Quant basket trader? I need your help!
Page 1 of 842 12 LastLast
Results 1 to 10 of 14

Thread: Quant basket trader? I need your help!

  1. #1
    Are there quant basket traders out there? I need a hand from those of you who use qualitative methods to size and time the opening and the closing of the basket trades --instead of I shut when I feel pleased with my profit--

    I'm not myself a basket trader. My motivation comes from the fact that I'm trying to develop a system that a number of instances, mainly at the ranging phases, is equivalent or at least very similar to basket trading.

    I tried to find whether there is an optimal method to start and shut a commerce under the premise that the instrument is mean reverting. I found a very complied method seeking to address this dual stopping problem (in the pdf). The alternative is based on the subjective cost of remaining in the market more versus the anticipated increase of the payoff. But the interest rates are low today. It can be interesting for people using choices, due to the time decay, but it does not appear to match my needs.

    Additionally, I found a very easy method that comes in keeping up a lot size linearly proportional and contrary to the space of this price to the expression measured in deviations. In brief I add on my losers at every standard dev. A grid. But I don't get the rational behind the thought. Say the price is at the mean. I'm flat. It extends 1 SD below the mean. I start a lot long. It goes to the next SD below the mean. I add another 1 lot long. If the price returns at 1 SD will I shut one lot to keep the linear exposure?

    In case the price goes into the 3rd SD I have 3 positions open and a floating reduction of (2 1 0=-RRB- 3 times the SD. I just get 3 SD if it returns right to the mean but a 6 SD floating reduction should it hit the 4th SD. At first it seems better to hold on and wait the price to come back to the mean and find a 6 SD profit. But since the price is wandering up and down before returning to the mean possibly the method is logical. If price goes -1, -2, -1, -2, -1, -2, -1, 0 the profit is 4 instead of 3. For sure I dislike the idea of adding about the losers while the price is moving against my positions. Especially because I'm aware the range won't last forever and I certainly don't wish to add against a breakout. I assessed the ranges don't continue long enough to offset this reduction.

    Because of the mean reversion assumption the probability of winning can't be utilized: per assumption/definition it is 100%! I kept thinking and discovered that it'd make sense to weight the exposure by the probability of being at a given level. I use the usual supply to keep it easy. The most probable place of the price is on the mean. But it is the goal level so no profit. The farther from the mean the more profit but also the probability of being there. It's better to be exposed when there is more potential. Let us multiply the potential profit (the space in SD) by the probability N(x), N the standard supply:


    The exposure is highest in 1 SD above or below the mean. Employing the formulation I scale my losers out. Sounds good in case of a breakout of this range. But I also scale out of the winners. And if the price ping-pongs between two amounts I take a reduction even when the price eventually reverts! Definitely not good. I feel like the exposure can't be solely dependent on the job of the price in the range.

    How do you handle it?
    https://www.forexforum.co.za/attachm...1757115554.pdf

  2. #2
    Quote Originally Posted by ;
    Simpleguy, Good idea... I will look at fashion, and also currency strength for the pair (AUD strength and USD strength for AUDUSD such as ). I'll post stats. I am adept in R and Excel... in case you've got any suggestions to test outside to gather stats, allow me to know and I will test it in R. What do you mean that a range into the data? So can help if you explain it longer, I know also. Best wishes, EZ
    I see the market in five countries: range, up, fast up, down and fast down. I am able to work out how to determine the condition of the market by looking at the charts but want to place it. So when I look at results and break it so I know what works best where. I can not figure out a fantastic formula for range to assess results.


    A negative question for the people in here. Does anyone know how to pull on renko data from charts into glow. What I am looking at is under the probability of continuation along with certain scenarios many bars we get.

  3. #3
    Quote Originally Posted by ;
    quote I see the market in 5 countries: range, upward, rapid down and rapid down. I am able to work out how to ascertain the condition of the market by taking a look at the charts but want to put it into excel. When I break it and look at outcomes so I know what works best where. I can not figure out a good formula for range to assess results. A side question for the people in here. Does anyone know how to pull on renko data from charts into glow. What I am looking at is under specific situations how many successive bars we get or the likelihood...
    Since reversal Renko pubs possess (range *two ) compared to same direction pubs (range * 1). The likelihood of an upward bar after a up bar isg, down bar after a up bar is3.

    So 2 upward bars at a row probability would be 0.67*0.67= 0.4489. 3 upward bars in a row could be 0.30 etc. I did the actual bars stats with tens of thousands of pubs time ago, and they always follow this probability rule (Random Walk). So it would be easy to simulate this in excel and get fairly accurate stats. You really wouldn't require the data.

  4. #4
    Quote Originally Posted by ;
    quote Exactly.. That would allow it to be better for the purpose that I think. I'm glad you enjoy it. . So far as making it better, unfortunately I do not have any ideas right now. New ideas might emerge after you apply this procedure, I expect!! Great luck Pip. .
    The result simply translates to follow the trend fool!! . A result which is logical since we espouse the candles to ch the swings which breaks the prior one.


    EDIT: The zones on top and at the base come from too few samples in the intense (and infrequent ) conditions.

  5. #5
    Quote Originally Posted by ;
    quote The result merely translates to follow the tendency fool!! . Sense since we labelled the candles to ch. picture EDIT: The zones at the top and in the bottom come from too few samples in the intense (and infrequent ) conditions.
    Well, what do you think? Room for improvement? Or is this the eureka moment??

