NG collaboration development

Discussion in 'Journals' started by 40yotrader, Jul 11, 2006.

  1. m4a1


  2. bolter



    I became a 46yotrader yesterday, and I have the hangover to prove it.

    Probably not. Unless you're doing meaningful business with these guys they won't do you any favours. It will be interesting to see how the introduction of an electronic market impacts liquidty, volatility and slippage.

    You mentioned seasonal tendencies for NG. The following chart may be of interest.

    #22     Jul 13, 2006
  3. m4a1


    well, for what it's worth, here are a few things i found that may be of interest.

    it looks like tuesdays and thursdays have significance. thursdays are best. tuesdays are worst.

    here's a simple system using thursdays as the focal point.
    #23     Jul 14, 2006
  4. m4a1


    here's the same system using tuesdays.
    #24     Jul 14, 2006
  5. m4a1


    it also looks like short trades work better. here's a short only system that only trades during the June-Aug and Dec-Feb. the profit factor is well above 2.

    the same system on the long side only produces a PF of 1.04.
    #25     Jul 14, 2006
  6. bolter


    Very interesting. I take it these are end-of-day systems? In the interests of context can you disclose them?
    #26     Jul 14, 2006
  7. m4a1


    here's the short only model. not a big difference from 40's. this particular model may be overly optimized, but i tried a bunch of different things yesterday and the short trades seemed to work better. someone else testing a whole other set of methods may find the opposite.

    for the day of week models i looked at the website i posted and noticed that data is released on thursdays. assuming that the data is important (and i don't know much about natural gas data), i figured structural shifts in trends or fundamental changes in people's views are likely to occur on that day. so i made thursdays the focal point to test for certain things. then i tried every other day of the week and found tuesdays to be consistently worse. i don't really know why tuesdays are worse, but if i am correct in assuming that thursday's data is important, then maybe by tuesday the market is starting to just meander along waiting for thursday's data. kind of like the hours before the fed meeting kind of thing.

    my data seems to have holes. i could not replicate 40's ts report exactly, so the above may all be hocus pocus.

    what does an "end of day" system mean?

    #27     Jul 15, 2006
  8. hi bolter ... do you mind me asking for how many yrs you have traded NG ? don't tell me since the early 90's ?

    are there ways to reduce the risk in trading this product?

    I do not like to see massive dips in my P+L

    within the space of 30 minutes

    #28     Jul 15, 2006
  9. bolter


    Nice work - you've identified what appear to be some important characteristics of the NG market. If you were considering incorporating them as filters or rules in a system, I would take the next step and validate your findings with empirical evidence.

    End-of-day system refers to your data interval. But clearly you are using intra-day data.

    #29     Jul 16, 2006
  10. bolter


    The NG market has been around since early 1990, but it was really thin for quite a while. I've been trading it for about 5 or 6 years.

    One way to smooth out your equity curve with a volatile market like NG is thru position sizing. I use a measure of volatility as the denominator in my bet size calculation. When volatility spikes my bet size decreases and vice versa.

    I wouldn't use exchange margin as a true measure of relative volatility. Exchanges can play games with their margin calculation.

    The other suggestion I would make is to avoid trading the front month. This is the most volatile contract.

    Good luck,
    #30     Jul 16, 2006