Trading Systems in 'New Concepts In Technical Trading Systems' by J. Welles Wilder

[kaalilaatikko]

Hello!

The time for getting into using SIS is approaching for me little by little. I reread the last 15 pages and collected the recent posts about the peculiarities or open questions of SIS. I think the earlier ones were pretty well summed by this comment by Dale:

After spending the entire week (as you all well know) going over the SIS etc. etc. etc. I've reached ONE FIRM conclusion:

It does not matter WHICH system you're trading from 'the book': you NEVER place an order or base ANY trading decision until after the close of the daily bar!!!

This makes a lot of sense and makes the system more consistent as well. Personally I'm going to study the system by using this rule.

This one I think has not been commented yet:

Originally Posted by dpaterso

This is where the (my) confusion is STILL ever present:

Take a look at Figs. 8.6, 8.7, 8.8, and 8.9 (pages 97 - 98).

The way I look at those diagrams we're being told that your entry orders are AT THE EXACT POINT where the ASI PASSED the previous (SIGNIFICANT???) HSP or LSP. Right??? In other words according to these diagrams the ONLY WAY you get 'get in' at the entry points indicated would be if you had already placed your orders the day after the previous (SIGNIFICANT???) HSP or LSP had formed. In other words: NOWHERE on these two pages does it say anything about waiting for the CLOSE of the ASI above or below the previous (let's forget about SIGNIFICANT??? here i.e. I'm tired of typing it but you know what I mean) HSP or LSP. Nor does it say anything about 'for the first time the ASI exceeded (passed) the previous HSP or LSP'.

This is 'significant' (DAMN!!!) BECAUSE:

Now turn to the work sheet example on page 103.

'Day 10' SPECIFALLY (another DAMN!!!) says: '... On Day 10, the ASI exceeded 104 for the first time. We therefore entered an order on Day 11 ...'.

Now my point is this:

Take a look at the associated chart.

Yes I agree that on 'Day 10' the ASI DID INDEED exceed 104 for the first time and I can see where the order was placed on 'Day 11'. NO PROBLEM SO FAR EXCEPT FOR THIS: WHAT IF the price had gone WAY PAST the high on 'Day 11'??? At that point the ASI WOULD STILL have 'exceeded 104 for the first time'. However: there is no way that if that had happened you would have been able to 'get in' at '46.05' UNLESS you had ALREADY PLACED YOUR ORDER on 'Day 7' i.e. the day after the HSP was formed!!! Do you see the difference??? Do you see the 'issue' here??? Which is it??? Do you place your order the moment a previous HSP or LSP has been (completely???) formed and is clearly visible OR do you place your order THE MOMENT the ASI exceeds (passes???) the previous HSP or LSP OR do you wait for the ASI to CLOSE above or below the previous HSP or LSP and ONLY THEN place your order???

Take a look see!!! I'm DYING to get and answer on this because I believe this is where the 'issues' with this system lie.

I think that 'Day 7' is not the correct time to place the order. It might be possible that the order would be hit by a peak, while the ASI would not necessarily ever cross contrary the previous significant HSP or LSP.

The second option "do you place your order THE MOMENT the ASI exceeds (passes???) the previous HSP or LSP" is contradictory with the conclusion not to place an order until after the close of the daily bar, as far as I have understood all this. So I would ignore it as well.

So, if the price is then "way past the high on 'Day 11'", you are then left with 3 alternatives:

a) Do nothing, just ignore the trade and consider it as a lost one.

b) Place the order at the original entry point and hope that the price will dip enough to pick the trade. If it refuses to dip, see (a).

c) Place a market order at whatever the price is after the ASI close.

My choice would be (b), as I think it is best in line with the overall concept in these systems. Market orders have been earlier classified as inferior ones in this thread, and therefore I would ignore (c).


I hope I have now caught up with the posts about SIS and can start introducing it into my toolkit.

J.

