TH's data, conniptions and brain dumps

[Trembling Hand]

I made a comment on Monday about the way I expected the markets to react based on experience,

Nasty Monday,
They normally fail to follow through. I think I've said it a few 100 times mostly they are a buy but when they do keep on going they make for really bad weeks ahead.

So with another comment I made here,

TA doesn't work. Test it.

I thought I would hold myself to the same scrutiny that I was dumping on the H&S and technical analysis guys and test it.

I grabbed the index data for XJO, HSI, Nikkei, Dax, FTSE and S&P500.

I tested each index over the last 10 years. What I was looking for was the return by the close 5 days later (Being Friday if there was no hols) if Monday fell more than 1% lower than the previous Fridays close.

Then compared that to the 5 day return after any day except Monday were we get greater than 1% fall. The reason for the comparison is to see if there is any particular edge buying Monday drops compared to other days.


Results? Seems legit except for the Aussie market

 

[barney]

Appreciate the work there TH. I assume you are running the XJO data off the 4.30 close? Any chance of running the same test off the sycom close at 24.00 hrs .... be interesting if it skewed the numbers back into the same ballpark as the others?

 

[tech/a]

You must be bored.
Stuffed your shoulder again?

 

[fiftyeight]

Selection bias, testing one comment????

I have been pondering a few comments you have made over many posts but

TA doesn't work. Test it.

and

Yep the pleasures of trying to make a buck hey!! In spite of having a good start and the right plan for a gap fill I just got spanked on god knows what other than a good trap when the HSI closed the ON gap and keep going.....

Some it up.

Is TA just looking for repeating patterns and trading them accordingly? (Profitably or not)

If I called the above a "Gap Fill", have you not just used your own kind of TA setup? (Profitably or not)

Not trying to argue semantics, just wondering if I have missed something obvious?

 

[Trembling Hand]

Selection bias, testing one comment????

Hmmmmmmm you wonder why anyone would bother. I was just trying to show how to add hard numbers to trading ideas away from the feel good fuzzy BS newbs waste their time with.

Here is a question for you. What is the chance of a short working late Friday morning after a week of falls? How would you find that out? You could trade a "TA Pattern" that looks like a short setup but tell me.......

What are the stats on this TA Pattern you would use?

Ask yourself this - have you got 1 hard stat on a single pattern you think you trade? Either through any type of backtest or forward test?

 

[fiftyeight]

HAHA WOW

TH thought you of all people would of picked up on a little sarcasm...

Selection bias again destroying one trade I posted??? (Sarcasm, it was a horrible trade and I know it)

After my great trading tonight (sarcasm) I went and had a look at the weekly chart and noticed not many bars seem to finish on their lows. I was going to ask for advise on how I could look into this further and plot some distributions (not sarcasm).

This seems like it may turn into a great thread that may answer a lot of the questions I am coming up with as I begin trading more.

My original question was leggit though, I am not sure if I am talking semantics or I have missed something really obvious??

 

[burglar]

... Is TA just looking for repeating patterns and trading them accordingly? (Profitably or not) ...

Surely, if the patterns are repeating ... and one is trading them accordingly ... one must be making a profit.
Unless I am missing something?@!!

 

[Trembling Hand]

After my great trading tonight (sarcasm) I went and had a look at the weekly chart and noticed not many bars seem to finish on their lows. I was going to ask for advise on how I could look into this further and plot some distributions (not sarcasm).

I've also noticed that. And tested it here,

http://tremblinghandtrader.typepad....2010/03/where-will-the-weekly-highlow-be.html
And
http://tremblinghandtrader.typepad....high-and-low-for-the-hang-seng-index-hsi.html

Data,



The lows and highs are distributed to Mondays and Fridays which is a great bit of info to set up trading throughout the week.

This was done in 2010 and I might do it again with recent data but read the links and you get the idea.

 

[Trembling Hand]

Surely, if the patterns are repeating .


Unless I am missing something?@!!

Thats just it. Where are these patterns? Head and shoulder? Break outs/downs? Elliot wave? whateever? You will not find a nice constant repeatable pattern to shoehorn the market into. If you did you wouldn't need stoplosses, money management and have win rates up around the 90% rate.

