stevehopwoodforex.com
https://www.stevehopwoodforex.com/phpBB3/
Print view

Baluda Indicators
https://www.stevehopwoodforex.com/phpBB3/viewtopic.php?t=629		 Page 30 of 81
Author:  Jemook [ Sun Jul 29, 2012 3:26 pm ]
Post subject:  Re: Baluda Indicators
Cool thanks for explaining that Gary. I'm not too fussed about the lower TF. It's the D1 I'm concerned about. For example:

EurUsd and GbpAud have dropped into ranging on D1 based on what Bob would have seen in his testing. However if I 'fix' the values using Baluda's brilliant indicator it tells me that those pairs are still trending short.

Quite a big discrepancy too. EurUsd is showing me -.6 vs -.37 unfixed. Hmm, which one to follow!
Author:  garyfritz [ Sun Jul 29, 2012 4:04 pm ]
Post subject:  Re: Baluda Indicators
You'd have to ask Paul about that. I'm not sure his code does what I think it should. It looks to me like e.g. in GetSlope() he just shifts his starting point on Sundays -- but then he does a 100-candle iATR() call. 100 D1 candles is 20 weeks with a "proper" crim, but 16.67 weeks with a Sunday-candle crim. Probably wouldn't make a big difference, but...

In calcTmaTrue() he skips over Sunday candles, which is the right thing to do. He's only looking at the close of the candle so you can just ignore the Sunday candles. But if TODAY is a Sunday, he doesn't skip it. Paul, you might want to look at that -- your init of dblSum always starts with today's candle, even if today is Sunday.

Hm, idea...

Paul's SlopeValues uses a fixed 20-candle lookback length. I just did an experiment with NT's MTF TMA indi, which lets you specify lookback. His D1 slope value was -0.37, like you're seeing. Then I changed the tmaPeriod from 20 (== 4 weeks if you have 5 D1 candles per week) to 24 (== 4 weeks if you have 6 D1 per week). The D1 slope changed from -0.37 to -0.55, close to what you're seeing. The difference probably comes from Paul *ignoring* the Sunday candles, thus effectively using a longer lookback period.

So I'd say the -0.37 value is closest to what NBob would have seen, because he didn't have the fix. The -0.37 value results from including Sundays and not having the full 4 weeks. I'd say it's probably not "right," since I think he intended to measure 4 weeks, but that's what he would have seen.

A bigger concern, and a more fundamental question about how you interpret this: the TMA slope value seems pretty sensitive to the lookback length. Which is not surprising, considering it encompasses more or less market movement as you change the length. E.g. in EU the slope with length = 15 is nearly zero -- which makes sense, since in the last 15 D1 candles (6 D1 per week) the price has hardly moved at all. Increase the length from 15 to 20 and it goes from -0.07 to -0.37 -- because in the 7/4 - 7/6 period, which just got added into the lookback, the price moved down a lot. Increase the lookback to 25 and it's -0.53, and so on.

The key point: THE TMA VALUES ARE NOT GRAVEN IN STONE. There really are no magic levels. The very same chart can produce different slope levels depending on how far back you look. The TMA slope just shows a measurement of how much the market has moved in the period you specify. How you interpret it -- what levels you choose to act on -- depends on what you're looking at and how you want to respond to it.

The "unfixed" levels are smaller (in this case) because it's effectively a shorter lookback, and the recent market action has been flat(ish). The "fixed" levels are higher because they look back farther, and that includes a sharp down move.

So rather than worrying about the exact slope levels, you should think about what it's telling you, and how you want to respond to that.
Author:  Jemook [ Sun Jul 29, 2012 4:22 pm ]
Post subject:  Re: Baluda Indicators
Great insights! Now everyone is going to yell "OMG the ranging and trending values repaint!!" haha. I wonder what happens if you make the look back value start from the beginning of the new trend (ie in an uptrend look back until the last low on daily, or where the uptrend began). Just a thought.. Will look into it more tomorrow.

