P & F & UK Shares

Hi,
gosh - this has been busy while I've been at work! The description of chart construction - 'extend in current direction as first choice, only check for reversal if the existing column doesn't extend by one or more boxes' - is correct for H/L charts. You are charting one bar at a time, so to speak, ie one set of O, H, L and C info for a chosen period... to see what happens over the course of a day you want intraday data and to chart based on the OHLC of each X minute period... pick your own X to suit how 'granular' (good word, well picked <g>) you want it to be. The docs with Pfscan explain the chart construction method. Again this'll look like an advert, but Pfscan includes a signal tester, that is still being developed further, to see how signals work out in practise... the current V2.79 'trades' all signals, the update I'm working on adds a few things to this and being able to select the signals to trade is - I hope - soon going to be up and running in there too.
To find 'failed' triple tops, for example, you'd run the tester and open the CSV file it generates (there are commentary logs output as well, outlining where 'decisions' are taken) and sort alphabetically to get the Triple tops - you'd then check the returns generated by your chosen exit tactic (short list of things like stops and retracements, V2.80 has ATR based exits, not in 2.79) and a 'good' failure would be tested to see that the T.Top had peaked at/near the signal level, then dropped a 'good' percentage say.
I trade large caps with P&F - darned infrequently over the past year due Pfscan etc admittedly - but my preferred spectrum of stocks is basically the FT350 (preference for the 100, but I will take a 250 share that looks good) and a group I made up from the US market that is mainly the top two or three stocks by mkt cap in each of around 50-60 sectors. Most of what I trade is longer term, held beyond a year sometimes, or as little as a week if the signal plays out quickly, and probably in the top 150 or so of the UK or US market by cap.
Dave
 
Hope you don't mind me chiming in here guys but I wanted to clarify one of Jonny's replies.

Fastnet wrote:

From your email I understand that if, for example, the price is currently describing an upward column but then, intraday drops once or twice and bounces off the same resistance that is more that 3 boxes away (assuming 3 box rev) BUT the price gave a high for the day slightly higher than the previous day then only the continued movement upwards would be added??

Jonny replied:
Correct, that is a rule of PnF charts

Jonny - in the example above (using Fastnet's proposed EOD system) are you saying that you'd actually plot the day's high and not the closing price (which itself may not actually extend the plot upward)? I always thought that it's the level of the closing price (not the intraday high) that determines whether you continue to plot upwards or instead start to plot a column in the opposite direction. Is this not the case? Happy to be corrected of course and have much enjoyed reading your automated PnF exploits on one of the other TTW boards...good luck with it!

Thanks in advance.
Chris
 
Hi Chris,

Proper PnF uses the days high and low prices and not the close price.

JonnyT
 
Seconded - you'll see (normally) P&F H/L charts, but some people plot close based ones instead. For the H/L you ONLY use the day's high and low prices. Take share A today, yesterday it ended in an X column say, today you check the High therefore, if the high is a box or more above the column top as it stood yesterday then you extend the chart by however many boxes and totally ignore the low, end of story. IF the high doesn't add 1 or more boxes to yesterday's last column, then (and only then) do you look at the low - if that price is low enough to cause a reversal (the amount being equal to box size x reversal count, so on a 2% log chart, 3 box reversal that'd be 2% x 3 = a low that is 6% below the X column top showing from yesterday) you stop charting the X column and draw an O column down from the box below the previous top... ie the O column heads south, and has a 1 box step down at the top compared to the X column it has taken over from.
If share B was dropping yesterday, and in an O column, then you check today's LOW first. If that low drives the O column a box or more lower then that's what you do, you then totally ignore the high price. If it doesn't set a new low you check to see if the High is high enough to create a reversal - if it is then start an X column up, if not then do nothing.
So it's ALWAYS (1) Extend current column if possible, (2) ONLY if you can't do (1) should you look for a reversal. If big enough swing occurs then start a new column.
The idea is more sensible than it sounds - if a share is climbing, and sets a new high, then we're weighting the chart in favour of the continuing upward pressure... if the peak was early in the day and a fall is starting then it'll become apparent in tomorrow's chart when there's no new high and the fall continues.
Dave
 
Thanks DJB for putting that one to rest!!

I am v new to P&F and, at this early stage at least, find them very exciting. As Marc Rivalland comments in his book P&F charts are designed to cut through the random noise and clearly show the balance of power between bulls and bears.

However, I do have reservations. I am sure that P&F charts if properly drawn give excellent and fairly reliable signals but I am concerned that updating every night for the following day will merely show you the excellent signals that you have missed!!

I am confused about the different methods used in drawing the charts. The method described above seems to be a modification to the underlying philosophy behind P&F which allows you to update day by day. However, as mentioned above, this introduces a time factor into the charts. The main idea behind P&F is that they are independent of time and are instead a graphical representation of price action whenever it occurs. To omit what might well be significant action just because it took place intraday would undoubtedly lead to a different charts and hence signals being created.

I wondered if anyone had looked into the extent of the effect of this modification on the results obtained?

