Interesting Initial Balance Statistic

[Dogpile 10]

Hi ant, great stuff.

 

I was wondering if you have ever quantified how often the market exhibits ‘open-drive’?

 

Maybe one way to look at this would be when the market makes its high or low for the day in the opening 15 minutes…

 

I don’t mean for this to be a forum where you do the work at our request – but in effect, that is exactly what it is quickly becoming. Nonetheless, I shamelessly ask you this anyway in hope you are interested in this also…..

:)

[ant 10]

Dogpile,

 

I haven't gathered statistics for "open-drive." It would probably be easier to test for "one-timeframing." I think what you proposed, which is not a bad idea, would be a subset of an "open-test-drive." If time permits, I may pursue this. But I would definitely look for high-confidence openings after testing the previous high/low, testing bracket or balance area extremes, or an opening gap after the market has been balancing.

 

By the way, providing IB stats is something I could do easily so I usually appease people that have contributed to this forum in a constructive manner. You qualify Dogpile

[jperl 11]

There are some interesting statistics being presented here that the probability of the high/low occuring in the first hour of trading is very high. The first hour of trading represents about 15% of the trading day. So the expectation is that the high or the low will occur in the first 15% of the trading day.

 

Here is something then to think about. If you think markets have a fractal behavior, then the time span for the high or low to occur should scale on different time frames.

Example:

Suppose my trading day is only 10 minutes!! If the 15% figure for the high and low is correct, then I should expect that within my 10 minute trading day, the high or the low should occur within the first 90 seconds of my trading day.

This should be correct for any time frame that you trade in. If you trade 30 minute bars, then the high or low for that bar should occur within the first 4.5 minutes of that bar's open.

 

Check it out.

[Dogpile 10]

one other request of ant...

 

I like to think from George Taylor perspective which is most easily thought about as you form a thesis about whether you think the days basic theme will be trade from a low morning level to a high afternoon level ('low made first' was how it was written in the book -- The Taylor Trading Technique) -- or whether you think the day will trade from a high morning level to a low afternoon level (high made first).

 

I generally see this 'reversal' occur in the 10am-12pm EST timeframe --- as it did today (Friday) and yesterday (Thursday). Was curious in the stats associated with a high or low made in precisely the 10am-12pm EST? btw, This is more for 'context' that for entries and exits. attached are last 2 days:

[colion 10]

In 2000, Larry Pesavento published "Opening Price Principle" in which he claimed that 65% of the time C will be near the H or L for the day which in turn are near the O 75% of the time. Sounds like the present studies might be consistent with these results.

[ant 10]

one other request of ant...

 

I like to think from George Taylor perspective which is most easily thought about as you form a thesis about whether you think the days basic theme will be trade from a low morning level to a high afternoon level ('low made first' was how it was written in the book -- The Taylor Trading Technique) -- or whether you think the day will trade from a high morning level to a low afternoon level (high made first).

 

I generally see this 'reversal' occur in the 10am-12pm EST timeframe --- as it did today (Friday) and yesterday (Thursday). Was curious in the stats associated with a high or low made in precisely the 10am-12pm EST? btw, This is more for 'context' that for entries and exits. attached are last 2 days:

 

 

Dogpile, using the buy day/sell day rhythm from the Taylor Technique and these statistics is a good idea. Personally, I don't use the Taylor Technique anymore, instead I use my analysis based on Market Profile for that bias. If I used Taylor, I know I will get conflicting information with my other analysis, which I would defer to anyway.

 

I think I captured the data you were asking for, but I also included the first 30 mins stats as well.

 

ES (Past 3 Years)

Highs in first 30 mins (9:30-10:00 EST): 149 out of 756 days or 19.71%

Lows in first 30 mins (9:30-10:00 EST): 205 out of 756 days or 27.12%

Highs from 10-12 EST: 151 out of 756 days or 19.97%

Lows from 10-12 EST: 146 out of 756 days or 19.31%

=======

High/Low in first 30 mins: 354 out of 756 days or 46.83%

High/Low from 10-12 EST: 297 out of 756 days or 39.29%

 

YM (Past 3 Years)

Highs in first 30 mins (9:30-10:00 EST): 138 out of 755 days or 18.28%

Lows in first 30 mins (9:30-10:00 EST): 200 out of 755 days or 26.49%

Highs from 10-12 EST: 151 out of 755 days or 20.00%

Lows from 10-12 EST: 152 out of 755 days or 20.13%

=======

High/Low in first 30 mins: 338 out of 755 days or 44.77%

High/Low from 10-12 EST: 303 out of 755 days or 40.13%

 