Punters waste time looking for the correct line up of magic tea leafs but thats no what makes a trader profitable. Their decisions are far more dynamic that.

 

[lesm]

..... and one is trading them accordingly ... one must be making a profit.
Unless I am missing something?@!!

Define 'trading them accordingly'

Will a 'repeatable pattern' be profitable in all scenarios?

The post implies a level of determinism, which may not be valid, with a potential caveat that appears to become one of trading them accordingly.

It would still be possible to slowly bleed to death. Otherwise, as TH states 90% of traders would be successful.

If the repeatable patterns have a high probability of being successful regardless of the instrument they are being applied to, then why are there not more algorithmic systems in use, even at the retail trader level?

Maybe it depends on what's in the line up of magical tea leaves

 

[John Swift]

I've also noticed that. And tested it here,

http://tremblinghandtrader.typepad....2010/03/where-will-the-weekly-highlow-be.html
And
http://tremblinghandtrader.typepad....high-and-low-for-the-hang-seng-index-hsi.html

Data,

View attachment 59588

The lows and highs are distributed to Mondays and Fridays which is a great bit of info to set up trading throughout the week.

This was done in 2010 and I might do it again with recent data but read the links and you get the idea.

Hmmm. So based on those blogs...

Would a possible trade setup be to buy on a break of the Monday high with a stop at the Monday low and holding until the close on Friday?

 

[Trembling Hand]

Hmmm. So based on those blogs...

Would a possible trade setup be to buy on a break of the Monday high with a stop at the Monday low and holding until the close on Friday?

That is not what I think of when I see those stats but we all view markets differently. I run stats like this to help me understand market moves. Frame trade ideas but not necessarily trigger or produce trades.

 

[CanOz]

That is not what I think of when I see those stats but we all view markets differently. I run stats like this to help me understand market moves. Frame trade ideas but not necessarily trigger or produce trades.

Sort of stuff to use in context later?

 

[ThingyMajiggy]

Great thread TH, more more more!

Hardest part I find is getting decent data and working with Excel, spend 90% of the time rooting around trying to reverse a column or figuring out some formula or some crap

 

[Trembling Hand]

Sort of stuff to use in context later?

Yeah, apparently we are meant to trade with the larger time frame trend . Its Wednesday morning and we have gapped lower Monday, flip flop around Tuesday and now tested Mondays lows and bounced back to the highs, knowing these two data points (Monday gap downs usually fail and weeks usually trend) you would lean towards a week trend higher and be cautious shorting breaks to new highs.

 

[ThingyMajiggy]

Been working on trying to find some stats all weekend on the DAX(as well as learning Excel!) and have got this so far, nothing really jumping out at me, other than seems to close it's gap a hell of a lot of the time at some point during the day and if we get ranges towards the 100 ticks, it's likely to keep going...I think lol. My brain is starting to hurt keeping track of what I've done and all the formulas

Not sure what sort of stuff to be looking for, wouldn't mind figuring out % of time it closes the gap within the first hour etc.

Do you have certain things you look for TH or just keep fiddling away until you find an edge somewhere?



P.S - Some of this might not be hugely accurate as I might have made a few formula mistakes in there, so don't go basing any trades off my findings whatever you do, need to run over it all again to make sure it's all good. Just showing how hard I'm working on it for not much reward haha

 

[tech/a]

I would think that GAP information is pretty handy

 

[Trembling Hand]

Been working on trying to find some stats all weekend on the DAX(as well as learning Excel!)


Not sure what sort of stuff to be looking for, wouldn't mind figuring out % of time it closes the gap within the first hour etc.

Do you have certain things you look for TH or just keep fiddling away until you find an edge somewhere?

Good stuff. Once you learn how to test for 'something' more things start to become possible. I really just use it to add weight to ideas or nagging questions. I think you find the most useful info by sticking to what makes sense from your observations. Like how quick a gap does close. I might have a look at that one later today.

 

[Wysiwyg]

I would think that GAP information is pretty handy

I think the gap 99% is meant to be open > ref(close,-1) or open < ref(close,-1). There certainly isn't daily gapping on the DAX 98% of the time.

 

[tech/a]

I think the gap 99% is meant to be open > close or open < close. .

Agree.
So would be handy I'd have thought