Technical Analysis Point & Figure Gone High Tech

There are a large number of traders who look at Point & Figure charts, with "look" being the operative word. That is, for most traders, once the market data gets represented as a P&F chart, they are at the end game. If the last x entered is in one box higher than x's in other columns, you've got a "buy signal".

Why is that so? Because someone eons ago said so. What's the proof? There is none; that's just the way it is. Now, don't go thinking that I am going to trash P&F. Au contraire. I'm going to show you why it's a great tool, how to test its efficacy and use it more effectively.

Before I really get started, let me explain that not everyone considers P&F charts as something only to look at. One extremely respected investment advisor counts the P&F buy and sell signals of each one of the NYSE stocks and creates an index out of those signals. And he has built a very good record managing hundreds of millions of dollars with that as his market index. Another very successful fund manager says that his confessions to using P&F usually get comments like, "How quaint", as though he was left over from the Paleolithic. But when he tells them he uses P&F-smoothed data as a neural net input, they're quiet with incredulity. So P&F data does not have to be the end game. It's just that most traders are not aware of the benefits of using P&F. And as we all know, the ignorance of something in the trading arena can cost you money.

Why P&F is a Great Tool

As a trader, you have developed a set of tools that work for you. And let's say that they work most of the time - except when the market goes flat and boring. How do you ignore flat and boring? Point & Figure! P&F is always active when the market is active, and does not register when the market is quiet. That is, only market movement beyond a certain noise level counts, making the filter non-linear. Additionally, many P&F chartists change their box heights depending on the price level, making it adaptive. Would you have ever expected Charles Dow to invent an adaptive non-linear smoother?

It gets better: P&F is asymmetric. As you are rising, you keep making x's each time you achieve a new box level. But you don't start a down column (making o's) until the price goes down by your pre-determined reversal level (traditionally 3 boxes). That is, at a given point in an up move you only need 1 up-box to make another x, but 3 down-boxes to make an o. The combined effect of the noise filtration and the asymmetry is that if the market moves in trends, a P&F trading tool should make you money. That helps to explain its popularity and longevity.

Because P&F eliminates time periods, time is no longer relevant. That's why the bottom scale (the x-axis) is frequently left blank or numbered with ordinals (1, 2, 3...). Time compression is not a problem; many traders look at weekly data in addition to their daily charts. P&F just compresses the data according to activity instead of the calendar.

How can you put P&F data to use?

First, let's look at what P&F data really is. Start out with your normal P&F chart as presented in Figure 1.

image001-1.gif

caption: Figure 1

image002-1.gif

caption: Figure 2

Then look at Figure 2. The price activity in Figure 2 is identical to that in Figure 1, except that instead of keeping our x's and o's in columns, we have placed each new figure in a subsequent column. Next, move on to Figure 3, where we have replaced the x's and o's with a line.

image003-2.gif

caption: Figure 3

Now that you have freed your data from its graph, it becomes a time series just like any other. You can now smooth it with a moving average, calculate relative strength, cyclic analysis, whatever. We have even created neural net forecasts based entirely on P&F data, and more interestingly on P&F data commingled with bar chart data. That is, you have removed the mysticism from P&F data, and see it for what it really is: a high-tech smoothing tool (adaptive, non-linear and asymmetric).

PF Line vs. PF Time-Extended

There are two ways to work with the reconfigured data: time-compressed, or time-extended. With the time-compressed version you are just using the data as illustrated above. That is, your data is a point & figure line, where each new point on the line represents a movement of the P&F data, rather than a new date or time period. Figure 4 shows a standard P&F chart of the Dow Jones Industrials.

PF_Figure4.jpg

caption: Figure 4.The Classic Point & Figure Chart. DJIA

Figure 5 shows the same market index plotted as a line, with a 10-period exponential average applied to the data, and with a 25 period moving slope plotted as an oscillator. Notice the dated horizontal scale: some months have a lot of activity and some not. That's the time compression.

PF_Figure5.jpg

caption: Figure 5. Reconfiguring the P&F data makes it usable

Suppose that you're an eclectic sort of trader who mixes and matches his signals to "confirm" his opinion. Some people do that by looking at daily and weekly charts, but you want to mix P&F signals with bar chart data. The two have different time scales, so you have to extend your point & figure lines horizontally across a bar chart until you get another P&F price change. Figure 6 shows a bar chart for the DJIA over the same period with the P&F time extended line superimposed.

PF_Figure6.jpg

caption: Figure 6. P&F Time-Extended

The chart is interesting because it shows you the smoothing effect of P&F, but more importantly, you now have the capability for testing various P&F parameters. For example, the "classic" P&F format such as the chart illustrated uses price highs and lows, and 3-box reversals. But some traders construct their P&F charts from closing data, and others use 2-box reversals, to name a few preferences. In total there are four parameters: box height, box reversal amount, data input (e.g. close vs. high/low), and scale (i.e. linear, log or adaptive). With the data in this digital format you can run trading simulations to see which works best.

