trend following delusion shattered

Status
Not open for further replies.
What I see is a variety of random distributions. This is quite similar it seems to what a Monte Carlo engine would develop as it ran series after series. If on the other hand you were to let one series run indefinitely, it should cycle through a variety of distributions, eventually resulting in a "normal" or "gaussian" distribution, so I would say that the odds of obtaining a profit using a systematic approach would be near 0.

On any single run basis, the chance that a systematic approach would obtain a profit is greater than 0. I don't know what it is however. I will think about it.
 
if you change the randDirBias input to false, it goes to the moon!!!

A value of 0.65 for a trend continuation factor is more than enough. My simulations indicate discernible trends with the naked eye at values of around 0.55. They become quite obvious at 0.6. At 0.65 they kinda slap you in the face!

Of course, as Lefty correctly pointed out earlier, this is all highly idealised, and the market will do what it wants to do (see my sig below for a classic line by Tomas Bjork about this).

Cheers.
 
I understand what you are saying (I think). I would probably do much better communicating about this if I had less to drink. What I am trying to say is that if you were to let this program run one series after another, at various points you would see results that clearly are trending. In many other series however you will see what most folks would take to be chop or trendless movement. The dispersion of these results should be along a gaussian curve.

Responding to the question, I would expect to see trend (that could be identified by eye) occur more often (better than random distribution) above .55
 
Quote from Lefty62151:

I understand what you are saying (I think). I would probably do much better communicating about this if I had less to drink. What I am trying to say is that if you were to let this program run one series after another, at various points you would see results that clearly are trending. In many other series however you will see what most folks would take to be chop or trendless movement. The dispersion of these results should be along a gaussian curve.

Responding to the question, I would expect to see trend above .55
More...

LOL. I'm with you Lefty. After all this was just to see if someone could trade the price model specified by Mr Subliminal - with his parameters.

I'm about to head out for a few drinks myself - and lets not forget some cheekiness with the ladies! :cool:

Cheers.
 
Quote from Mr Subliminal:

And the second....
More...
Hi,

do these have the 0.65 trend continuation factor and constant bias as per your original spec? They don't look like they do - or are you zoomed right in? Just curious.

Cheers.
 
Well, if I had to disarm a bomb, I would have blown up by now. I will retire from the field before I embarrass myself further. A lot of fun to consider the questions however. Thanks to Mr. Subliminal for bringing it up in an interesting way, and to Mr. Atavachron for his "addendums".

Hope you are all having good weather this weekend.

Take care out there.

Lefty
 
Yes, they do have the 0.65 factor and constant bias as per original spec. It was run for 300 periods with a starting price of 1170 and imported into Excel to chart the bugger.

Now what was that about some cheekiness with the ladies?

Drinks on me!!!

Cheers, Lefty....
 
Quote from Mr Subliminal:

Yes, they do have the 0.65 factor. It was run for 300 periods with a starting price of 1170 and imported into Excel to chart the bugger.

Now what was that about some cheekiness with the ladies?

Cheers, Lefty....
More...

LOL! "The Motel" bar-club in Sth Melbourne. Although I always get the "crazy" ones. Ehh, what are you gonna do?
 
Status
Not open for further replies.
Back
Top