traded on every signal fired which may actually stay (or would have disappeared based on future action), how successful would that study really be? Any ideas would be great.

 

[wtf_dude]

Is there a way that we can force a study to not repaint once a signal fires? We all know that repainting ****s and especially with MTF studies it makes false signals. What I am trying to figure out and backtest is if I forced it to not repaint, and actually traded on every signal fired which may actually stay (or would have disappeared based on future action), how successful would that study really be? Any ideas would be great.

The thing is, things repaint based on what the conditions are when a candle completes. If a signal is generated midway through a candle and then is repainted later, it means that conditions are "not true" upon completion. What you are basically asking is if a signal can be kept as "true" if its true at ANY point through out the duration of the candle completing. Does that make sense? and is that really what you want? A signal that is true midway and then disappears later is, by definition, a false signal.

 

[JoshF]

Makes sense and yes, I think that is what I am trying to do. Assess the true profitability even with false signals included. The study is already set such that a signal does not appear until the previous bar is closed with a "true" condition, however could disappear later based on future moves. I am trying to prevent these signals from disappearing after already being "confirmed". Since we can't see the future and I would have acted based on this "confirmed" signal I would like to be able to backtest based on this.

 

[wtf_dude]

Makes sense and yes, I think that is what I am trying to do. Assess the true profitability even with false signals included. The study is already set such that a signal does not appear until the previous bar is closed, however could disappear later based on future moves. I am trying to prevent these signals from disappearing after already being "confirmed". Since we can't see the future and I would have acted based on this "confirmed" signal I would like to be able to backtest based on this.

hmm I think your best bet would be to see if you can do a multi-time frame setup. So lets say you're on the 5 minute chart, the signals would be generated from the 1 minute chart and if any of the 1 mins had a signal, it would show the 5 min candle as a signal triggered. I'm far from an expert on multi-time frame work, but we have a few guys on here who can probably give some advice.

 

[JoshF]

Yep, and there is the issue. Its a custom study Ive been building and testing for months based on a MTF TMO using convergence/divergence of the signal/main lines for advanced signals (works really well imo). I'm using a 1 min chart, with 1min, 5min, 15min plots and based on a combination of criteria I could get a "true" signal based on current 1min criteria, but can change later if criteria based on the 5 or 15min go "false".

Anyway thanks for the help, was just curious if there was a way to lock-in a signal once fired, regardless of true/false criteria in the future.

EDIT: I would be interested in doing a collaboration with someone if they are interested in the study

 

[wtf_dude]

@JoshF what if you changed the signal to a value. So lets say on the 1min you fire a signal then the signal trigger = 1 and any additional signal is another 1. Tthen for the 5 minute candle you would just put your signal as value>=1. Does that make sense?

 

[Tidan]

What if you used open instead of close and possibly a read ahead value?
Something like:
If close <= open[-1]...
Or
If open[-1] > x...

 

[MerryDay]

What if you used open instead of close and possibly a read ahead value?
Something like:
If close <= open[-1]...
Or
If open[-1] > x...

Recommend you put it on your chart first. Nothing beats personal experience The only way you will know what works is to play with the settings. Repainting doesn't just occur on the current candle, in the case of MTF, they repaint going back 16-20 candles and highest/lowest repaint going back even more. So I would think that a read-value would add lag and have no effect on repainting.

 

[rottentrade]

What if you used open instead of close and possibly a read ahead value?
Something like:
If close <= open[-1]...
Or
If open[-1] > x...

By definition, close = open[-1], unless there's a gap, so it wouldn't make any difference.

Anyway, has there been any fix to the repaint issue?

 

[MerryDay]

By definition, close = open[-1], unless there's a gap, so it wouldn't make any difference.

Anyway, has there been any fix to the repaint issue?

There is no way to 'force' repainting indicators to not repaint. It is the basis of the code.