[randont]

I've had an epiphany! Ok, so how many of our RTS trades hit their B1/S1 etc and go instantly green? NONE, maybe you've had another experience, but I've never had my entry hit and then an instant rebound to TP! How many hit their entry and then go on to the next level and then some, most if not all. Another thing. How often does it rebound and hit the original TP at B1/S1? Not often. AND once it does hit your TP at B1/S1 it then reverses quite often.
To sum up 95%+ of our RTS trades go red quite awhile sometimes weeks before they turn for TP, even those that turn for profit instantly sometimes come back to entry anyways. So... I've been thinking why not trade it as a breakout? At (asuming they're at the outside LRC's)B2 you sell and at S2 you buy, ride it for a nice profit to then next level or so and then TP and reverse, AND on the reverse you ride it for some more profit? The market just seems too wild atm for a more normal aproach. Also well its what Wilder has been telling us only he says to enter at B1/S1 and then reverse at HBOP/LBOP but that was in alot "tamer" market. Let me know what you think. I'm going to give this method a try this week I'll keep you updated. Another thing that made me think of this is that I demo traded and entered 15+ rts trades all but 2 were red as of today, half entered last week the other half this week. I'm sure they'd turn green eventually but that takes time and time is money my friends. Another thing if it does instantly reverse why the hell couldn't we ride it out just like we do now? Asumming ADX is low it should reverse anyways with time.

[kaalilaatikko]

I implemented the Wilder smoothing for ADX so that it shows on the Delta platform. A couple of screenshots are attached as starters. I hope this will spawn some discussion about the good old subject.
J.

[dpaterso]

Good (Saturday) afternoon everyone.

kaalilaatikko:

Well: all I can say is WOW!!!

It's quite obvious that there are MAJOR differences between the 'smoothed' and 'unsmoothed' ADX/ADXR to the point where it's obvious that using the 'unsmoothed' ADX/ADXR is quite possibly signalling the incorrect trading system to be using at a given time i.e. in many cases the 'unsmoothed' may be signalling the use of (for example) the RTS whereas the 'smoothed' ADX/ADXR is signalling the use of one of the trend following systems (and maybe so far THIS is why Randon's RTS trades are not turning to profit as speedily as he would like). Of course it's also very obvious that the 'smoothed' ADX/ADXR cuts down on the number of 'whipsaws' given by the DMS.

Nice work!!!

There is just one thing that is concerning me though (and I'm not sure if it will affect the 'bigger picture' of things):

kaalilaatikko:

You'll notice that my 'unsmoothed' ADX/ADXR starts on day 14 right from the beginning of the chart. Yours does as well BUT you'll notice that the ADX/ADXR are not being calculated correctly for (I would imagine) at least the first 42 bars. Now I know WHY this is happening BUT are you sure that this is not affecting the indicator from 'way back then' BECAUSE remember now that (because you're 'smoothing') every last bit of data is being taken into account (unlike the 'unsmoothed' version where anything prior to a 14 day period is being ignored). Actually: OBVIOUSLY it's affecting the current data BUT just how much of a difference this makes to the current data IF IT'S A FULL CHART i.e. 999 days I would imagine is negligable BUT if it's a relatively newly added instrument where the chart is NOT a FULL CHART then the difference will be pronounced (and as a matter of fact the monthly charts for many of the instruments are not FULL CHARTS as yet e.g. AUD/NZD).

I'm also assuming that somehow you've managed to test this and compare it to what the 'smoothed' ADX/ADXR is supposed to be in 'the book' (unlike me who ASSUMED that the ADX/ADXR that I was using was correct)!!!

[kaalilaatikko]

Originally Posted by dpaterso

There is just one thing that is concerning me though (and I'm not sure if it will affect the 'bigger picture' of things):

kaalilaatikko:

You'll notice that my 'unsmoothed' ADX/ADXR starts on day 14 right from the beginning of the chart. Yours does as well BUT you'll notice that the ADX/ADXR are not being calculated correctly for (I would imagine) at least the first 42 bars. Now I know WHY this is happening BUT are you sure that this is not affecting the indicator from 'way back then' BECAUSE remember now that (because you're 'smoothing') every last bit of data is being taken into account (unlike the 'unsmoothed' version where anything prior to a 14 day period is being ignored). Actually: OBVIOUSLY it's affecting the current data BUT just how much of a difference this makes to the current data IF IT'S A FULL CHART i.e. 999 days I would imagine is negligable BUT if it's a relatively newly added instrument where the chart is NOT a FULL CHART then the difference will be pronounced (and as a matter of fact the monthly charts for many of the instruments are not FULL CHARTS as yet e.g. AUD/NZD).