  6. #6
    Quote Originally Posted by ;
    quote Well, what do you believe? Room for improvement? Or is this the eureka moment??
    I always discovered the trend was statistically quite powerful. No matter oscillator I could test/invent I always realize that the signs are only always correlated to this trend (buy signs win when the trend is up). Even when the oscillator is detrended. Even when the oscillators have no advantage their likelihood distribution is skewed to the trend direction (0 mean with skewness the indication of this trend).
    The eureka moment will probably be if some day I find an edge in the ranges.

  7. #7
    Quote Originally Posted by ;
    quote I constantly found the trend was statistically quite powerful. Whatever oscillator I really could test/invent I constantly realize that the signs are just always connected to the trend (buy signs win when the trend is upward ). Even if the oscillator is detrended. Even if the oscillators have no edge their likelihood distribution is skewed to the trend direction (0 imply with skewness the sign of the trend). The eureka moment will probably be if some day I find a border in the ranges.
    Could you distinguish or prediction ranging and trending markets using a version? And change your trading style accordingly?

  8. #8
    Quote Originally Posted by ;
    Can you differentiate or prediction trending and trending markets with a version?
    Yes and no. I can easily see a non-trending market such as a MA having a slope of a size smaller than any (adaptative) threshold. But the real definition of this range is based more on the process generating the range than. With a real definition of a range we could have a fantastic version.
    Quote Originally Posted by ;
    And alter your trading style accordingly?
    I can't alter the trading style accordingly because I really don't understand the style of trading to embrace. Buy the support/sell the resistance is flawed if the market is not mean . For today statistics proof that it is not. The price comes back to the center line but nothing forces it to do so.

  9. #9
    Quote Originally Posted by ;
    quote Yes and no. I can easily spot a non-trending market such as a MA having a slope of a magnitude smaller than a (adaptative) threshold. Nevertheless, the actual definition of the range depends more on the process generating the range than. Just with a definition of a range we can have a fantastic model. quote I can not alter the trading style accordingly since I do not know the style of trading to adopt. Buy the support/sell the resistance is faulty if the market is not mean . For today statistics evidence...
    I see. I'm looking for the square root period method, and comparing the projection into ADR(30). If the projection to the day (live updates every minute) is far lower compared to the ADR, I can conclude the day has a long way to go. If the projection surpasses the ADR with a lot, than that I could conclude the day might become ranging for the rest of the time . I am going in this direction, although I don't have any results yet. If this can give me a two state model somehow I will be glad.

  10. #10
    If you hunt for mean reversion on google you always find a reference to the Ornstein-Uhlenbeck process (or the Vasicek model in fund ). You'll also find reference to the discrete equivalent AR(1) autoregressive process. The Ornstein-Uhlenbeck process is devised as

    Where is the mean, #952# and; 963; are positive parameters. Wt is the Wiener process. I prefer writing it this way:

    because today we view dx/dt as a speed and dW/dt as noise. Wikipedia offers an image of the evolution of this process:

    Wherever the starting point the process returns to its own mean. The further away from the mean the stronger the next move is expected to be toward the mean.

    The image attached below is made with R. I utilized EUR/USD H4 close prices. The first horizontal chart indicates the first 1000 samples. I added a low pass filter to simulate some long term tendency (purple). Because this isn't about trading or calling but about numbers I will alter the filter to cancel the lag. The next level graph is the signed distance of the price from the moving average. The neighborhood signed distance to the mean: positive = previously, negative = below. As it is extremely noisy I filtered again (reddish ). The next horizontal chart is merely random white noise to be contrasted with the fourth horizontal chart that's the residuals to the red filter. Definitely not white it is at least almost gaussian as shown on the histogram along with the QQ-plot.

    When the space to the mean follows a Ornstein-Uhlenbeck process we should have, generally, a negative pace when the value is positive (and the other way around). I utilized the value of the filter and its incline to assess the position and speed. This was done in order to get rid of the noise element as a consequence, generally, it is zero. The speed should be proportional to the space with an opposite sign. The past two graphs on the right hand of the image show this isn't the case. The regression of the speed (y-axis) awarded the standing (x-axis) is not so strong and what's more when the price is away from the mean, generally, the speed has the exact same sign. The further away, the quicker off! The last chart is the acceleration (second derivative of the filter). We can see a relation. What I see is that the speed isn't proportional and opposite to the space to the mean but the speed is. We have reversion with momentum.

    On a practical point of view the presence of momentum tells us
    - if an average reversion trader's place is in drawdown that this drawdown is expected to increase. Averaging down at this time isn't a good idea.
    - the target of the mean reversion trader is no more the mean price but the time when the momentum dissipates. This really doesn't depends too much on the job of the price or below the mean.
    - for the exact same reason the entrance doesn't depend on the distance to the mean but on the change of sign of the detrended momentum.
    - but the further from the mean the more probable the price will probably move in favour of the trader since the acceleration increases with the distance.
    - regrettably if the price is quite far from the mean we can not guarantee that it is not since it is the mean that's changed and thus the tendency has turned.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  
This website uses cookies
We use cookies to store session information to facilitate remembering your login information, to allow you to save website preferences, to personalise content and ads, to provide social media features and to analyse our traffic. We also share information about your use of our site with our social media, advertising and analytics partners.