Gary I love reading your posts. Sometimes I wonder who is the bigger nerd for time spent on this forum and I think we might be equal in that respect. But when it comes to divulging great insights I think you take the cake. Cheers.
Author:  garyfritz [ Sun Jul 29, 2012 6:59 pm ]
Post subject:  Re: Baluda Indicators
My pleasure, Jem. You've shared some great insights too!
Author:  Baluda [ Sun Jul 29, 2012 10:00 pm ]
Post subject:  Re: Baluda Indicators
Jemook wrote:I was just having a think about the AddSundayToMonday fix on SlopeValues. I am pretty sure Bob used Cowboy IBFX wen he came up with the ranging and trending values (.4/.8) and he didn't use the AddSundayToMonday fix when coming up with his results as it didn't exist.
So .4/.8 are the levels to use. Good thinking Jem.
garyfritz wrote:It looks to me like Baluda's code only pays attention to the AddSundayToMonday stuff when timeFrame = D1.
Correct, D1 is the only timeframe that is really messed up by those short Sunday candles.
garyfritz wrote:t looks to me like e.g. in GetSlope() he just shifts his starting point on Sundays -- but then he does a 100-candle iATR() call. 100 D1 candles is 20 weeks with a "proper" crim, but 16.67 weeks with a Sunday-candle crim. Probably wouldn't make a big difference, but...
The iATR isn't corrected for Sunday Candles. But I don't think it is an issue. Would you like me to try to solve the iATR too, Gary? Jem?
garyfritz wrote:In calcTmaTrue() he skips over Sunday candles, which is the right thing to do. He's only looking at the close of the candle so you can just ignore the Sunday candles. But if TODAY is a Sunday, he doesn't skip it. Paul, you might want to look at that -- your init of dblSum always starts with today's candle, even if today is Sunday.
The last version of SlopeValues has this fixed.
garyfritz wrote:A bigger concern, and a more fundamental question about how you interpret this: the TMA slope value seems pretty sensitive to the lookback length. Which is not surprising, considering it encompasses more or less market movement as you change the length. E.g. in EU the slope with length = 15 is nearly zero -- which makes sense, since in the last 15 D1 candles (6 D1 per week) the price has hardly moved at all. Increase the length from 15 to 20 and it goes from -0.07 to -0.37 -- because in the 7/4 - 7/6 period, which just got added into the lookback, the price moved down a lot. Increase the lookback to 25 and it's -0.53, and so on.
These are treacherous waters you sail in, Gary. A debate on this can and probably will last a very long time. Let's just say that I like the default 20. (And, let's not try to overthink this.)

Paul
Author:  garyfritz [ Sun Jul 29, 2012 10:35 pm ]
Post subject:  Re: Baluda Indicators
Baluda wrote:The iATR isn't corrected for Sunday Candles. But I don't think it is an issue. Would you like me to try to solve the iATR too, Gary? Jem?
There are two aspects to consider:

1) the different length (6 D1 vs. 5 D1 per week), which is probably a very small factor. I doubt a 20 vs. 25-bar ATR would change that much.

2) The short Sunday candle, which may be more significant. That short candle will reduce the ATR somewhat, artificially inflating the Slope. I don't offhand know how much difference it would make but I suspect it would be noticeable. I'll pull together a study and find out.
These are treacherous waters you sail in, Gary. A debate on this can and probably will last a very long time. Let's just say that I like the default 20. (And, let's not try to overthink this.)
My point was, the lengths are arbitrary, which means the amount of history you look at is arbitrary. I think it's best to be informed about what the indicator is really telling you, rather than treating it as a magic number. The length of the lookback can make a significant difference, and that's the biggest change between the "fixed" and "unfixed" versions.
Author:  garyfritz [ Sun Jul 29, 2012 11:21 pm ]
Post subject:  Re: Baluda Indicators
OK, the difference between 20 and 25 candles makes very little difference in the ATR -- generally less than 2-3%.

But the short Sunday candle makes a significant difference in ATR. The Sunday candle is only 3 hours long on GMT+0 crims like Cowboy IBFX, so it almost always has a very small range. That drops the ATR by about 10-20%, which inflates the slope by about 10-20%.

That's fairly significant. It inflates a slope value from 0.6 to about 0.7.

Notice also the "with Sundays" ATR is noisier, because those short candles throw off the average.
Author:  Dixpat [ Mon Jul 30, 2012 8:58 pm ]
Post subject:  Re: Baluda Indicators
I enclose my set file for MPBM

I was in a trade last night which closed out at the AdaptivePipsTakeProfit of 50

However on its way there the EA did not make any change to the trade for the BE setting or the Jumping SL setting

Do I have my set file correct for a basket of trades:

TP 50
SL 20
Set to BE + 10 at 25 pips profit
Jump the SL every 15

Is my logic warped or is the EA not doing what it should

Thanks
Author:  Baluda [ Wed Aug 01, 2012 10:51 am ]
Post subject:  Re: Baluda Indicators
Dixpat wrote:However on its way there the EA did not make any change to the trade for the BE setting or the Jumping SL setting
Dixpat,

Do you realize MPBM is a basket manager. This means all trades are handled as a whole rather than individually? MPBM will never set the SL or TP of a trade.

Paul
Author:  gms [ Wed Aug 01, 2012 3:02 pm ]
Post subject:  Re: Baluda Indicators
I have currencies showing up twice in the SlopeValues indi (v1.0.5.).

And I checked it is only once in the symbolsToWeigh string.

gary
All times are UTC Page 30 of 81

Powered by phpBB® Forum Software © phpBB Limited