When we dissected ADX last time, there was quite extensive cross-checking of the implementations by different people and re-reading of the calculations by more than one person. As a result, I became very assured that my calculations are done in accordance with Wilder.

What I did this time was to port the same calculations to use Delta's data structures. Then I checked the values of a few sample +DI/-DI crossing points against my text-form analysis dump, which used my original implementation, and all of the values matched at least to the 2nd decimal (I did not dump more in the text-form analysis). So I was satisfied with this, and forgot to look at the beginning of the chart.

But indeed, there was a bug in this new implementation (but not in the benchmark one). I forgot to divide the sum of the first 14 DX values by 14, and this caused distortion to the beginning. I added the missing division, and now the calculations look as they should right from the beginning.

After the correction, my checkpoints still matched completely with the reference data. So a full chart was able to totally smooth out the initialisation bug.

Originally Posted by dpaterso

I'm also assuming that somehow you've managed to test this and compare it to what the 'smoothed' ADX/ADXR is supposed to be in 'the book' (unlike me who ASSUMED that the ADX/ADXR that I was using was correct)!!!

I hope this got answered by the above.

I will mail you the corrected version in a minute.

J.

[dpaterso]

Good (Sunday) morning all!!!

kaalilaatikko:

Again: well done and thank you for your hard work. And yes: you've answered all my questions beyond doubt!!!

HOWEVER folks:

Now that we're forced to revisit ADX/ADXR I've done a comparison between my 'unsmoothed' ADX/ADXR and kaalilaatikko's 'smoothed' ADX/ADXR and come up with what I perceive to be some very interesting results. (Before I continue let me say the I believe that the 'smoothed' ADX/ADXR is 'the way to go').I've attached a chart (that you have to download to view i.e. you'd not be able to see my markups clearly if I simply uploaded the chart).

What you'll find is this:

First instance:

Using the 'smoothed' ADX (which started here at 14.92) I'd have been looking to use the RTS for trades and needless to say I'd have lost money BIG TIME i.e. although the 'smoothed' ADX was telling me that the market was ranging it's quite obvious that is was not and had indeed broken out into a trend (down). The 'usmoothed' ADX IN THIS INSTANCE was giving the correct signal / value for a trend following system.

Second instance:

Using the 'smoothed' ADX (which started here at 31.78 BUT moved down to 17.38) at some point I'd have started using the RTS. Using the 'unsmoothed' ADX at no point would I have even considered using the RTS and would have gotten 'eaten alive' again. The 'smoothed' ADX IN THIS INSTANCE was 'closer to the mark'.

Third instance:

Using the 'unsmoothed' ADX (which started here at 32.07) I'd have been looking to use one of the trend following systems which, from what I can see, would have been the correct choice of systems to be used here. Using the 'smoothed' ADX (which started here at 17.38 but did eventually get to a value of 32.57) I'd have been looking to use the RTS and, again of course, this would have resuled in losses.

My conclusions are these:

First:

Don't get too happy in the thinking that the the 'smoothing' is going to 'miraculously transform' the profitability of the RTS. I don't see it happening (even although as I have said I believe that the 'smoothed' ADX of kaalilaatikko's is 'the way to go').

Second:

I now find myself thinking ONCE AGAIN that maybe we have to start 'playing around' with the period again e.g. using a 7 day period as opposed to the 14 day default period. The reason I say this is that just judging from my above observations it's really does appear (once again) that ADX is very slow to react in fast moving markets.

Third (and last):

What's interesting to note is that the VS would have got you into a short trade pretty early on in the downtrend (even before my 'first instance' as indicated) and you'd STILL BE IN THE SHORT TRADE right up until this past Friday with not a single stop and reverse!!!

Just to reiterate:

I'm by no means questioning kaalilaatikko's work here at all and I believe it to be correct. I AM INDEED now questioning the overall reliability of the ADX itself as it pertains to our markets and the use of the RTS. One 'curve ball' I'd like to 'throw into the mix here': as you are all fully aware I have the AO and AC on my charts all the time and I've noticed that when an instrument or pair is range bound the actual height (values) of the oscillators are reduced whereas when an instrument or pair is trending the actual height (values) of the oscillators increases. Could THIS be an 'ADX Backup' or 'ADX Confirmation'??? I'm not sure so take a look and see what you think.