The Proof is Always in the Trading

Let me give you an example. I started out with the S&P 500 index (SPX) from 1996 through the present; 9 years and 8 months of data. I smoothed the closing price with a 25-day exponential and then initiated a trade every time the close made an excursion through the 25 exponential smoother. The beginning price of SPX was about 600, and over the time period the system made 60 points in profit. Nothing to get excited about; I just needed a base for comparison.

Next I took my original SPX data, and created two P&F lines. They were identical in all parameters except one: The first was based on the "classic method" of using the high and low, whereas the second was based on using the closing prices only to create the P&F data. I then applied the same 25-day exponential smoother to each P&F line. Lastly I made two trade simulations of the SPX, this time using the P&F data to generate the signals: one using the high/low method, and the other using the close-only method. Here's the quick summary:

PF_Results_Table.jpg


Wow! Using classic high/low P&F as a filter increased gross profits from 60 to 178. And switching the P&F price input to close-only ratcheted gross profits up to 550. That's a whopping difference, and the summary statistics at the end of this article are worth your review. Now, note that the best system shown here is still not worth trading. It made almost 100% in the 9-plus years, about an 8% compounded rate per annum, but the maximum drawdown that you had to endure was almost equal to the eventual profit.

My purpose here is not to give you a winning system, but to show you that Point & Figure can be a very effective tool, and not just something to stare at. Also, you can see that it makes a big difference which P&F parameters you use. Once you open your eyes to see what P&F really is (an adaptive, non-linear, asymmetric smoother), you can pick the best parameters and start generating profits. If you are going to use P&F, either for charting or digital data smoothing, wouldn't you prefer using the best parameters?

Summaries of Trading Results

PF_traderesults_base.jpg

caption: Trading results - 25-day exponential

PF_traderesults_hilo.jpg

caption: Trading results - 25-day exponential of "Classic" P&F based on Highs, Lows

PF_traderesults_close.jpg

caption: Trading results - 25-day exponential of P&F based on Close-only
 
Last edited by a moderator:
Point & Figure charting is not just some quaint old notion according to William Rafter's latest article.
 
Is there any software which will allow me to use IB data to plot P & F charts?Paid or free?Thanks
 
Thanks for the plug joules! :)

As for the article - quite good! I didn't quite understand what he did when he said that he smoothed the pnf chart using a 25 period average. I can guess that he simply used a 25 period average to smooth the data that was plotted, but is that what he meant?
 
Author’s elaboration:

In the case of a Point & Figure line, each new point on the line represents movement, either a new x or a new o. Dates are irrelevant and ignored. This is classic P&F turned into a line. The chart mentioned by “Effkay” illustrated a 10-period exponential “average” and a 25-period moving slope (see my previous T2W article). That is, the data input was from the P&F line. Once you have turned the P&F price data into a line, you can do anything you want with it, like use it to produce indicators or a trade simulation.

In the case of Point & Figure Time Extended, each point represents a new day (or week, etc.) and the last x or o for that day. If there were no new boxes recorded for that day, then the line would be continued horizontally. The chief reason for doing P&F Time Extended is to enable you to compare the P&F signals with a bar chart. Then you could compare the trading records of a “signal” generated by a bar chart with the “signal” generated by the P&F data. The bar chart signal would usually move first, but the real comparison would be the profitability of each.

Too many traders use untested ideas, or ideas which they have only tested once. BTW the software I use for this is Financial Data Calculator.
 
I love p&f and it's interesting to read that the original one box reversal is covered in the du Plessis book (which I haven't read) as there is very little written elsewhere about it.

I'm assuming JoulesMMI that you are Jeremy du Plessis's literary agent! (smile)
 
Not specifically on this article but a comment on the book by Jeremy Du Plessis
Description of The Definitive Guide to Point and Figure
I have read most of the book.. I say most because I got very bored with it in the end because of its very repetituve nature which became annoying.
Yes it is a good book for point and figure and I agree with that.
..but..
it could be condensed into far smaller a size without losing any of the information.
It reminds me of trying to use big writing to fill the pages in a school essay, so it impresses the teacher
 
Excellent new view of one of the oldest technial tools. Would also love to see some software to produce such charts.
 
Would also love to see some software to produce such charts.[/QUOTE]

Although J de P never mentions it in the book he is a director/shareholder of Updata.

Updata, not unexpectedly, has a geat point and figure facility. What is especially clever is the way you can (a) seach for favoured p&f patterns and then (b) whilst paging through the shortlist the software will automatically draw trend lines and do the upward/downward target counts.

