System Building and some potent thoughts

Discussion in 'Strategy Building' started by momotrdr, Nov 18, 2003.

  1. I wanted to share some key points that hit so close to what it
    is that I've experienced in system trading this past year.

    I own maybe 4-5 books on daytrading, most in my opinion
    are complete shit and a waste of my time to even expend energy
    to pick up.

    However, I was recently given a recommendation by a friend to go take a look at Thomas Stridsman's new book. Trading systems that work. The title alone made me feel ill.

    However, my ill feelings have since faded and I wanted to share a few of these points that I feel are important in gaining perspective on going about building yourself something that works in today's markets.

    Here are those points:

    Thomas Stridsman's notes on System development, performance, and expectations.

    TOTAL NET PROFIT: of very little value when it comes to evaluating a trading strategy's estimated future performance, no matter how rigorous the testing or how robust
    the system.

    Average Trade: of the most important things to consider before starting to trade any system is the estimated average profit per trade in the future.

    The Profit Factor: build a robust trading system that is likely to continue to perform in the future, make sure that the underlying logic is sound and Simple,
    that the trading rules are as simple and as few as possible, and that the gross profit and the gross loss are evenly distributed through time
    and in relation to each other.

    Slippage and commission:
    ....First calculate the expected move you are likely to catch, then transform that move into dollar terms in today's market by multiplying by
    today's market level and point value. The deduct the proper amount for slippage and commmission. If the dollar value still looks good enough,
    you should take the trade.

    on data:
    ....non-adjusted data is best for day traders or other commodity futures traders with very short trading horizons, who more often than not close
    out all their trades at the end of the day.

    on system building:
    ....In a correctly built system, your're interested in exploiting something that should, on average, work well in several different markets and over
    several different time periods, Not with any specific type of anomaly that might or might not still exist in a particular market.

    on systems:
    ....For a system to be profitable, however, it must be traded in a market where the moves it is designed to catch also are worth taking.
    This has nothing to do with the system, but rather with the level at which the market currently is trading and the dollar value of that level.

    On perspective of market:
    ....No matter what the long-term mode the market is in, the short-term statistical characteristics are likely to still look the same and be close to
    stationary in nature. This is a very important conclusion, because if this is true, the only way to build a mechanical trading system that holds up
    and behaves the same way in the future as it does during testing-no matter what the longer-term underlying trend looks like-is to focus on the shorter
    time perspective, with trades lasting for no longer than approx. a (insert time period) and using as little historical data for the signals as possible

    on systems:
    ....Because a system is working well on all markets, it does not mean it will be tradable on all markets, because not every system will produce big enough
    dollar profits to over come the costs.

    on system building:
    ....before we can put together the system, we also need to understand what it is we are trying to achieve, and consequently, how to go about doing it.

    on bad systems:
    ....that the model is not working for these markets, but simply that these markets, at this particular point in time, are not trading at the appropriate
    price levels and with a high enough point value to generate a profit after slippage and commission. Hence, a good working model is Not the same as a
    profitable model.

    on what it all boils down to: (MY Favorite set of statements in the entire book as of yet)

    ....The really big winners are usually few and far between;the rest are just a bunch of look-alikes there to confuse you. By staying away from all of them,
    you are freeing both your mind and capital to milk that big winner several times over in A SEQUENCE OF AVERAGE TRADES, while at the same time allowing
    yourself to step in and out of other markets as well.

    ....Translated into trading terms, for a system to be both as robust and as profitable as possible, you should strive to make each trade as similar to all
    other trades as possible, EVEN if it means that you have to cut your profits short every now and then.
  2. Banjo


    Thanks for the review momo, just sliding into the systems world myself, looks like I'll get into tradestation and start the game from there.
  3. Ah at least two things I can agree with : TOTAL NET PROFIT is far from enough to judge a system, it is the details that count - but people are sometimes too lazy to look at details. Average is of course important but associated with average I think he forgot or it is not listed the variance wich defines the consistency of the average. Also something that is forgotten by many authors : the sampling aspect all the more so that the system is based on stochastic framework - and that market has some memory effect but this comes from stochastic farima model or chaos theory it doesn't matter for this subject if it is stochastic or deterministic what matters is that both recognise this memory effect which acts statistically as a variable which restrict the validity of sampling - that is to say sampling in a region where the context due to this memory effect is the same is like making a survey within a restricted region which is not representative statistically and so biased.

  4. Smarter Trading: Improving Performance in Changing Markets by Perry J Kaufman is a very good book on system trading in my opinion.
  5. Harry, I think variance is covered by the following 2 quotes :
  6. A few comments:

    1) Net profit is useless... I agree
    2) Profit factor is useless too, learn to calculcate everything in terms of percentages
    3) Most measures like profit factor, net, etc, suck because
    they are endpoints. Better to look at picture curves of these
    values over time.
    4) You MUST simulate the REAL equity curve using your actual
    money management rules, against your selected stock universe
    to know your REAL drawdowns and behaviors of your systems.

    I have found that simply looking at an equity curve, from
    a "true" simulation, with nice fat margins for commissions/slippage
    tells me almost everything I need to know.


  7. According to Toby Crabel,


    So what the inter-day time frame requires you to do is make adjustments constantly, more than any other time frame.

    You're constantly having to adjust your systems because of the changing nature of the market.


    :confused: :mad:
  8. Does this then argue in favor of systems that are not out to capture the big trend, but rather chunks along the way, which may actually be contra to the long term trend? It seems to me that most of the systems being sold are out to capture this home run type of trend. And does this also tend to turn the old saw into something like: cut your losses short and let your profits run only so far?
  9. Yes and No. Your point is valid and I struggle with this on a daily
    basis, especially given the tool set I have to work with.
    Since throughout the day you can have conflicting signals, literally 1 Tick Bar apart, the trader must be able to recognize
    the "type" of market he/she is in. Unless the moves are huge
    and the market is painting with very broad brush strokes can I expect to pull a 10+ point trade in the NQ. I see guys who
    hold all day and capture the bulk of a move on hard trend days.

    What I'm trying to do, is to capture small bite-size "Order Flow". If the move turns out to be worth 15pts. to the down side, then great. I should hopefully grab maybe 6 pts of that on two trades.

    I guess what I like about what Stridsman is saying is take the freak'n pressure off yourself to get the whole damn enchilada.

    Take what you can, with what your system allows for, then move to the next trade. It is great to have a big 10pointer every now and then. But, I'd take 15 ...(3.5 to 4.0) winners anyday over that one big one...know what I mean.

    Here is an example of how I traded yesterdays market. A few nice ones that really ran hard. But mostly really small gains per trade. they just add up.

    I expected today to not be to giving considering yesterdays plunge. So, I'm mentally hunting for even smaller moves to materialize.


  10. potent.
    #10     Nov 19, 2003