[kaalilaatikko]

I moved to looking at CSI before jumping to SIS. I implemented the following based on the downloaded data on top of what I already have in my program (this applies for Delta only; I used 10000 as lot size):

- Margin requirement = lot size / leverage; this is for the quote currency and needs to be converted to the standard currency used.

- Value of 1 pip move = lot size / pip factor; again for the quote currency (pip factor = 100 for jpy and 10000 for the others) and needs to be converted.

- Value of commission = value of 1 pip move * spread.

The result is finally multiplied with the pip factor, which gets around the related differences in ATR. The multiplication with 100 in the formula is still there on top of that.

I retrieved the values for all available pairs and applied Wilder's formula on p. 111 as such (except that I used euro as the reference currency for all values). The results are stored in the attached file csi_rank.txt. What strikes on my face is the dominance of ZAR pairs in the beginning of the list. I think that the constant 150 is causing this. Wilder does not explain at all what this constant means or how he has defined it. I suppose that it somehow balances the values he has for the commodities. But apparently it does not work properly for forex.

So I experimented a bit with that constant and tried to put it in some proportion to the commission. Wilder's constant was roughly 3 times the commissions he was having. The commission grows with the lot size, and looking at the values that the calculation yielded for the pairs, I tried to use the value (lot size)/2000. The resulting rank is stored in the second attachment.

What do you think, would either of these ranks be usable for selecting good pairs?

J.

- Edit: I removed the attachments after finding out an "issue". See my later post where they are updated.

[dpaterso]

Good (early for me Tuesday) morning all!!!

kaalilaatikko:

Nice work although you MAY want to check this out:

I TOO had an 'issue' with the CSI and the ???/ZAR pairs UNTIL I 'modified my thinking' a little bit.

What I believe you need to do with forex pairs is this:

You need to 'level the playing field' for the ATR FIRST and only THEN perform the rest of the calculation. After doing that you'll find that the ???/ZAR pairs are NOT ALWAYS 'top of the pops'.

In other words:

I did like this:

For any pair quoted with four decimals you multiply the ATR by 10 000 and for any pair quoted with two decimals you multiply the ATR by 100 and so on and so forth (this has NOTHING to do with a 'PIPFACTOR' or 'Quote Price Factor' by the way). Ignore all decimal places after doing this. The idea is to get the ATR's to 'jive' (Craig???) with each other. Once this has been done you then perform the rest of the CSI calculation as per 'the book' and you'll see the difference. The 'skewing' of the CSI calculation is because of the ATR and has nothing to do with the spread or anything else. As a matter of fact: it is BECAUSE of the spread and therefore the amount payable in commission that GBP/ZAR should NOT always be 'top of the pops'. Try it out and let me (us) know what you think. It works for me. The only reason I don't use it on a daily basis is because it's a REAL pain to have to update the margin requirement every day (remember that the margin requirement in USD changes on a daily, actually on a 'per second' basis, because of currency rate fluctuations at Delta. At a broker like GCI where the margin requirement is 'fixed' and cannot vary this is of course not so much of a problem but then of course the value per pip movement also changes on a daily or 'per second' basis. How MUCH of a problem this is with forex pairs I'm not sure i.e. these fluctuations may be negligable at best given the number of decimals places being used and there may be a sort of 'one size fits all golden figure' that can be used).

Stock note (I'm going to start including 'Stock notes' now and then if it's OK with everyone because as many of you are aware I'm shifting my focus to stocks or at very least 'diversifying my portfolio' as it were):

A nice little 'tidbit' for this morning:

Bloomberg.com: Worldwide:

Now what's interesting to note here is the fact that even although the fnancial stocks are taking another 'pounding' there are some large names that are INCREASING their stakes in some of these companies and NOT DECREASING their stakes!!!