With regard to the book, yes, it is long.

However, some of the later chapters, such as the use of p&f on OBV, are worth a visit - as is the use of Bullish Percent and Market Breadth. I know that keeping an eye on the latter is going to improve my trading.

Tony
London UK
 
Currently your choice for P&F software that can pattern find and scan for interesting charts is probably limited to Updata TA or Bulls Eye Broker (archer analysis, an Australian outfit).
BeB allows data from Metastock, TC2000, Amibroker or Ascii text according to the website - Updata used to be tied to their own data, I don't know if that's still so but it might be relevant to a decision.

I've not read J Du P's book which I'm sure is very thorough, Carroll D Aby's 'Point and Figure Charting - the complete guide' is also a very good book on the subject, probably the best I've read so far. Those interested in the 1 pt charts looking to see early work on this might like to order up 'The Point and Figure method of Anticipating Stock Pricew Movements' by Victor De Villiers and Owen Taylor, from 1934 - both books were available recently (and still are, I think) from Global Investor... the people the T2W booksite link takes you to.
 
Yup,
Updata have a trial as well - I'd strongly suggest those looking for good P&F try both out to see which suits them best. The chap running Archer is a nice chap, and provided you have a decent datafeed already then it's economical to just add the P&F software to it. If you don't have data, and perhaps would like other TA facilities along with good P&F, then Updata is pretty good.... on top of which I'd suggest that 'how it feels in use' will probably determine the ultimate choice between them.

It's a bit surprising to me that so few 'big' programs out there do a decent job of P&F - they incorporate it as a chart, but either ignore it when providing scanners or implement it so poorly (I'm thinking of Metastock as I type) that it's horrible to use. It's relatively straightforward to code up a scanner for the main patterns - around two dozen or so - yet none of them seem to bother.

Dave
 
p/f charts are now available on quotetracker.it includes historical and intraday tick charts.covering all us stocks,indices..uk stocks,spot and fx futures and index futures.cost approx a tenner a month to include 10 days intraday
free eod p/f charts for uk and us stocks,futures and indices and spot fx from prorealtime.com
 
osho67 said:
Is there any software which will allow me to use IB data to plot P & F charts?Paid or free?Thanks
quoteracker can use data from ib.they have p/f charting now
 
I'm monitoring this thread as I am a long standing user of Updata.

Nothing more to add other than that anyone with any queries about it should feel free to ask them here.

Tony
London UK
 
Free EOD data

DaveJB said:
Yup,
Updata have a trial as well - I'd strongly suggest those looking for good P&F try both out to see which suits them best. The chap running Archer is a nice chap, and provided you have a decent datafeed already then it's economical to just add the P&F software to it. If you don't have data, and perhaps would like other TA facilities along with good P&F, then Updata is pretty good.... on top of which I'd suggest that 'how it feels in use' will probably determine the ultimate choice between them.

It's a bit surprising to me that so few 'big' programs out there do a decent job of P&F - they incorporate it as a chart, but either ignore it when providing scanners or implement it so poorly (I'm thinking of Metastock as I type) that it's horrible to use. It's relatively straightforward to code up a scanner for the main patterns - around two dozen or so - yet none of them seem to bother.

Dave

If you want free eod data for your p&f charts go to www.thegrafster.com This site has a free data downloader which uses Yahoo historical eod data for up to 100 symbols over a date range of 5 days to 10 years
 
Hi,
personally I'd avoid Yahoo data... I had to program around it's dodgy figures a while back - you'd find the 'High' wasn't the highest figure in the OHLC array, or the Low was actually above the close, or something equally stupid, in a surprising number of datafiles. Also in disucssions of index prices a few months ago one chap was quoting figures nobody else saw and it turned out these were Yahoo numbers - can't remember the exact details, but it was something like the daily high and low were made up by adding all the component stock highs and lows, rather than keeping a running total through the day. Obviously not all stocks hit their high or low at the same time - if you are trading the index then it's the tick by tick total that counts, not some shortcut to make the sums easier.

The shareware 'free' data downloaders all, as far as I am aware, grab Yahoo data and suffer the same data problems as a result (although some will connect to a broker feed if you have one) - with a UK EoD datafeed running at £15 or less, perhaps £25 for the US markets, and realtime data kicking in for a few tenners a month it's a false economy to base trading signals on prices that aren't accurate. You get what you pay for <g>
 
I am very keen on p/f charting and have been using it exclusively and successfully for quite a while. I use updata which is by far the best for these charts. Its pattern searching and scanning lets you set your own formulas to describe any pattern you want. That is besides log scaling, automatic trend lines, automatic targets. Try it and see for yourself. As J deP says, you have to beware that so many programs draw the charts incorrectly, specially 1point reversals. Updata seem to be the world experts.
 
Top