Thanks for your patience - especially DJB!!

Fraser
 
To omit what might well be significant action just because it took place intraday would undoubtedly lead to a different charts and hence signals being created.
The same thing happens when you consider a 1 minute chart compared to a 15 minute chart etc. Yes - you will miss intraday events by only using end-of-day data. But you will catch events that occur over longer timescales instead.

It is said that Technical Analysis is scale-invariant. That is to say, it works on all timescales. So if you use end-of-day data, you will still find TA signals, but it will take longer for them to form.

Even on a 1 minute chart, you are missing some of the action. Read the posts about Level 2 and you will realise that by the time a price is on the chart, the opportunity that existed at L2 is gone!

Hope this helps,

-svengali
 
Hi Svengali -

I am afraid it looks like we might have to agree to disagree on this one.

With all other charting forms I am familiar with your latest comment would indeed be true and I agree with you entirely.

However the main difference with P&F is that (in theory at least) many days or even weeks can pass by without any new columns being added. If the price traded within a tight range that was inside the reversal point (eg 3box) but also never exceeded previous highs then nothing would be added to the chart.

The only thing that moves the chart along the x-axis is price movement of significance as defined by you. If your boxes are large (2.5-3%) then there will be less action to update.

Time does not come into it. This is why it is artificial to update using just the High/Low of the day just gone.

I thought for a long time that P&F was simply another way to graphically describe price action with time. I thought that one column was added per day. The only benefit, I believed, was that the random noise was screened out and levels of support and resistance were easy to spot.

P&F should record any significant (as defined by your chart) price action whether it happens intraday or over a number of days. The point is that days are a measurement of time and time is not relevant with P&F.

The method described above by DJB is a modification to P&F theory to simplify the P&F chart construction process.

My question is whether this modification makes an appreciable difference to the results. If it does not then I will happily use this method and stop mithering about the minutia of P&F chart construction !!! ;)

Cheers

Fraser
 
Hi Fastnet,

I think the issue you raise is addressed by Dave's last post.
If I'm correct then your concern is that using an EOD system will mean that you miss signals generated by the intraday activity. While this is probably true, as I see it we are always going to miss signals generated within timescales shorter than the one we choose to work with. Hence Jonny's point about granularity of charts and Sven's point about missing signals generated within a 1 minute timeframe when you are only looking at 15 minute charts.

Dave's point about using the day's H or L for your PnF plot ensures that you are at least reflecting the day's strongest sentiment, whether in the long or short direction. So while operating within the constraints of an EOD system, you could still be generating signals from the most significant of any intraday H/L movements.

Remember, trends don't just exist in the minute-to-minute timeframes...they also exist (and will likely cover a greater range) in daily, weekly, monthly and even yearly timescales. Your challenge therefore is to determine which timeframe you want to work with. If you can stay glued to your screen then minute-to-minute timeframes will certainly throw up (shorter term) signals that you would otherwise miss. However, as a newcomer like myself to the PnF arena I would suggest working with timesframes that give you an opportunity to think, analyse and respond appropriately.

Time does come into it even if the chart doesn't show this directly e.g. daytraders using PnF (jonnyt) obviously need to use price points across a much shorter timescale (every 5 mins) if they are to identify signals, though these don't necessarily make it to the PnF chart. So while they may use 96 x 5 minute prices to determine how their chart looks during the day, you are going to use just one H/L price to create yours.

Hope this helps....feel free to question anything I've said.

Chris
 
Last edited:
Hi Nautical and thanks for your reply -

I think I'm done this one to death to be honest and I'm wondering how much it really matters to the signals that are generated.

I know what I'm trying to explain but am having great difficulties getting this idea across in words . . . . . . guess that's ultimately down to me.

Basically time, whether it be minute, day, month is totally irrelevent with true and unmodified P&F. The chart measures price action alone independantly of when this action occured. So, in this instance, to say that one must chose a time period and try and trade it (whether it be 15min daytrading or looking at the weeklies) does not make sense.

Let's leave it there and agree to differ. . . . maybe it matters not - I don't know.

Thanks to all that have joined this discussion.

Cheers -

Fraser
 
[...] the main difference with P&F is that [...] many days [...] can pass by without any new columns being added. If the price traded within a tight range [...] then nothing would be added to the chart.
Yes.

P&F should record any significant (as defined by your chart) price action whether it happens intraday or over a number of days. The point is that days are a measurement of time and time is not relevant with P&F.

No. PnF should record any significant price action portrayed in the subset of data points that you choose to consider. Some people use 1 minute data to feed their PnF charts - they get very different charts to the people using EOD data for their charts.

Both are using a subset of all data points, and so both see features on a different timescale. If you use 1 minute data, you find features that emerge over minutes. If you use 15 minute data, you find features that emerge over hours. If you use daily data, you find features that emerge over weeks.

As with all charting techniques, you need to choose an appropriate timescale. Since you have already stated that you must use daily data, you cannot see features that occur intraday.

I hope this helps to clarify things!

