Support and resistance - Fib calc

[roughbert]

I've made a Pivot Point calculator which sources prices from Google and I want to add Fib levels to it - how are these calculated please?

 

[TheBramble]

If you're using google for your feed then 1, 1.5 and 2 should do you nicely.

 

[roughbert]

If you're using google for your feed then 1, 1.5 and 2 should do you nicely.

to what pray? closing price or pivot point?

 

[GammaJammer]

I think your question indicates that you don't really understand what Fibonacci levels are, how they are calculated, how they are used etc. Which begs the question - why on earth are you using them?

If, rather than asking people here for this information, you simply look it up yourself (there are plenty of web resources for this) you might actually learn something, rather than expecting people to spoon feed you.

Or, do it your way, but you will have learned little of import.

GJ

 

[roughbert]

...pedant!!

Of course I'm looking, I was just making conversation. Did I make your day?

I think your question indicates that you don't really understand what Fibonacci levels are, how they are calculated, how they are used etc. Which begs the question - why on earth are you using them?

If, rather than asking people here for this information, you simply look it up yourself (there are plenty of web resources for this) you might actually learn something, rather than expecting people to spoon feed you.

Or, do it your way, but you will have learned little of import.

GJ

 

[TheBramble]

The thing is, once you re-discover 'ignore' you find so many more to which to apply it.

Sort of harks back to the charity thread where, if you extend my (il)logic here, by removing the ignore facility, less individuals worthy of its utilisation would show up.

 

[HaloTrader]

I think your question suggests you don't understand Fibonacci levels at all - They aren't calculated the same way as pivots - They are used to measure the length of a retracement, therefore you need to take the high and low of the trend before the retracement and calculate percentages to represent important Fibonacci levels - I think its much easier to just draw fibonacci levels on the graph itself.... Which makes me think that maybe you have no clue what a fib. level is :S ?

 

[roughbert]

Thanks!

Apparently the high point of the range (previous day's high) is the 100% point. The low point is the 0% point. The other points slot in at the Fib numbers.

Note to self - never ever reply to an honest inquiry in a tightassed or demeaning manner. Got that? OK, well just make sure you don't forget it. Right, I'll try, boss. You do that, Roughbert.

 

[GammaJammer]

Apparently the high point of the range (previous day's high) is the 100% point. The low point is the 0% point. The other points slot in at the Fib numbers..

Which is 0 and which is 100 depends on whether you're plotting the retrace levels from an up move or a down move.

Oh, and it's not just about plotting the day's range. That's not a 'move' in itself. If you're looking at daily charts, it's likely the sort of moves that you will want to plot fib retracements (or maybe even projections) on are going to last multiple days.

Hence you can't really just bang it all quickly into a spreadsheet or whatever - you would have to define what you consider to be a move (i.e. coding for higher highs, higher lows etc).

 

[NetTecture]

actually no, this is not proper fiobonachi. Theyx draw trends. Even INTRA day, the low -> high movement may be constructed of more than one trend.

There simply is no way to calculate them for a daily base given only the previous days price. None.

 

[roughbert]

There simply is no way to calculate them for a daily base given only the previous days price. None.

Unless the previous day maintained the same trend, or mostly the same trend, all day?

GammaJammer - given the data source (open close low high for a series of days), one could look for a trend in previous data and calculate Fib on that.

What formula would define a trend? It would have to allow for small reversals. Maybe a slope calculation would do it as averaging would be built in. But a positive slope might well have two components, a gentle start then a sharp upturn, it would be a pity to mix them up. A turn like that could be picked up by a MA crossover. Would a MA crossover be a good place to start?

 

[GammaJammer]

Ok - start at the beginning. Do you know what a trend is - could you find one on a chart if you saw it? Do you know what you think constitutes the beginning and the end of a move?

 

[NetTecture]

Unless the previous day maintained the same trend, or mostly the same trend, all day?

No. For a trend valid all day before you have no idea whether this is true. If the last day is a continuation of a trand of the day before, you need the high/low of 2 days ago to calcualte fib retracement.

If you are in an uptrend, close at the high of the day, and this is the third day up... the fib levels would encompass all three days from the low to the high. So, if you only enter the last day, you still do not calculate the right values.

 

[roughbert]

No. For a trend valid all day before you have no idea whether this is true... So, if you only enter the last day, you still do not calculate the right values.

Good. So a trend is not divisible, the start has to be detected. Thanks.

 

[NetTecture]

Well, it may be divisible. But the fact is that a fibonacchi retracement level is DEFINED as a retracement of the complete trend, and as such a fibonacchi retracement level can not be computed on a part of the trend 😉

 

[roughbert]

Well, it may be divisible. But the fact is that a fibonacchi retracement level is DEFINED as a retracement of the complete trend, and as such a fibonacchi retracement level can not be computed on a part of the trend 😉

Identifing the start of a trend is the problem. Would a "three-line break " color change be a good place to start?

 

[NetTecture]

Hm, that is a good question. It really is - I fight the same thing. I would say some "higher high, higher low" mechanism plus some stoachstic filter to filter out miniscule movements (i.e. below 5% of average bar) may be in order. This is something I am trying things out myself at the moment 😉

 

[roughbert]

Hm, that is a good question. It really is - I fight the same thing. I would say some "higher high, higher low" mechanism plus some stoachstic filter to filter out miniscule movements (i.e. below 5% of average bar) may be in order. This is something I am trying things out myself at the moment 😉

you could use an averaging method (including best fit to a line) to get around small reversals and so on, but then to detect a change of slope (either best fit or MA) or best fit to a regression line, for example, I think you would need an iterative method.

 

[NetTecture]

I think more about a haiken ashi indicator and when it turns "around" use extreme points plus a percentage filter to find out when the reversal move qualifies as "significant enough". This also would work very nicely as stop point for positions 😉 Heiken Ashi ways of calculating trends are pretty efficient.

 

[roughbert]

I think more about a haiken ashi indicator and when it turns "around" use extreme points plus a percentage filter to find out when the reversal move qualifies as "significant enough". This also would work very nicely as stop point for positions 😉 Heiken Ashi ways of calculating trends are pretty efficient.

That's interesting. I shall try it out.