Saturday, January 28, 2012

Introducing ACH

I've decided to continue writing the blog. I have some new plans, I have refined some tools and also got new ones. I hope to make it interesting and useful for the readers.


A new tool that I want to present is the Analysis of Competing Hypotheses (ACH). This is a framework in which decisions can be made by comparing different hypotheses (scenarios) with the help of evidence/observations at hand. It is a method inspired from the book Psychology of Intelligence Analysis, written by Richards J. Heuer, a CIA analyst. The book can be downloaded for free at the CIA website.

The method relies on identifying possible scenarios from a certain point and then trying to invalidate each scenario using the available evidence. The scenario that is the least negated is also the most probable. I adapted and simplified a lot the way it is done in the book as it will be seen. I have used it for some years but never methodically. Recently, I noticed that, if used methodically, the ACH greatly improves the process of decision making by focusing attention on multiple pieces of continually changing evidence, by eliminating the "confirmation bias" and by leaving a trail that can be reviewed later.

Here is an example. (This would make this post quite lengthy but it is just for the first time as I introduce the method.)


ACH on Jan28


Will the market correct more from these levels, maybe 40-50 points, or will it start rising after the ~ 20 points correction on the daily chart without going lower than ~ 1307 ES? (the analysis is done in the context of a developing bull market - this is in itself a subject of analysis but the evidence is overwhelming against the bear market so I am taking it as given).



Hypotheses/scenarios:


H1 - the rally continues from here, the correction on the ES daily chart will not be greater than the standard  ~ 20 points;
H2 - the correction will be greater than 20 points, probably reaching the 20 DMA.


Evidence/observations:


E1 - CPC 5d ema pulled back enough to support a rally;
E2 - NYHL finally spiked confirming the SPX rallly;
E3 - the market sold off 20 points after only a 27 points rise on the daily chart;
E4 - Thursday's session was all weak except for the last 30 minutes;
E5 - ST TRIN (3 day moving average of TRIN) is oversold;
E6 - the GDP report showed improving residential and durable goods consumption contributions;
E7 - there was weakness in the GDP report at the Final Sales to Domestic Purchasers chapter;
E8 - Friday's session was not weak with the European and US lunch entries strong, and with buyers coming in after the US lunch;
E9 - the market did not correct more than 20 points after the January break-out and now it is not against any big resistance - the next resistance on the weekly chart is around 1340. 




Matrix: (the way to read and fill in the matrix is to ask the next question for each piece of evidence: does this evidence confirm (+), invalidate (-1) or partially invalidate (-0.5) the corresponding hypothesis?)



E1 E2 E3 E4 E5 E6 E7 E8 E9 Total
H1 + + -0.5 + + + + + + -0.5
H2-0.5 + + + -1 + + -0.5 -1 -3



The winner is H1, since it is the least invalidated. This means that I will base my trading on this scenario next week.The conclusion is not automatic and before deciding on the scenario to bet upon, the critical evidence should be checked for accuracy. I will not complicate things further though.

Also, as part of the conclusion two questions can be asked:

(1) If H1 is happening what should we/should we not see happening from here? Sometimes, the fact that "the dog didn't bark" represents very strong evidence.
(2) What kind of event would invalidate the choice made? In this example I would say that this event can be a decisive move below ES 1307.

Updated here.

3 comments:

Anonymous said...

Love it! Great work!

Tony

Adrn said...

Thanks Tony! I am glad to see you are still around.

Anonymous said...

I read through your whole blog many times and found great help for my trading. Thank you for your providing such wonderful tools!

Tony