-svengali
 
Yep,
on a quick scan I think Chris made the point that matters here... and I think it's a point you've not quite grasped, it isn't that intraday moves are ignored at all... the H and L prices show how far in each direction the pressure went, whether it bounced 5 times enroute is immaterial incidentally, it's which way the supply : demand argument was decided that counts. Where the HL chart 'goes wrong' is it assumes a single box in the existing direction outweighs a 20 box move reversal (to exaggerate)... Tuesday had an X column up to 100, (2 pt box) so the Weds high at 102 sets a new box end of story - disregarding the fall to 50 on the same day! <g> So on Thursday either the price rebounds to 104 region OR the reversal check is made (no new box unless 104+ hit by the high). So you might be a day late with the reversal signal - to overgeneralise a bit this is a price you might, now and then, pay for taking the unimportant moves out... you occasionally get the signal a day late. Not often though, and mostly it's a case of throwing out 98% rubbish - that 2% good goes out as well doesn't matter, it's the ratio of good to bad that counts!
If you want to ensure you don't miss short term intraday then use intraday prices for P&F instead... you make up H and L price data for the chosen time period, and use a small boxsize. Using EOD data you're trading over days+, as you increase the boxsize so you miss trading the zig zags inside the longer term up or down move.
Just one tiny issue - I haven't modified P&F theory at all in the charting I described, that was 'plain vanilla'... some of Pfscan's stuff DOES alter the accepted ideas (never been keen on stone tablets with zilch to confirm authenticity) but the HL charting as described is the official 'right' way to do it.
Dave
 
Originally posted by nautical
Hope you don't mind me chiming in here guys but I wanted to clarify one of Jonny's replies.

Fastnet wrote:

From your email I understand that if, for example, the price is currently describing an upward column but then, intraday drops once or twice and bounces off the same resistance that is more that 3 boxes away (assuming 3 box rev) BUT the price gave a high for the day slightly higher than the previous day then only the continued movement upwards would be added??

Jonny replied:
Correct, that is a rule of PnF charts

Jonny - in the example above (using Fastnet's proposed EOD system) are you saying that you'd actually plot the day's high and not the closing price (which itself may not actually extend the plot upward)? I always thought that it's the level of the closing price (not the intraday high) that determines whether you continue to plot upwards or instead start to plot a column in the opposite direction. Is this not the case? Happy to be corrected of course and have much enjoyed reading your automated PnF exploits on one of the other TTW boards...good luck with it!

Thanks in advance.
Chris
 
JonnyT said:
Proper PnF uses the days high and low prices and not the close price.



Not true.......Indexia Technical Analysis package and instruction offer two methods based on High/Low and Close prices only.

As with most things in TA there are two schools of thought when it comes to PF charting. Indexia provides a unique ability to draw PF charts based on either of the two methods......Indexia concede that "Many Analysts" prefer Hi/lo P&F....But to try both and draw own conclusions.........

I use Indexia 4x1 P&F using a free slightly delayed feed from NTV which still produces good signals.
 
dolton said:
JonnyT said:
Hi Chris,

Proper PnF uses the days high and low prices and not the close price.


Not true.......Indexia Technical Analysis package and instruction offer two methods based on High/Low and Close prices only.

As with most things in TA there are two schools of thought when it comes to PF charting. Indexia provides a unique ability to draw PF charts based on either of the two methods......Indexia concede that "Many Analysts" prefer Hi/lo P&F....But to try both and draw own conclusions.........

I use Indexia 4x1 P&F using a free slightly delayed feed from NTV which still produces good signals.


Correction...{typing again} .should read 4x2 P&F.....patterns with single box reversals are not allowed!!
 
Dolton,
hi - I think I have to support Jonny T here a bit - whilst Indexia allows close based charting, although not uniquely (despite the claims in their promo material other programs - including Pfscan <g> - had some of their 'unique' features before they did) ... the 'how to do it' literature going back the past 60 years or so is fairly consistent on the HL method, although many of us accept that close charts are both possible and useable. (My own opinion is you just get a different view of reality, which will vary from 'better than' to 'worse than' the H/L variant, just as changing box size or reversal is often as much personal preference as anything else). Consequently perhaps the more appropriate word would be 'traditional' rather than proper?
1 box reversals might not be supported in Updata TA, but they are 'allowed' in P&F, they are the only version of the charts that allow both X and O boxes in the same column in fact.
Dave
 
Thanks for the support Dave.

I have seen evidence that high and low prices for PnF was used by the Chinese in the 1800s for trading spices etc.

Personally I see no advantage to using close prices.

JonnyT
 
The first charts I ever saw were Close... it was only when I got into P&F properly that I discovered H/L! I like the thinking behind H/L, which I can make a logical case for - I find it harder to justify close... I think yu could perhaps make a case for starting a new column if (despite going a box further) there was a huge opposite move AND the price closed in the reverse direction, but ultimately the 'intraday' problem will either resolve itself or reverse next day... which is, to my mind, why we use P&F - it takes the wobbly bits out and when the smoke clears you're off again.
 
Top