Pair Trading Strategy Journal

Discussion in 'Journals' started by jonnysharp, Aug 18, 2008.

  1. tough not to play with momentum in these momentum friendly, trending markets but if you can also take something out with these relatively stress free, market neutral strategies, tear it up, dude:D

    also, good to hear that you held onto some of it. i think to trade some and hold some for the bigger moves if the commish is not cost-prohibitive.

    wondering on a pair like MAS-WHR, how much do you (or anyone; Johnny, Total, Saico, trading2) consider the aspect of biz practice correlation or do you just go with the differential correlation and cointegration? do you think that in some cases, it may be just coincidental or do you just work it until it doesn't work anymore? if WY wasn't coming out with earnings tomorrow, what do you think of that with MAS as a better fit or even better, TWP is fundamentally better than WY, though the spread is not as extended on a weekly thus not quite ripe?

    thanks much and good luck increasing the probability of stumbling upon more nuts
     
    #1421     Jul 30, 2009
  2. ej420

    ej420

    If you are not using intraday data, coding this up is a few hundred lines of python (or another high level language) code AT MOST (getting data from yahoo.) (visualization/charts are also trivial to add.)
     
    #1422     Jul 30, 2009
  3. It maybe ok for a seasoned programer but still a couple of 10 hours.
    Just think that there are holes (missing dates in one stock and not another) in the Yahoo data and that you have to account for it otherwise all is shifted. Not that easy. And that is assuming that you know the problems in advance ;-)
    You also need an intraday update.
     
    #1423     Jul 30, 2009
  4. a couple of ideas:

    long ATI - RTI
    short BMS - SEE

    SEE got hammered on OK earnings and in line guidance yesterday and barclay's downgrade today.

    long term, ATI has been trending up against RTI.

    fundies are about neutral on both pairs, if anyone considers them.
     
    #1424     Jul 30, 2009
  5. ej420

    ej420

    Yeah, you need to be a somewhat competent programmer. You should always join timeseries using their date (or timestamp). That would add an extra 2-3 lines of code. Anyway, I'm just saying this (the pair selection part, not the live signal generation) is a really easy task for a lot of people.
     
    #1425     Jul 30, 2009
  6. academic

    academic

    I figured it wouldn't be too hard.

    Are you using intraday data? Does intraday data help if you are typically holding trades for several days?
     
    #1426     Jul 30, 2009
  7. Just got out of KGC:HMY at a little loss. Was in since july 7.

    I use EOD data + an intraday update before scanning at around 15:00. It helps if you place your trades before the close.

    Just give us word when you finish your scanner. We will see if it was that easy.
     
    #1427     Jul 30, 2009
  8. ej420

    ej420

    I was describing something I had already done. Maybe rudimentary, but very useful for me.
    If you use ptfinder, what are the features you find most useful for your trading?
     
    #1428     Jul 30, 2009
  9. I dont use PTF. I have my own code.
     
    #1429     Jul 30, 2009
  10. My bias has been opposite, short ATI, long RTI 160x200, got flat today around 5 in the spread. It does look about ready to flip the other way. I would have additional capital ready to layer this one though as it has the potential to continue moving against you. I use distribution levels of around 2 dollars.
     
    #1430     Jul 30, 2009