Regular visitors will have noticed ever increasing amounts of spam on the Trading Gurus Community Forum. On November 24th 2012 we implemented some changes:

1. New users will be moderated 2. Registered users who have a post count of zero will not be able to view attachments or download code. 3. Registered users who have a post count of zero will not be able to create a new thread

As an additional countermeasure we anticipate introducing a modest paywall in the near future.

If this is your first visit here our apologies for the inconvenience.

Existing users may wish to contribute their ideas to this thread.

Page 1 of 2 12 LastLast
Results 1 to 10 of 13

Thread: Seasonalities

  1. #1
    Junior Member Frank's Avatar
    Join Date
    Sep 2013
    Location
    Germany
    Posts
    15

    Seasonalities

    Not sure if I am right here but I didn't find a subforum for erm "general trading concepts". Feel free to move this Jim in case I missed an appropriate sub forum...

    Does anyone of you have experience with seasonal patterns i.e. repeating patterns? Larry Williams is one who wrote about them in various books and I am currently backtesting some of these.

    For example: backtests show that the 8th trading day and the last trading day of a month in the S&P 500 have a very high likelihood of being positive. So if you buy on the open (new bar around 5pm est I think) and close at the "close" you would have quite nice little system.

    However am I being "Fooled by randomness" / Data snooping? I mean... if you dig enough you will find some pattern in data that is that consistent (like if you throw a dice 10000000 times there will some impressive streaks where you will only roll 1's).
    or could this be to some real / prevalent fundamental reasons like funds investing to allocate their assets or whatever ...

    I am actually seriously considering to test for further signals and try these (with a stop of course in case of the ol' swans...).
    What do you think?

  2. #2
    Senior Member
    Join Date
    Dec 2011
    Location
    London, UK
    Posts
    139
    Hi Frank,

    Actually your observation is similar to something that I've noted. I've been playing with time of day systems. Whilst testing these live, I noticed that they tended to enter a losing streak around the 15th of the month, with more than compensatory gains towards the end of the month.

    This prompted me to look at Day of Week results. These showed patterns of behaviour specific to each day of the week. I haven't repeated this for Day of Month yet, but it's on the list.

    I wondered the same, am I just being fooled by randomness? However, I did start to consider whether real events might occur around these times... settlement dates for certain types of contract perhaps?

    Overfitting is always a danger, but I've looked at some pretty simple systems (i.e. open and close at certain times of day) and the indications are that effects are real. However, what I also noted was that when trading costs (slippage and spread) were taken into account, the profitability was not as impressive (although still there) and draw-downs were a little alarming.

    That said, I think there is mileage here. It's part of my current line of testing.

    Cheers,
    Pete

  3. #3
    Administrator Jim's Avatar
    Join Date
    Apr 2010
    Location
    South West England
    Posts
    494
    @Frank - Thanks for starting this discussion, and you were right about the absence of an appropriate sub-forum! I've created this new one which hopefully will suffice, for the moment at least.

    When I hear the word "seasonality" I immediately start thinking about commodity futures. Growing then harvesting crops, or the US summer driving season. Have you experimented with any instruments other than the S&P?

    @Dr_T - Having started thinking about futures I next start wondering about things like contract rollover, options expiry, triple witching etc. etc. Have you tried correlating your results with any of that sort of stuff?
    Reality is merely an illusion, albeit a very persistent one - Albert Einstein

  4. #4
    Junior Member Frank's Avatar
    Join Date
    Sep 2013
    Location
    Germany
    Posts
    15
    Thanks Pete and Jim,
    Yes I have experimented on a couple of other instruments and got some astonishing results. Try buying Gold Friday morning and closing on the close for example. I couldn't believe my eyes... If you then filter out months that are historically weak you get scary equity curves even with 30 USD Commission + Slippage per round trip. And it goes on an on ... You will find some of these on almost every symbol. Even some currency futures.
    However, when it get's that "good" I am always a bit sceptical....
    Like "Long the EC (6E) every Thursday but only in October and June". You get fab results but the question is: How robust is that? I mean on ten years of backtesting there are 20 trades which are hardly significant. However the Friday story with Gold for example are almost 400trades which could be some real "quirk" of this market may it be rebalancing of ETF's or institutions hedging themselves for the weekend or whatever the reason could be ...

    I quickly put together a system that tries to exploit the "quirks" that seem to be halfway significant and put them together into a portfolio of ES, CL, GC, US, EC, BP, SF

    Have a look at the results:
    https://www.dropbox.com/s/99b9gtft2i...2014.38.22.png

    Looks awfully curve fitted ... however I am putting this live on my sim account and will see what happens


    Edit: I have to mention that I have used stops and with the ES system I used Larry Williams original "exit on the first day where Close > Entryprice (first positive day)" Just in case you look at the trades in the screenshot.
    Last edited by Frank; 10-24-2013 at 12:41 PM. Reason: Clarification

  5. #5
    Senior Member
    Join Date
    Dec 2011
    Location
    London, UK
    Posts
    139
    Hi Frank,

    Thanks for posting that... interesting result. Like you say, there are some similar findings with other instruments (although my equity curves are a little more choppy!).