What's more:

ONCE AGAIN money mangement is 'key'!!! This may NOT be the 'bottom' for financial stocks BUT then again it COULD be!!! The point is that most of the financial stocks have lost most of their value to date since last year so with sound money management it's possible to 'buy and hold' (many of the financial stocks are actually 'undervalued' i.e. the share price does not reflect 'fair value' and for the most part these stocks have been sold off because of 'sentiment' which translates to 'fear' 'in my book')!!!

Also:

Did you know??? If you're long a stock and the company declares a dividend you 'receive' the dividend (at Delta anyway).

Back to forex:

Why should any of the above be of interest to you as a forex trader??? Well for ONE thing: if the Dow Jones Industrial Average is 'brought to its knees' then so is the USD.

[kaalilaatikko]

Originally Posted by dpaterso

I TOO had an 'issue' with the CSI and the ???/ZAR pairs UNTIL I 'modified my thinking' a little bit.

What I believe you need to do with forex pairs is this:

You need to 'level the playing field' for the ATR FIRST and only THEN perform the rest of the calculation. After doing that you'll find that the ???/ZAR pairs are NOT ALWAYS 'top of the pops'.

In other words:

I did like this:

For any pair quoted with four decimals you multiply the ATR by 10 000 and for any pair quoted with two decimals you multiply the ATR by 100 and so on and so forth (this has NOTHING to do with a 'PIPFACTOR' or 'Quote Price Factor' by the way). Ignore all decimal places after doing this. The idea is to get the ATR's to 'jive' (Craig???) with each other. Once this has been done you then perform the rest of the CSI calculation as per 'the book' and you'll see the difference. The 'skewing' of the CSI calculation is because of the ATR and has nothing to do with the spread or anything else.

Maybe I did not write it clearly enough in my post, but the end result was exactly as you described. I may have used the term "pip factor" incorrectly here, but anyway I used exactly the same multiplication factor as you mentioned. I just mentioned doing this multiplication last and not first, but it's the same arithmetics anyway.

But I had another look at my tables and started wondering why the ADX/ADXR values looked so odd, and here I had an issue of my own. The values were calculated correctly, but my parametrisation was not! There was mixed use smoothed and unsmoothed values as well as differing period lengths because I happened to use an input file which I had been using for backtesting and forgot about these parameters. Latest, corrected tables are attached.

Originally Posted by dpaterso

As a matter of fact: it is BECAUSE of the spread and therefore the amount payable in commission that GBP/ZAR should NOT always be 'top of the pops'. Try it out and let me (us) know what you think.

Here I must disagree with you. I think that it is because of the huge ATR that these pairs are risen on the top, rather than lowered as much down because of the big commission. The big divisor offset 150 causes that even rather big variations in the commission do not affect that much in the overall result. To prove this I prepared a small Excel worksheet that is attached.

Wilder's example commissions were in the area 45..85, and apparently that is much more optimal for his formula. But the values I got were in the area 2...40, weight on the lower end, and I am afraid that this distorts the results somehow from the desired one. And that's why I think that to get this working better for our purposes, the offset needs to be adjusted in an appropriate way. Unfortunately Wilder does not give any hints how he selected his offset.

Originally Posted by dpaterso

The only reason I don't use it on a daily basis is because it's a REAL pain to have to update the margin requirement every day (remember that the margin requirement in USD changes on a daily, actually on a 'per second' basis, because of currency rate fluctuations at Delta.

This is a great example where I do believe in the power of automation. Once implemented it is a matter of seconds to get the values. There is no magic in calculating Delta's margin requirements for each pair, it can be done with simple arithmetics once you have the daily prices at hand. And I don't think that you must stick to converting all values to USD, I'd rather do that in the base currency of your account, which happens to be EUR in my case.

J.

[dpaterso]

Hello again,

kaalilaatikko:

I'm not going to post any comments YET although I have read your post and need to 'digest' it.

I just wanted to post something 'quick' here:

It would seem to me that using kaalilaatikko's now 'fixed' and 'smoothed' ADX/ADXR the only SURE FIRE WAY of KNOWING WITHOUT A DOUBT that an instrument or pair is trading in a range for the RTS is WHEN ADX IS BELOW BOTH +DI AND -DI (AND < 20) as per 'the old man'!!! Take a look see!!!