Van Tharping Our Way to Financial Freedom

[FetteredChinos]

Evenin' all,

been having a think today during a few quiet moments having read Van Tharp's laugh-a-minute tome.

Has anyone done any backtesting/fwd testing to try to prove that his basic Random-Entry system (coin flip with trailing stop loss of 3*ATR) works to a degree? A quick Google reveals that there appears to be a lot of contradictory information and theories about this.

the purpose of this thread is partly to establish if this method has merit (searches on here havent revealed too much), and then secondly to see if this theory is fractal in nature. ie whether it can be worked on all time frames (tx costs notwithstanding). Can anyone shed any light on this?

Im tempted to trial this on 15/30/60-min charts as a dry run for a few weeks to build up a fairly large sample size. But i would appreciate it if anyone could save me the work and tell me if it is completely fruitless?

thanks a lot, all


FC

 

[charliechan]

i think the time frame and the stop are both irrelevant assuming exit = r

his point is that random entry, when exit = r (from memory - could be wrong) will yield a win rate of 50%, BUT after costs, and filters, the win rate comes down to only 30%.


the point isnt about a profitable system based on random entry.

the point his how futile the effort is in finding the golden entry point is. therefore, we can go 1 step further and ask why bother backtesting entry points at all?

remember that van tharp isnt a trader. also, imo, he is a hell bent trend follower like its the only game in town. this is contradicts his own viewpoint that we should all keep open minds to other ideas.

for those with time to waste on reading this 'book' then you should go to www.elitetrader.com

there is a link to a free pdf version. i wont tell you where - you must do some work

 

[FetteredChinos]

thanks CC, thats kinda the viewpoint ive come to grasp from "impartial" BB posts.

might make an interesting experiment though, in the short term.

FC

 

[donaldduke]

The smaller the time frame you use the less likely it is to work, because the spread takes up
more as percentage.

3*Daily ATR for say EUR is 250 pts, and the spread is only one or two ticks.
Thats an insignificant cost.

But if you stop is less than 100pts...

Slippage is also less of a percentage when you use the 3*Day ATR.

I had my doubts about this random stuff too.

I think Tharp used $100 dollars for costs in his 1994 study, nowadways costs are likely to more like $30.

So its even more likely to work today.

There is a newletter article on this random system, number 23a, it will cost you $20 to buy from IITM.

 

[donaldduke]

Also i think if you modify the system so that the stop is allowed to get wider when the ATR expands,
then the system has a better expectancy.

 

[Trader333]

There was a lot of discussion on random entry and whether it was successful or not on a previous thread (which I cannot remember the name of). Maybe a search will yield some results on this.


Paul

 

[FetteredChinos]

thanks paul, i had a look for it on here, but couldnt find anything entirely relevant.

there are lots of coin tossing threads, which dont work in quite the same way, as they are relying on a fixed profit target and stop, and thus are likely to have a negative expectancy after costs.

i did find a wealth-lab script on this (do a search for Tom Basso) and it does appear to be profitable on certain stocks and indices (go figure?) and yet horrendous on some of the others.

im just trying to satisfy my curiosity that something crazily simple can beat transaction costs.

fc

 

[peterpr]

For what it's worth, this has been a 'pending project' of mine for ages too - although I haven't read Van Tharp.

I've long suspected that exit criteria have a bigger impact on system profitability that entry - If only because I struggle to maintain my own winning trade percentage above 50% and yet manage to remain profitable. However, I also suspect that it might be a lot less if the entry was random. Whenever I start pondering the matter, I discover that the whole issue is in fact horrendously complex , making it near impossible to decide upon a reasonable research methodology. For example, what exactly IS a random entry? you have to make your system do so many a session, or at a particular time, or whatever - and bang goes randomness.

Net result? I'd be interested to hear what you come up with because I doubt I'll ever get around to it

 

[FetteredChinos]

well as a small experiment, im toying with something similar. each day when i flick on the pc i pick a direction via an "electronic coin toss". instead of using a trailing ATR, im using the Jenkins method for stops of under the previous-but-one swing high or low.

Jenkins suggested that whilst a previous swing high/low can be taken out, the trend can still be intact. if 2 points are taken out, then the trend is likely to be over. will be interesting to see how ig goes on FX.. -30 yesterday, but currently +35 and counting today

putting the KISS back into trading...

 

[FetteredChinos]

right, as a follow up to this thread, those that are interested may want to read the last few pages of this thread, where Acrary has done some backtesting on random entry, Van Tharpy stylee..

http://www.elitetrader.com/vb/showthread.php?s=&threadid=10890&perpage=6&pagenumber=10

tested on a couple of time frames too, which shows that the idea is fractal.

time to bin those charts completely now!

fc

 

[chump]

FC,
Besides Jenkins have a try based on the Gann 3 cycle method and see if it improves your results or not..