+ Reply to Thread
Page 2 of 3 FirstFirst 1 2 3 LastLast
Results 11 to 20 of 23

Thread: Tick Status / Interpretations

  1. #11
    Join Date
    Dec 1969
    Location
    Vienna, Virginia
    Posts
    603
    Quote Originally Posted by brrim View Post
    Can you point me to a post that indicates what explains looks like MA on the charts?
    Much of the setup is derived from Billy's published work in this forum as well as the VIT forum.

    Basically, do the following:

    1) establish a cumulative tick trace that is essentially an always-running accumulator. I use $TICK, which represents all the securities trading on the NYSE, $TIKSP, which is unique to TradeStation and represents the ticks of the S&P500, $TIKRL, which is also unique to TradeStation, and represents the ticks of the Russell 2000 small caps, and $TIKQ, which represents the ticks of the NDX-100.

    2) Apply a simple moving average (SMA) that corresponds to 1/2 day in length. Hence, if you are on 1-min bars, this would be 195m.

    3) Apply a SMA that corresponds to 1 day in length. This would be 390m

    4) Apply a SMA that corresponds to 10h in length. This is 600m.

    5) I take this further, and I also use 780 (2d), 1170 (3d), 1560 (4d), and 1950 (5d).

    I also have another display that I am working on and will share that when it is finished. Basically, it is the slope of the MAs above, and it is FAR easier to see signals using slope than looking at the moving averages. In addition, I also am plotting the slope of the slope, so we can see the ebb/flow of momentum in the tick indicators, and initial backtesting has shown this to be interesting (but unfinished) and somewhat profitable work.

    Regards,

    pgd

  2. #12
    Join Date
    Dec 1969
    Posts
    1,585
    I had experimented with a detrended presentation (4th subgraph) and like the slopes idea as well (5th, bottom).

    Name:  es 1 min liq ticks 10-6-11.PNG
Views: 2245
Size:  86.7 KB

    Here are the settings for both, respectively.

    Name:  custom3 cume tick detrended.PNG
Views: 2241
Size:  23.0 KB

    Name:  custom3 cume tick slopes.PNG
Views: 2276
Size:  27.1 KB

  3. #13
    Join Date
    Dec 1969
    Location
    Vienna, Virginia
    Posts
    603
    Quote Originally Posted by EB View Post
    I had experimented with a detrended presentation (4th subgraph) and like the slopes idea as well.
    Hehehehehe, thanks Bob. I really appreciate all your help here, as your initial shorthand is what sparked my present work (indeed, I use your shorthand on all my displays, as you can see). The slope work lends itself to strategy development very well too, and I'm finding that the underlying asset on Data1, and Data2 = cum(MA($TICK)), Data3 --> $TIKSP, Data4 --> $TIKRL, and Data5 --> $TIKQ can provide additional confirmation. They also can take you out early, so still a work in progress.

    My work as of late has concentrated on using the tick indicators to move in, and then timing the individual security with it's own behavior, e.g., much like how the GGT system is conceptualized. Contra ETFs, such as SDS, TWM, TZA, etc. are not faring all that well in backtesting, and I think this is because the primary trend has been up since March 2009. With the primary trend down now I'm getting PF ~ 1.5 or so, but the sample size is so small it's hard to draw valid (statistical) conclusions. I note that the distribution of returns is not Gaussian, so it's not a robust system as of today.

    Work in progress....

    Regards,

    pgd

  4. #14
    Paul; Thank you. Exactly what I was looking for!
    Robert

  5. #15
    Join Date
    Dec 1969
    Location
    Vienna, Virginia
    Posts
    603

    Update for Friday, October 7th

    There is an interesting divergence setting up and in looking back through a year's data, I can't find a close duplicate, so time to watch it unfold.

    Primarily, the $TICK indicator is strong:

    Name:  11OVT06-TickFinal.jpg
Views: 2102
Size:  209.1 KB

    As with all my images, right click to open in a new tab or window.

    The cumulative tick pattern on the bottom is showing the "bullish rollover" that a ribbon band gives -- the instantaneous cume tick is in white, and the moving averages are in various shades of light purple (shortest) to dark purple (longest). The green line is the 10h MA line.

    Immediately evident is that the first "breech" of the previous down leg occurred when the cume tick crossed the 1/2d MA from below around 10 a.m. on 10/5. It had not done this in several days, so this was an initial confirmation that the previous trend was in jeopardy.

    The second head's up was the crossing of the 1d MA from below which occurred around 11 a.m. on 10/5. Note that both the 1/2d and 1d slopes had started to point positive shortly thereafter, and if the previous end-of-day-reversal didn't close your short positions, some backtesting that I've done with this "failsafe" slope method suggests that these two limit the downside damage if you miss an exit. These are not good exits by themselves, but they do help.

    We all were challenged when the cume tick line reversed and headed down -- the aforementioned slopes also reversed which is why these are not a great exit indicator by themselves.

    Finally, around 14:00 on 10/5 the cume tick line crossed the 10d MA from below, tested it a few times shortly thereafter, and then it's been off to the races since.

    Of all the tests I've done all week, signals based on the cume tick line are the most robust compared to $TIKSP, $TIKRL, and $TIKQ or $TIKND. Hence, you'll see me using this more going forward (thanks Billy!).

    =====

    The aforementioned divergence is present in the $TIKSP (S&P500) and $TIKRL (Russell 2K small caps):

    Name:  11OCT06-SPXFinal.jpg
Views: 2077
Size:  216.7 KB
    Name:  11OCT06-R2KFinal.jpg
Views: 2092
Size:  221.2 KB

    The SPX is on top and the R2K is on the bottom.

