Mathematically Predicting the Future?

Discussion in 'Automated Trading' started by MarkBrown, Oct 27, 2007.

  1. my2cents...very interesting links you have posted ealier. I also believe that some systems leave "footprints" in the market, especially when they are traded big enough so they can be far as I can tell this is usually the beginning of the end of a system...just like systems that are scientifically published have only a limited time span. (For e.g. read the interview with Mr. Shaw in the third Market Wizard book by Schwager)
    Thats the whole point in evolving systems. I think the "smart" quant houses only play above the market radar when they
    a) have somethin better up their sleeve
    b) take advantage of others knowing what they do
    #51     Oct 30, 2007
  2. Pentaquark-

    Thank you very very much and thank you for elucidating my last post here. That was exactly the point I was trying to make when I said "that is until the sniffer programs track it down and the wonks start to program against it."

    #52     Oct 30, 2007
  3. couldn't resist posting the chart -- 'Dolly Graphics # 10' -- 8 MAs + 8 other indicators

    MetaTrader 4 has thousands of people creating indicators and Experts -- auto trading
    systems for the program all of which come onto the MT forums either free or as copies
    of original ideas -- Experts being sold

    I don't know the profitability of the many Experts but there are a few cash prize contests
    for the most profitable MT4 Experts:

    from the pov of profitability then a big yes, although I wouldn't call it 'predicting the future',
    it just isn't necessary to do that, in fact a liability — Advanced GET
    #53     Oct 30, 2007
  4. MarkBrown


    Other thought:

    Could something - seemingly not related to the market being traded - possibly provide prediction value?

    Like some forex market being traded off of cattle data? Could it be that the trading world has possibly looked over the not so obvious?

    Could it be possible that events in one market could lead to signals many days in advance of another? I have some evidence to convince me that this is at least possible to exist.

    Like some unknown never before seen variant specie of Tasmanian Tiger.

    Mark Brown
    #54     Nov 7, 2007
  5. Murray Ruggieri has done exhaustive research on intermarket analysis...much of it was published in Futures Magazine. In summary, correlations do exist, but all exhibit low coefficients (r-squared).
    #55     Nov 8, 2007
  6. gnome


    Futures Mag did a study on AI systems a few years ago. Their bottom line conclusion was that "markets have a modest degree of predictability for the next bar, only".

    So, I guess that's "predicting", but not very far into the future.
    #56     Nov 8, 2007
  7. RedDuke


    "next bar, only" is very general. If you put up yearly chart, the next bar will be the whole year. I personally trade short term intra day, and many times can guess what direction of the next bar will be when certain set ups take place. If I am wrong, I immediattly close the position and wait for next opportunity.
    #57     Nov 8, 2007
  8. gnome


    That's the conclusion of the study... if you're trading 5-min charts, next bar. If daily charts, again next bar only.
    #58     Nov 8, 2007
  9. MarkBrown


    in the old days i could predict the profitability in a position by whether or not i got a good fill. if the fill in the sp pit was quick and good price. the market was about to kill me. if the fill didn't take place then of course the trade: had i gotten it off would have made tons of money. ;)
    #59     Nov 8, 2007
  10. MarkBrown


    ok it has been said that someone possibly could predict the market one bar in advance. what about predicting the price action like 20-30 bars in advance?

    if you could do that you could make some money. but i dont even think you would have to be that accurate either to make big bucks.
    #60     Nov 12, 2007