ER2 (Past 3 Years)

Highs in first 30 mins (9:30-10:00 EST): 165 out of 756 days or 21.83%

Lows in first 30 mins (9:30-10:00 EST): 182 out of 756 days or 24.07%

Highs from 10-12 EST: 154 out of 756 days or 20.37%

Lows from 10-12 EST: 172 out of 756 days or 22.75%

=======

High/Low in first 30 mins: 347 out of 756 days or 45.90%

High/Low from 10-12 EST: 326 out of 756 days or 43.12%

[Dogpile 10]

thank you ant... awesome...

 

<<High/Low in first 30 mins: 354 out of 756 days or 46.83%

High/Low from 10-12 EST: 297 out of 756 days or 39.29%>>

 

just to be sure I am reading right, this is either/or such that:

 

1-[(1-.4683)*(1-.3929)] = 67.7% of days high/low would be before lunch (9:30-12pm)? or am I reading it wrong?

 

if right, this would imply that the market makes a higher high AND a lower low after lunch 1/3 of the time? that is surprising. guess that would include those times it run stops above/below a morning high/low by just a tick or two -- but still, seems higher than I thought even including those days....

 

btw, I use Taylor rhythm just for thinking about the days potential structure -- that is the thing about Taylor... a 'sell-short day' can quickly morph into 'sell-short day, low made first' -- which effectively equals a 'buy day' -- so he has covered all the bases with that book because you just insert the 'low made first' or 'buy made first' in after the fact....

 

therefore, I agree that you can't really rely on Taylor for much other than kind of a starting thesis with which to look for supporting/refuting evidence as the day progresses.

[Dogpile 10]

just took another look at that 8/30 day in my previous post... it never made a new low in the afternoon. therefore, statistically it would qualify as a day where low was made in opening 30-mins AND a day where high for the day was made before lunchtime. good example of a day where you could make money from both sides as there was not much Taylor bias -- 8/29 was a buy day and came into 8/30 with residual upside momentum from strong 8/29 action (can see how gap down was bought aggressively just after the open). curious what your 'post-mortem' on 8/30 is ant? how do you think about the structure for that day in context with previous action?

[ant 10]

curious what your 'post-mortem' on 8/30 is ant? how do you think about the structure for that day in context with previous action?

 

See the post here http://www.traderslaboratory.com/forums/f6/es-analysis-for-8-30-in-2371.html#post18091

[marketguy 11]

Just found this thread, Ant. Thanks so much.

 

Have you ever run stats on the Dalton's Value Area Trade (and the supposed 80 % likelihood of the market returning to the opposite value extreme)?

 

Wonder how close to truth that is? Larry Levin on Value Area Trade.doc

 

Thanks.

 

Bryan

[Minetoo 10]

Ant, good work. Your idea has been studied/proven a number of times in the past. Larry Pesavento's 'Opening Price Principle' (http://www.ensignsoftware.com/tips/tradingtips67.htm) and to some degree John Clayburg's DDF (directional day filter)(http://www.clayburg.com/ddf_home.htm) expound on the exact same phenomenon.

[idetsc 10]

All of the other day types have range extension, where one of the extremes of the IB is taken out. Since you're a betting man, always place your bet on range extension since about 95%+ of the time one of the IB extremes will be exceeded.

 

ant,

 

thank you so much for geneurosly providing these stats.

 

I am referring to your quote above. May I ask you if you could provide us with stats regarding the probability of reaching multiples of the IB? I often read that the 1.5, 2.0, 3.0 range extentions provide a kind of support/resistance. So since without any doubt the IB is very important, I thought it could also be valueable to know how significantly the IB is exceeded, respectively on which magnitude.

 

Thanks again and kindest regards,

idetsc

[ant 10]

Hi Idetsc,

 

I won't provide any stats, but will tell you that that is an area worth researching on your own. All the best with your trading!

 

Antonio

Edited May 12, 2011 by ant

[alleyb 10]

idetsc

Consider the following that range extention outside the first hour denotes a day structure that is anything but NORMAL. So if a ND is defined as first 60 minutes contains the day then consider how often that a ND actually occurs.. Looking inside my Data WareHouse I find that ND occur on 6.47%. Normal days are the exception range extentions is the Norm