Frosty Strategy Development

Discussion in 'Strategy Building' started by frostengine, Aug 3, 2007.

  1. bigbubba,

    Yes the hard stop and hard profit target seems logical as well. Actually I almost wonder if that is a better route. Simply because if I find an entry that has a edge using a hard profit target and hard stop, then it simplifies the next stage of development.

    For example, using the timed stop its you have no idea if the trade drops like a rock imediately after entering everytime and just comes back up above your purchase price often.... It could be good at picking a turning point, but who would want to sit through that big drop...

    If using a hard stop and hard profit target then you know that the position is not dropping below that hard stop. you then know that the position is more likely to hit the PT than the stop, which validates its a good entry. This also makes working on your exit after easier,because you have a potential hard stop you can work with, and then work on ways to change the hard PT to extract more points...

    I am going to try doing some tests with hard pt and hard stop as well, to see how that looks.
     
    #31     Aug 5, 2007
  2. I chose ES because its the most liquid contract. The competion in the ES is very fierce, which is also something I like.

    I would have chosen the ER2, but with the move to ICE, that contract may change substantially.

    I find YM to be too think,and I just don't like the way the NQ trades.
     
    #32     Aug 5, 2007
  3. Frosty,
    Here's another thought, keeping in mind I am not an ATS trader and have no experience programming ATS systems.

    The closest a discretionary trader can get to ATS is DOM-based trade entry/exit... trades can be set up in advance. Now then, there was a recent set of posts in the ES journal by Spectre2007 I think, not a revelation of any kind for me, just fresh in mind. It also reiterates what I already know regarding use of of DOM-based order entry.

    This example is crude and full of holes to be sure, the concept is what Im trying to illustrate. This can be easily setup IN ADVANCE in a DOM-based order entry platforms, and assumes a Long dominant intraday direction....

    L 1 ES @1440
    L 1 ES @1442
    L 1 ES @1444

    S 1 ES @1439
    S 1 ES @1441
    S 1 ES @1443
    S 1 ES @1445

    So my thought is to think about your strategies based on DOM-based order entry. Ease of programming when you land.

    No matter what, you have to decide whether your bot is designed to make money or maximize profit. The difference is subtle, but crucial.

    Osorico :)
     
    #33     Aug 5, 2007
  4. Great.

    You're picking markets to trade ATS based on how you "like" and "think"(opinion) of something.

    What you mention has no viable basis towards your trading decision. Liquidity? You should start worrying about that once you are doing enough size.

    Liquidity does not directly give you profits. It's the market tendency (condition... causality... edge... character...), the developer's ability to identify and expose them, and risk management that makes money in the ATS business. Liquidity is just one of the risks involved when trading size.

    If what you "like" is very important part of your decision process, I would suggest that you trade discretion. ATS may not be a good fit for your thought process.

    Another hint: If you going to limit yourself to trading a specific contract, I would start eliminating thoughts of how you prefer your models to be like.

    If the contract is not a trendy market, you shouldn't be forcing yourself to a trending strategy. If you want lower DD because you can't take it, don't force yourself to trade trends...

    I've mentioned this before... you can't force a strategy to work on a market that doesn't support it. You can only make money from what the market offers you. YOU have to get passed your own preferences and trade based on the market's preference.
     
    #34     Aug 5, 2007
  5. Running tests are not difficult. You build up a few codes and run it.

    I don't see anything difficult about that...
     
    #35     Aug 5, 2007
  6. If you've got SE background you should know that planning and designing is most important.

    What does a SE need to plan or design a project?
     
    #36     Aug 5, 2007
  7. Regarding ATS, I don't think planning and designing from a software standpoint is important, other than planning and designing for modification and enhancement without need to refactor. The trigger of an ATS may change, but the execution is constant regardless of how the trigger is determined. X=1=Buy, X=0=Sell.

    IMO, ATS planning and design is not the same as developing a fully featured, ever enhanced application or game, dependent on specific OS abilities as well as consumer trends.

    Osorico
     
    #37     Aug 5, 2007
  8. maxpi

    maxpi

    Planning and designing?? Get NinjaTrader, use the wizard... it's incredibly flexible after that, you can edit the C# code, it's compiled, you can't make anything that is significantly faster. The trading rules are 99% of the game, the software is almost nothing once you find a package that does what you need. I would get some trading rules that work when I trade them manually first, maybe augment that with chart indicators. There are always things you never thought of, just always. After that I would dig in and do the serious software work for the ATS, and that would start with the wizard......
     
    #38     Aug 5, 2007
  9. Before you write a program or code up a test routine, you need to have a concrete objective of what the code does or what you're trying to test.

    This is acrary's thread:
    http://www.elitetrader.com/vb/showthread.php?s=&threadid=33654

    acrary's post:
    http://www.elitetrader.com/vb/searc...=1895394&sortby=lastpost&sortorder=descending

    I don't expect anyone to accept what I say. Maybe people need to hear it from someone else. But I'll still bite:


    At this point, frosty is trying to trade what he likes, the way he likes it. He's running tests which make little sense to me or himself. Frosty needs to start off by asking himself with something like:

    "OK, I'm going to develop a trend-based model. To do so, I will have to trade a market that trends. How would one test if a market is trendy?"

    or....

    "OK, I'm going to trade the ES. I need to run tests to find out what type of models are more likely to succeed. I'm going to start running multiple tests to find the tendency of ES. What tests should I run??? ... ..."

    The above is just an example. It's not exactly where I would start but I think it's good enough for frosty's learning curve...

    Like I've been saying, coding up and testing the actual system (entry/exit, pos. sizing... etc. etc.) comes in a lot later.
     
    #39     Aug 6, 2007
  10. MGJ

    MGJ

    All markets trend ... ... ... but they only trend some of the time.

    A different approach is to say "I'm going to develop a NOUN-based system, but only take trades when the market is obviously in a NOUN. When the market is not in a NOUN, I'm going to stay out. Perhaps for many many bars. Therefore I am going to develop a NOUN-based system and I am going to develop a market-state filter that tells me whether or not the market is in a NOUN.

    It's up to you whether you choose NOUN=TREND or NOUN=TRADING-RANGE or some other approach to trading.

    One classic approach advocated by many authors, applies when you have chosen noun=trend. In this case they suggest you only take signals to go Long when the next higher-level time frame trend is Up. If you're trading 60-minute bars, only take long entry signals when the trend on the daily chart is Up. If you're trading daily bars, only take longs when the trend on the weekly chart is up. Etc. Some even suggest you require the TWO higher timeframe charts to be trending in the same direction, before you accept an entry signal on your lower timeframe chart.

    Just because some bozo wrote it in a book (or an ET posting) doesn't make it correct, of course. But it might just possibly serve as a catalyst for your own thinking and research.
     
    #40     Aug 6, 2007