    Having seen that, and following on from Jim's 'buy the dip' video, I couldn't resist but post this:



    Cheers,

    Pete

  6. #6
    Administrator Jim's Avatar
    Join Date
    Apr 2010
    Location
    South West England
    Posts
    494
    Hi Frank,

    Thanks for the heads up! I have to admit I've never looked closely at day trading gold. Maybe now is the time?

    Pete's video notwithstanding, has "buy gold on a Friday" worked just as well during the recent "dip" as during the preceeding long bull run?
    Reality is merely an illusion, albeit a very persistent one - Albert Einstein

  7. #7
    Senior Member
    Join Date
    Dec 2011
    Location
    London, UK
    Posts
    139
    Although busy, I've found a little time to modify code and let a computer chug through days of the month (DOM) since 2009. On each day, it computed the slope of the regression line between each hour of the day, averaging these for each day of the month (1-31). I've attached the results. The colour maps show the mean regression slope multiplied by the number of minutes over which it was calculated... it's a kind of estimate of the expected return if you successfully traded each trend. (I've only shown days 1-30, starting in the top left and working left to right 1,2,3... then top to bottom!)

    Day of Month - Hour of Day - Expected Return.jpg

    Like my previously posted results for the Day of Week (DOW), these don't average to zero - but each shows distinct 'seasonality'!!


    Ok, it could all be random, but I'm suspecting not. What interests me is whether a trader should be more concerned with DOM or DOW? Which is most important, should a trading decision be conditioned upon DOM for example and then DOW?

    Exploiting these results would require some way of finding a good entry point. Simulations of just opening a position at a specific time of day (TOD) do show that just opening and closing at specific minutes of the day (MOD) or hours of the day (HOD) can result in a positive expectancy. However, the regression line data suggests that better entry points do exist. I wonder if Ray II could help here?

    What I'm suggesting is that the colour maps show us where breakouts are most likely to occur at specific times of day, days of the week and days of the month. They also show us the most likely direction of the breakout (or at least the dominant direction). So, armed with this information, perhaps Ray II could take his performance up another notch?

    Is it now time to develop Ray'n'DoM...?

    Have a good weekend,

    Pete

    Edit:
    I forgot to mention the following:
    Currency pair simulated is EURUSD, from 2009 to 2013 (Dec to Jan).
    x and y axes represent hours of the day (1-24). The colour scale is in pips (-50 to 50), 1 pip = 1e-4 change.
    Red represents positive change (i.e. long position would profit) and blue is negative change (short position would profit.)
    Last edited by Dr_T; 11-02-2013 at 09:43 AM.

  8. #8
    Senior Member
    Join Date
    Dec 2011
    Location
    London, UK
    Posts
    139
    ...and for completeness, I've included the plot for the 31st day of the month, I've even labelled the axes!

    EURUSD Expected Return for 31st Day of Month.png

    Pete

  9. #9
    Senior Member
    Join Date
    Dec 2011
    Location
    London, UK
    Posts
    139
    Quote Originally Posted by Jim View Post
    Having started thinking about futures I next start wondering about things like contract rollover, options expiry, triple witching etc. etc. Have you tried correlating your results with any of that sort of stuff?
    It's something that I've thought about... but to be quite honest, I didn't know where to start! I naively did a quick google for 'futures expiry dates' some time ago, hoping that something useful might turn up - but all I think that I discovered was that different futures exchanges have different expiry/rollover dates. Is this correct, have I even got the terminology correct? Having read quite a lot over at LinkedIn, I've got the impression that these things are important - I just have no idea where to get the information from and how to incorporate it into my testing. Certainly on the list... for me it represents another part of the learning curve.

    Any pointers would be most welcome!

    Cheers,
    Pete

  10. #10
    Senior Member
    Join Date
    Dec 2011
    Location
    London, UK
    Posts
    139

    Analysis paralysis perhaps?

    Ok, so now I'm confused. Take a look at the colour maps below (scale and axes etc explained above). Each set shows when each day of the month (days 1-30) falls on a particular day of the week. So, the first set shows results from Mondays on which days 1-30 of the month fall!!! Clear?

    Anyway, visually, there isn't an immediately obvious pattern. I've posted them here though, just in case somebody else spots something that I've missed. It's worth noting that these plots actually only comprise 5 or 6 actual days each over the 2009-2013 period.

    Nonetheless, I'm finding increasing support for the idea that breakout systems (i.e. Ray II) confined to specific sessions should work. That shouldn't be a surprise because Ray II does appear to heading in the right direction!

    Anyway see plots below...

    2009-2013
    EURUSD

    Mondays on which days 1-30 of any month fall...
    Monday - Day of Month - Hour of Day - Expected Return.jpg

    Tuesdays...
    Tuesday - Day of Month - Hour of Day - Expected Return.jpg

    Wednesdays...
    Wednesday - Day of Month - Hour of Day - Expected Return.jpg

    Thursdays...
    Thursday - Day of Month - Hour of Day - Expected Return.jpg

    Fridays...
    Friday - Day of Month - Hour of Day - Expected Return.jpg

    I'm clearly meant to be doing other things, so no real time to decipher what any of this means - if anything at all. Anyway, I may set my scripts to churn out other colourful images whilst I prepare various teaching material.

    Cheers,
    Pete

Page 1 of 2 12 LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •