Point & Figure charts

[brujam]

Re: Reply to post 'Point & Figure charts'

Fabrizio, thanks for your contribution. I understand
what you are saying and I appreciate what the Esignal
programmers want to try and do in developing new P&F
charts. I was trying to explain my idea of the correct
format for an intra-day P&F chart. My point was that
as far as these P&F charts are concerned the only
condition that dictates a new column is a price
reversal, time should be irrelevant. It confused me
that I could change the intra-day chart entirely by
clicking a 1 or 5 minute chart. I understand the
importance of time in market profile and Gann charts
but I am looking at intra-day P&F charts and time
should be irrelevant. The most recent chart I posted
on this forum gives me serious doubts about the
validity of the P&F charts as displayed.
regards
Bruce
--- [email protected] wrote:
> Hello brujam,
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>

[Dave_Pfscan]

p&F charts

Hi Brujam,
the idea that you can change to a 5 minute (etc) chart is okay, and that it'll provide a different chart is also okay - it's no different to using daily data in combination with weekly or monthly data to get different timeframes to assess the trade over. By using tick charts you ought to get the sort of chart you are expecting - personally I'd suggest a tick chart ought to start on today's first tick - by swapping to using a 10 minute bar (for example) you're aggregating all the trading over 10 minutes, taking the high and low of that period, and charting as per an End of Day datafeed version of the P&F chart....ending up with a longer term view of what the stock is doing. You might use a combination of timeframes, where tick charts show every swing whilst the chart based on 5, 10, 30 or whatever minute bar data gives you the trend over periods of fractions of an hour up to days at a time.
Time is not totally removed from P&F charts, I'm from an EoD trading background using P&F and when compiling a daily P&F chart you use daily bar data (primarily the high and the low), it's been done like that since it all started over a century ago. If anything using tick data is the new kid on the block.
As for the charts being accurate - I'd agree with you there I think, I've a suspicion that maybe there's some sort of aggregation process going on and the P&F routine isn't getting the data you'd expect it to be fed with... having some experience in developing for P&F you'd be surprised at how easy it is to find the display one box out due a simple rounding error or similar.
Dave
Disclaimer - dunno if this is required here, but other BBS's are happier if you do so - I'm a P&F software developer (small time) so have a commercial interest in this area.

[brujam]

Re: Reply to post 'Point & Figure charts'

Hi Dave,
thanks for your contribution. I can acknowledge that
there is a great deal of benefit being able to switch
from a 1min to a 5min (or any other timeframe) chart
to get a different view. I do it all the time with
candlestick charts.

My problem is that when the plot for the P&F chart is
calculated at the end of the period it either takes
price at the CLOSE of the period or some average (high
+ Low / 2), you can select what you want in the "edit
studies" menu. This misses the actual highs and lows
when they are reached DURING the period and this is a
critical error. Look at the last chart I posted on
this forum, the high and low at some time between 8am
and 1pm on that chart shows 110.82 but the actual
values according to snapshot window were 110.84 &
110.77. This is because the actual high and low was
recorded some time during one of the 5min periods and
not at the end of it (when the plot values are
recorded). Consequently the P&F plot does not show the
correct high and low and therefore the chart is
inaccurate and misleading!

If the plot at the end of the period can calculate an
average (high + low / 2) for the period why not show
the high and low on the chart? That may cause an issue
when the market gaps but I think that would be a far
smaller problem than showing P&F charts with
inaccurate highs and lows.

Another issue is where the X & O are positioned on the
chart. You can see by the dotted lines where 110.80
and 110.82 are on the chart. Can you tell me which X
or O relate to 110.81?

Esignal, are you reading this? Have you got any
comments on this issue? I wanted to pinpoint the exact
5 min period when the high of 110.84 and the low of
110.77 was made but the 29 Dec 03 seems to have
dissappeared from security BL H4-DT. What happened to
it? I realise developing the software is difficult but
it has to be right.

regards
Brujam


--- [email protected] wrote:
> Hello brujam,
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>

[Dave_Pfscan]

P&F charts

Hi,
I think we're pretty well in agreement then - I have to confess that I'm not using eSignal P&F (I'm busy writing my own with the API) so hadn't noticed they weren't using H/L on the non-tick charts. Close charts have, it must be said, also got a following so eSignal on this point are just deciding to go with CL charts - I personally use HL. (HL themselves aren't idea, in that a continuation in the existing move direction can sometimes ignore a large countermove... eg 'it added one more X box so I'll ignore the fact that the bar low dropped 20 boxes'. (Well, a bit of an exaggeration perhaps).

My own plan (so far) is to use bar data to construct HL charts off 1,3,5,10,15,30 and 60 minute data bars, and for the tick chart to use individual trade prices... I'd have thought that was more logical - use the single price when the data itself is a single trade price, and use H and L when the unit of data has H and L info in it.


That might all be amended in due course as it's early days yet and I'm still learning (I'm from an EoD background as I said, so whilst I'm not a total newbie with RT charts it hasn't been my main focus up to now).


As for the chart accuracy - I'm not arguing with you....having managed to program some absolute horrors in the past myself(upside down charts, for example - it helps to remember which way the Y coords are numbered in a Visual Basic window) it's surprisingly easy to muff something up and astonishingly difficult at times to spot quite obvious errors . The real test here, I'd say, is whether the problem is fixed quickly or not. It looks a wee bit like there's not a lot of P&F experience perhaps, and they're trying hard to supply something they're not completely familiar with? If so then I fully expect they'll get the hang of it and produce something useful. (Optimism <g>)

Thanks for the reply, HAppy New Year,
Dave

[Dave_Pfscan]

P&amp;F Charts

Hmmm,
Further to my last effort,
I was expecting tick data ("dTickTrade" in the API doc) to be single trade occurrences, but the history retrieval suggests tick data is also clumped together - looks like I'll have to experiment a bit to find out what exactly is coming in, this looks like I'll be doing HL for the tick charts too. No doubt I'll find I need a supercomputer to do the stuff I'm hoping to do just to add to the fun
Dave