Research project - Technical Analysis

[crazybrab]

Scenario is as follows: Postgrad at ICMA Centre, have to complete a 20 credit point research project/aprox 8000 words on something finance-related. I've always been a chartist at heart for the past 8 years, so I hoping to munch through charts/technicals galore.
The effectiveness of technical trading rules and chart patterns has been done to death by many, so that is a no-go area.

Here are a couple of options that I am considering, and any feedback or other interesting ideas would be much appreciated: (not actual questions, just a bunch of ideas lumped together)

1.Is the taking-out of techncial-based stops, reversal just after signal, distortion of trading signals etc. random or occuring frequently to be attributed to chance? Has probably increased with growth of electronic communcation networks?

2.Might have a dig through literature on Market profile. Evidence of value areas in the stock market.

3. Anyone have anyhting they have always wanted to answer and never had the time??

Thanks,

Martin Brabenc

 

[dcraig1]

One thing that I do think would be very interesting (and quite possibly very useful) would be to examine market microstructure (If I'm not abusing the term) at S/R levels. ie the behavior of the order book and T&S. S/R levels could be pivots, MP levels, gap fill points, previous high/lows, the edges of volatility bands - anything you like really.

It would be a non-trivial exercise and you would need to crunch quite a lot of data. You might be able to get the data for free (including historical book data) from http://www.opentick.com

 

[ZDO]

How about exploring / testing the actual feasibility of trading perfect rings in FX? PM me if interested and need further info...

 

[blackcab]

The effectiveness of technical trading rules and chart patterns has been done to death by many, so that is a no-go area.

Has it been done well though? Every paper on TA performance I've seen takes an idea like a pattern or indicator, exhaustively tests permutations, but has some glaringly dumb component, often in the performance measurement, like just taking a snapshot of price after 10 days to measure efficacy and leaving it there. You can even see it in some of the charts the authors exhibit - some great exit signals completely missed, and the performance ruined because of some stupid arbitrary exit criterion. As an idea, how about fixing that and doing it well? (If it is a widespread problem that is - it could just be the couple of dozen papers I've looked at).

Here are a couple of options that I am considering, and any feedback or other interesting ideas would be much appreciated: (not actual questions, just a bunch of ideas lumped together)

1.Is the taking-out of techncial-based stops, reversal just after signal, distortion of trading signals etc. random or occuring frequently to be attributed to chance? Has probably increased with growth of electronic communcation networks?

Seems difficult - you'd need an accurate model of stop placement, signal recognition etc. that matched reality, and you probably couldn't verify its accuracy.

2.Might have a dig through literature on Market profile. Evidence of value areas in the stock market.

I'd like to see something on MP. And like dcraig said, what about an analysis of internals? Could be good.

 

[crazybrab]

All replies have certainly got my thoughts going again.

Looking at the 'order book' is a mammoth task given I would need to take some mammoth sample.

Blackcab,
Thanks, I'm going to have a dig through academic literature and see if this is the case. This topic was my initial choice and I may well return to it! (assuming I can find a better way of doing things!)

ZDO,
I've emailed you..

I've also got access to the Eurekahedge database, but looking at hedge funds means I would be up against perhaps 80% of other students on the course. I have no comparative advantage in this area, so it is probably best avoided!