trend following

tabris · Post by **tabris** » Wed Aug 31, 2005 8:30 pm

I agree with Moreliver and my place do put on mean reversion trades. (and yes for those who met and/or knows where I work, let the jokes begin!) While not everyone trades mean reversion or trends here, the information given initially by your question is at best laughable. If you want help, make it a serious question with approriate information.

Maltese · Post by **Maltese** » Wed Aug 31, 2005 9:34 pm

Point taken. Let me make my question more specific. I want to build a trading machine that trades and asset driven by a single BM with drift. Drift and vol are given for simplicity. What I'd like to know is this:

[list]

[*]how do I parametrize the machine
[*]has there been any work done on effect of frequency of trading and transaction costs
[*]what optimization methods work best (I was thiking of using RL methods)
[/list]

As you see this is a bit open ended. I'm just looking for any pointers in terms of what research is being done in the area.

James · Post by **James** » Wed Aug 31, 2005 9:35 pm

"my place do put on mean reversion trades."

I think most "statistical arbitrage" and "capital structure arbitrage" trades are really "mean reversion" trades in disguise.

The question is: the bet is the statistical probability that both the *volatility*, and the value, of the trade will "mean revert" within a real TVM framework. That often is not the case.

Good article here:

http://www.variancecapital.com/newsletter/july-2005.htm#1

akimon · Post by **akimon** » Wed Aug 31, 2005 10:00 pm

the turtles guide, as cheesy as the name sounds, is not a terrible read

the stuff inside are mainly risk management methods applied to trades (stop loss, size of bet, entry, exit etc). nothing magical. nothing much that one wouldn't know if spent a little time on a dealing room

FDAXHunter · Post by **FDAXHunter** » Thu Sep 01, 2005 6:21 am

[b]Maltese:[/b] [i]has there been any work done on effect of frequency of trading and transaction costs[/i]

Indeed there has been. Countless pieces. I could randomly link some papers. Try google.

Optimal Trading of Multiple Predictable Assets under Transaction Cost

Maltese · Post by **Maltese** » Tue Sep 06, 2005 6:52 am

Thanks for the paper. But it seems to me that it is focused on asset allocation. I'm rather interested in a single asset. I have tried google and citeseer -- nothing comes up or I got the wrong key-word. Could you please post a couple more links? Your help is appreciated.

Regards,

MoreLiver · Post by **MoreLiver** » Tue Sep 06, 2005 7:06 am

I really don't understand your issue here. Googling "single asset trend following strategies" gives over a million hits. If you are looking for academic papers, use Google's scholar service. if you cannot find documents with Google, what exactly are your chances of developing a trend-following system that could add any value?

cap · Post by **cap** » Wed Sep 28, 2005 12:27 am

Here are some papers on technical analysis. Perhaps you can adjust these techniques for high fequency. I think its questionable if these things can really return profits (taking into account trading costs, lower volatility, etc.)

Jagadeesh and Titman 2001

Profitability of Momentum Strategies: An Evaluation of Alternative Explanations

Mamaysky, and Wang (2000)

Foundations of Technical Analysis: Computational Algorithms, Statistical Inference, and Empirical Implementation

shavinOccam · Post by **shavinOccam** » Wed Oct 12, 2005 11:38 am

the basic of trend following is fat tails. the trader places bets in order to capture such a tail. hit ratios are usually below 50%, offset buy high payOff ratios (=avgWin/avgLoss). the turtle paper gives good introduction. the system as it is described won#t work in todays markets - one (really very small) ingredient is missing. you might want to search for "curtis faith" on this.

most trendfollowing signals are astonishing simple - yet they are not the key point about trend following. more sophisticated time series analysis will lead to similar results (IMHO), thus better shave, mr occam. money management, which is the combination of bet size, stop loss and portfolio constitution, is the key. take profits do not work - they cutt the tail too much.

RFMontraz · Post by **RFMontraz** » Thu Oct 13, 2005 12:00 pm

[i]the system as it is described won't work in todays markets - one (really very small) ingredient is missing. you might want to search for "curtis faith" on this.[/i]

Could you elaborate pls? I know Curtis Faith was one successful turtle but apart from that I find your statement pretty obscure (as to why the system doesn't work today, what is the missing ingredient, why Curtis Faith is the key to understand all the above). Thx.