Author Topic: Trading logic  (Read 6610 times)

Alexander Horn

  • Newbie
  • *
  • Posts: 2
  • Karma: +0/-0
Trading logic
« on: September 26, 2014, 08:07:37 pm »
Just questions out of curiosity from own coding, not suggesting these for implementation.

 - Have you ever tried defining different entry/exit levels, e.g. enter at 1 /2 / 3 SD whenever first downtick appears, or even staggered allocation at all entries as money management? So making sure you get regular returns from lower levels, but also catch the juicy higher ones?

 - Have you tried limiting pairs from trading when one of the stock is before/after earning reports / dividends or trading above certain MA volume? To get out of “bumpy zones”?

admin

  • Administrator
  • Sr. Member
  • *****
  • Posts: 338
  • Karma: +13/-0
    • Pair Trading Lab
Re: Trading logic
« Reply #1 on: September 27, 2014, 09:53:49 am »
Hello Alex!

Have you ever tried defining different entry/exit levels, e.g. enter at 1 /2 / 3 SD whenever first downtick appears, or even staggered allocation at all entries as money management? So making sure you get regular returns from lower levels, but also catch the juicy higher ones?

I was thinking about that usually the other way around (I trade mostly uptick signals) - so about order layering/ position scaling. But the problem is the money management. You need either to expose too much of the capital to one pair or crumble the slot to layers. So I never tried to implement this because of these problems, but maybe I will look to this again in the future.

Have you tried limiting pairs from trading when one of the stock is before/after earning reports / dividends or trading above certain MA volume? To get out of “bumpy zones”?

I did not but some of my friends tried to avoid trading in earnings times. The problem is that you filter out too much. Most of the time this filter reduced both losses and profits.
Personally I used to prefer another approach - trade only at the end of the day. This means that sometimes these "black swan events" can go in your direction as well and you can exit at the other side of the z-score with huge profit. In the long term, profits and losses from these events should cancel out.

What is critical to watch is whether one of these events also broke the cointegration relationship of the pair and get rid of the pair if this happens. This is very difficult :)

I don't care about dividends at all - sometimes you pay them, sometimes you gain them. Nothing to worry about.