    Note how the price of the SPX was steady or moving upward and it's cume $TIKSP pattern was dropping, causing the majority of the cume MAs to reverse too. Also note in the middle of the SPX pane the intra-day filter on the SPX showed clear selling pressure starting like clockwork at 14:30 EDT and price held relatively steady. I don't like steady prices on selling pressure -- this is churning -- and it shows great indecision.

    The bottom figure is the Russell 2K and while we're clearly above the 10h MA (green), we seemed to have stabilized around noon and then experienced a slow drift down in selling pressure while prices remained more-or-less constant. This reversed around 15:30 pm, but it didn't give me the greatest confidence about the internals yesterday afternoon.

    Consequently, I took 2/3 of my profits off the table at the end of the day, and while I'm glad I left 1/3 on the table, as I write this (10:15 am on 10/7), we're not moving much.

    I'm keeping my eye on the ball to see if we stall here or inch higher. There's no reason to sell to lock profits, but I've not seen any compelling reason to enter new long positions either.

    ===============

    Regards,

    pgd
    Last edited by grems8544; 10-07-2011 at 10:18 AM. Reason: typo ...

  6. #16
    Join Date
    Dec 1969
    Location
    Vienna, Virginia
    Posts
    603

    Update for Thursday, October 13th -- Crossroads

    Overall, I think the markets are out of sync -- rationale is in my post here.

    ===========

    Overall, the cumulative $TICK remains very strong:

    Name:  11OCT12-TickFinal.jpg
Views: 2066
Size:  205.3 KB

    Note that:

    1) the cumulative tick in the lower pane is constantly above all the moving averages
    2) all the moving averages have a positive slope
    3) all the moving averages are more-or-less constant-spaced, which means we do not have acceleration

    I consider this an orderly movement upwards of the broader markets.

    If we take a deeper dive into the SPX, R2K, and NDX, we see that the SPX is showing the most cumulative tick volatility, the R2K is less so but still volatile, and the NDX is impressively strong:

    Name:  11OCT12-SPXTickFinal.jpg
Views: 2064
Size:  201.0 KB
    Name:  11OCT12-R2KTickFinal.jpg
Views: 2053
Size:  190.0 KB
    Name:  11OCT12-NDXTickFinal.jpg
Views: 2046
Size:  191.1 KB

    The SPX ($TIKSP) is on top, the R2K ($TIKRL) in the middle, and the NDX ($TIKND) on the bottom.

    My view of what we're seeing here is a rotation from the SPX into the Russell and technology stocks, and this bodes well overall despite being very overbought at this level.

    In my spare time (ha!) I have been working on strategies to trade these cumulative signals, and the most stable, longer-term in-sample/out-of-sample consistency appears to be using the cumulative $TICK as the primary market trending entity and entry signal, with additions and exits from those positions determined/confirmed by behavior on the individual cumulative $TIKXX indicators. The individual cumulative $TIKXX indicators are incredibly more sensitive to news events than the $TICK, but they do fare better than simply relying on the $TICK for exits. A further path for exploration here that I'm working on is to use the $TICK for confirmation of entry and using the individual security for timing the exit. This has worked better with the leveraged instruments than the unleveraged underlying.

    More (and official results) to follow as time allows.

    With respect to the above charts, even though we're clearly due for a pullback, today (Thursday) will be the first indicator of such a pullback according to the cume tick displays. Right now they are all indicating bullishness, so the depth of the pullback, e.g., does the real-time cume tick cross certain MAs, will determine whether we should remain bullish or start transitioning to a bearish stance. In the present displays, there really is no overwhelming bearishness...

    ===============

    Regards,

    pgd

  7. #17
    Join Date
    Dec 1969
    Posts
    1,585
    Paul, without getting into anything proprietary, could you share some of your methodology for in/out of sample testing? I know you've touched on this before with respect to testing EMAs. I'm more interested in your general approach to robust system testing.

  8. #18
    Join Date
    Dec 1969
    Location
    Vienna, Virginia
    Posts
    603
    Quote Originally Posted by EB View Post
    Paul, without getting into anything proprietary, could you share some of your methodology for in/out of sample testing? I know you've touched on this before with respect to testing EMAs. I'm more interested in your general approach to robust system testing.
    Yes. I'll write something up -- and I'll try to get it done next week since I'm home and have a bit more freedom. Backtesting correctly is a big issue and I think it's important to do it correctly (or at least, my *view* of correctness <grin>).

    As an aside, I had dinner last night with several GGT/EV/HGSI followers and this topic came up there too, so it will be good to document some of that discussion with examples.

    Regards,

    pgd

  9. #19
    Join Date
    Dec 1969
    Posts
    1,585
    Great...look forward to it.

  10. #20
    Join Date
    Dec 1969
    Location
    New Jersey
    Posts
    189

    Matlab backtesting

    Quote Originally Posted by grems8544 View Post
    I too use Matlab for all my GGT work (2 production machines running 24/7, 1 devel machine), so let me know if I can be of assistance.

    Regards,

    pgd
    Hi Paul,

    Last week I sent you a PM regarding Matlab - didn't want to bore others with my entry level questions. Not sure if you are swamped or if the PM never came through? Anyway, figured I better check before I begin my backtesting.

    Thanks,
    Harry

+ Reply to Thread

Posting Permissions

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