mark brown said he found the holy grail.

Quote from MarkBrown:



the model uses data 2 to trigger the trade and then when it fails it is using the data 1 series to determine the failure - is that slick?

m
More...

Ok, and I'll restate it in my own words to be sure I'm on the right track:
- data 2 gives the long entry signal, because presumably I've found data 2 leads data 1 by some amount of time
- data 1 has the stop or "non-confirmation level" or "relationship no longer holds level" to determine the failure

Sorry if it was stated before, but which data series gives the signal to exit for a profitable trade?

thanks
 
Quote from MarkBrown:

i don't look at percent gains because i ignore drawdowns for my accounts. so it's not really relative to me but for the institutions big money loves 12% returns with 4% drawdowns. anything other than that will not attract big money.

liquidity ceilings are always a concern and probably my most time consuming focus these days. that and anonymity of the business.

no re-investing only last a short time as you probably are aware sounds like you are or worked around some professionals. your guesses are almost concerning me i have a leak somewhere. smirk.
More...

My big leak is my big mouth.:D
 
Quote from MarkBrown:

here is a way to start the framework for a dual chop to trend system.

http://www.oddballsystems.com/download_code.php?id=27&file=326shortcut.zip
More...

Hello MarkBrown, very interesting. One problem I had with my backtesting software was it would only give results for one strategy on an instrument at a time. So I built my own tool to get the results from 2 strategies on the same instrument and see how they performed together.

I've had a problem with the "switch". First switch was a "race switch" -- the dual strategy would go with the first signal it got from strategy 1 or strategy 2. Sometimes improved, sometimes not, and in the end I reckoned, how much confidence could I put in the signal that came first? Seemed kind of random.

I found a few indicator-based switches that improved the dual performance over either's individual performance. However, it occurred to me, the switch could improve the dual performance as a case of curve fitting. I've tried a few of the curve-fitting protection techniques (eg, OOS sampling, across instruments, etc), but still seems to me as "just an indicator signal", like any other indicator that gives signals, they work over backtesting but fail after a couple of weeks.

Now I'm trying to figure out if there is something else the switch should be based on. Any clues or suggestions what area I should investigate next?

thanks for your help and interest
 
Quote from zedDoubleNaught:

Ok, and I'll restate it in my own words to be sure I'm on the right track:
- data 2 gives the long entry signal, because presumably I've found data 2 leads data 1 by some amount of time
- data 1 has the stop or "non-confirmation level" or "relationship no longer holds level" to determine the failure

Sorry if it was stated before, but which data series gives the signal to exit for a profitable trade?

thanks
More...

data 2 gives a signal long or short

data 1 you got correct

if it is a data 2 trade data 2 runs the show if its data 1 trade then data one runs the show however there are 4 switches

data 2 switches

switch 1 + switch 2 = 2 then buy or sell

reset switches to =0 now

switches 1 and 2 are now looking to turn back on and may each independently do so - hold this in memory.

now data 1 switches

switch 3 + switch 4 = 2 then disengage position and reverse entire position.

reset switches to = 0 now

and so on
 
confusion made simple

there are a pair of conditions to go long and short always in the market. those conditions once met reset so they have to be met again for the next trade etc. both conditions have to agree to take a long or short, but not at the same time. one condition maybe be met an hour ago and the second condition maybe now. when they both agree they trade. one condition could be on and off 3 or 4 times and if the second condition is not on when the other condition is on it will not trade.

there are a pair of conditions to override the first pair of conditions much like parents watching their children play. they work in unison like the other pair being on and off to trigger a reversal event. also one of these conditions may pair up with one of the original conditions to perform an evasive act. this is the type of flexibility that not many simple minded trading algo's have and the reason this is a much better method to building a lasting model it's adaptive and not static.
 
I, for one, appreciate Mark's posts. One of the few on ET worth reading. I have come to realize ET is like a raucous party, sometimes a cesspool, where you sift for a nugget here and there.
 
Quote from dandxg:

I, for one, appreciate Mark's posts. One of the few on ET worth reading. I have come to realize ET is like a raucous party, sometimes a cesspool, where you sift for a nugget here and there.
More...

and so the nugget in living color, the results of the code i posted today showing how robust the model is using different settings. i will continue to build on this and show you what other varieties can accomplish.

in the end you all will be confronted with the fact it's only your fearful mind that keeps you from implementing a strategy that makes money.

http://markbrown.com/et/

again for the non programmers we are simply buying when the advancing issues are above a number else we are selling if not above that number. it's that simple, is everyone with me?

btw the red dashed lines are losers and the blue dashed lines are winners.
 
Quote from stock777:

yep , all these aliases just keep on 'acomin in to join the pump party
More...

your pathetic look how many post and how long these people have been members and see if they ever posted about me before. i don't know them anymore than i know you. but they know my work and that's something eating your pea brain up "jealous" just shut up and learn something if your capable. i won't even ask for an apology, just shut the fuck up and stay out of the way.
 
You must log in or register to reply here.
Back
Top