extreme events in historic data

M

man

I have observed several times that statistics for the sp future differ a lot if done including the crash of 1987. if my system was long on that day with three or four times leverage, it was down by 80% at the end of the day.

now it would have been good before and good after, but how do i deal with the fact that i would have been thrown out on that day? one way is to squeeze and filter as along as this day is gone without affecting the rest too much, another would be ignore, a third to start testing after 1987.

how do you guys deal with that? assuming that there will never be a crash again?


peace
 
tough question. ignoring it could be fatal, but so could fitting your formula to that specific event - seeing as how the next time it is likely to be very different in all but it's devastating effect.

however, check your sample - I have heard that crashes like '87 invariably occure during a major bull market, which we are unlikely to see for quite some time
 
further more this biggest bull run in history did not end in an crash. but maybe the fed will let it burn out next time - one never knows. next fed chairman might not be willing ot bail out the market ...


peace
 
good point. finally i want to end up with that, though i feel that it is really a challenge to short the sp future on a purely quantitative basis using eod data. for each profitable idea on the short side i find four on the long.


peace
 
You need to add another element to your approach.


Most boutique operators were well aware of the pending difficulty you have. What is the maximum range of time you are willing to move along in your analysis approach?
 
bubba
i have data on the future since 1982. there is no longer history and the index is a non-adequate proxy. i am actually not sure that i understood you correctly.


peace
 
Quote from man:

bubba
i have data on the future since 1982. there is no longer history and the index is a non-adequate proxy. i am actually not sure that i understood you correctly.


peace
More...


All people who are in aggressive stategies for making money (boutiques trade smart money and expect ROI's several times what you read about as normal) have forewarning indicators for everything. Some of their accounts, those run by the owners of the accounts, set their own commissions and have direct access to trading. And they are informed all the time about the contemporary climate even though they are not being managed per se.

Most forewarning indicators use data from a set duration of time.
Perhaps you do too. If you do, how have you been choosing these set durations?

I cannot remember one past situation that the boutique in Irvine Cal that I used did not have forewarnings that were not correct and that there always was a chain of warnings.

As in the homeland crap, there were always levels as well. They related to leverage, margin and cash levels.


The next level up from these considerations is how to make use of indicators that cull out the player sectors. If you do not do that, you are unable to "see" your contemporaries in action. Most extreme actions are precipitated by various player sectors. Therefore it is best to be "pushed" by these guys and their actions.

You may not be using indicators of the ilk I am mentioning. A lot of ET crowd has very short fuses so they do not make money in trends. Another portion have stayed with indicator defaults designed for use before the PC. I had a post on bridging removed once because I think i critgued a guy with defaults that were bridging and his site was commercial it looked like. he was definiteltin the dark ages and the moderators allow that crap here and they delete posts that are critical of crap. lol.
 
bubba
i am extremly sorry, but i have never been to a trading boutique and have no idea what they use to forecast or indicate extreme events. thus, either elaborate or accept my humble request for pardon that i do not know what you are talking about.


peace
 
You must log in or register to reply here.
Back
Top