Search results

that's not necessarily true. system's ability to over-optimize is a function of a) the number of trades and b) the number of system parameters (ie degrees of freedom). therefore, by adding parameters, it is perfectly possible to over-optimize a system having any number of historical trades. -...

sure, but that's not what i meant by "intrinsic value". i mean when you buy a stock in a company, you actually buy real value -- a portion of company's assets and future profits. hence the "intrinsic value" of a stock. whereas, buying a futures contract won't give you such "intrinsic" value for...

well, think of it this way: there are always two parties to each futures contract. one agrees to deliver the goods on the expiration date, and the other agrees to pay for the goods on the same date. the latter is the buyer (of the contract) and the former is the seller, and the transaction value...

FWIW, with Datek you can flip the position in one order, but no, your order cannot exceed your current buying power. - jaan

i thought i'd mention the ultimate source for historical tick data (stocks): http://www.nyse.com/marketinfo/taqdatabase.html each month is about 15 CD-s of data, so it's HUGE. the disadvantage is that there's only a monthly subscription and it takes several weeks for the previous month's...

here's another idea: why not make the number of displayed threads equal to the number of new messages on the forum, with some min/max limits applied? for example, for each forum display max(2,min(N+1,10)) messages, where N is the number of new messages in that forum. on a related note, it...

i really wonder how can they (the bullet firms) do that? the only way i can imagine to avoid paying the spreads (both on the put and long stock) would be to loan them out of the inventory. is that what the bullet firms are doing? also, if the long component of bullets is loaned stock, then i...

actually, there is a way around this (this requires additional development, though). basically, you add an 'attach' button next to the browse form that would attach the current file to the message being composed, thus freeing the form for next file. that's how webmail programs do it. let me know...

this has been covered ad nauseum, but at least admit that it's risk adjusted P&L that matters. otherwise i could go out, buy $10k worth out of the money options, and have a fair shot at being your semi-god for january 2002. :) - jaan

hitman, please ignore FIREHAWK's post -- that was a textbook example of a flamebait. - jaan

precisely! that's the very thing that i keep telling people. while i would never argue which approach - discretionary or mechanical - is "better", i would like fellow traders to recognize that they are very different. that's why discussions about expectancy, statistical relevances, and...

hehe, freaky, isn't it? basically, there are numerous experiments that demonstrate this. a typical experiment involves firing an electron from point A, measuring it at point C, and forcing it to make a "decision" at point B inbetween (for example, at point B there are two slits, so the electron...

i see, LOL. nono, the QM most definitely does not challenge the scientific method! i believe you are misinterpreting the "act of observation" in quantum physics. to put it simply, quantum systems are so delicate that it is not possible to observe them without profoundly disturbing them. you...

dottom, just wanted to use a little bandwidth to thank you for your effort of putting together what in my view was the most interesting and inspiring post of the week. - jaan

well, i claim that they are mostly random, and the trick is to take advantage of the small non-random component that's there due to the inefficiencies dottom covered in his post. this has been widely accepted among the academics, too. heh, don't get me started on this -- i'm a theoretical...

thanks, based on the above i know i need it, at least in the short term :) peace & good luck to you too. - jaan

hitman & magna, i think you are slipping into patronizing here. i for one find this thread both interesting and relevant. sure, i have an academic degree and do automated trading as opposed to being with you in the "trenches" (ie so what do i know), but i believe more tolerance towards...

voodoo, i don't disagree with much of what you say, but i think we're getting entangled in details. so let me distill my point: trader's performance is a function of: 1. luck 2. market conditions 3. skill therefore, to really find out the amount of skill involved, we need to...

umm... i can see that i'm having difficulties making myself understood. so let me rephrase: it is perfectly possible to specify a system (complete with detailed entry/exit criteria, involving both shorts and longs, etc.) that will a) pass your skill test, b) be really trivial, so no skill is...

still, i don't think this requirement will save your test. in my "raging bull" example above, it is really easy to define a corresponding system: buy a basket of nasd100 stocks at the open, sell N days later, repeat. profitable, yes. your test passes, yes. skill involved, no. - jaan