Auto Mechanics vs. Automated Traders

Discussion in 'Automated Trading' started by chaostheory, Apr 18, 2009.

  1. What do good mechanics and successful automated traders have in common?

    It must be that both have complete training and education. That is, each must fully know their profession in order to be successful.

    For automated traders that includes scaling into/out of positions, money management, risk management, position sizing, order tracking, self psychology, portfolio management, diversification, and much more.

    Now, in contrast, what's the major difference between the training of a mechanic and the education of a successful trader?

    It appears to be the shocking contrast in the amount of misinformation. Newbie traders undergo an avalanche or false, misleading, misguided, or confusing information. At least, that's the experience of most.

    Personally, after finally breaking free from the standard "school of thought", I now trade only systems that profit at least 98% of every month. Month after month.

    In fact, the individual systems in my portfolio statistically all trade at 100% win rates. Some play the long trends, others short trends. Some play the reversion to mean and more.

    I used to think that traders who claimed 90% plus win rates were either liars, weren't really trading only "curve fitting" or else using some kind of martingale betting strategy with exponential risk.

    But I don't use any martingale strategy. I simply scale into and scale out of position in, admittedly, a rather clever technique.

    One fact seems to really stand in the way of many succeeding at this arena and that would the be the tools.

    How many tools make it effective and easy to thoroughly simulate the market using tick data with portfolios of strategies on a single instrument so you can study and refine your diversification and scaled position sizing ability?

    Anyway, I was forces to build my own platform to achieve these.

    So perhaps another difference between mechanics and auto traders that succeed are the tools available.

    Mechanics with the modern electronic and power tools will repair more cars faster and with greater accuracy that those without the proper tools.

    Just some thoughts for discussion.
     
  2. I'm not disputing any of your points, but you could take pretty much anything and say:

    [random] vs Automated Trader

    Let's do an example:

    Gay Prostitute vs Automated Trader

    - Gay prostitutes and Automated Traders both need to assess risk. For Gay Prostitutes, it's the risk of AIDS; whereas for Automated Traders, it's the risk of the exchange pushing out bad packets with erroneous sequence numbers.

    - Gay prostitutes need to choose their clients successfully. They need to have a long run expected value that's positive on income. Automated traders need to choose their trades correctly, in order to have a long run positive expected value.

    - Gay prostitutes need to use computers, to flag their interests to the sex exchange (craigslist). Automated traders have to use computers to flag their interests to the stock exchange (NYSE/NASDAQ).

    ...

    I mean if we want to quantify the intersection of some arbitrary quantity with automated trading, we could go at it all day. Let's talk numbers and quantitative strategies, or technology. :)
     
  3. travis

    travis

    I read about this pyramiding concept before. I remember reading Jesse Livermore about it, and a chapter in Nazer J. Balsara's "Money Management Strategies For Futures Traders".

    I can't fully grasp it - I can't grasp very much in this field, as I am quite ignorant on formulas and math in general. By what you are saying it sounds like it can produce miracles (not to say that I am skeptical about it). I don't understand how it works, besides the fact that its implementation seems to me very complicated (also the back-testing of it).

    If you or anyone else could write a simplification of it or a dummies guide (even just a few sentences), I'd appreciate it.
     
  4. I don't do pyramiding which is similar to the Martingale betting approach.

    The above refers to simple scaling into or out of a position. I may wax poetic about it but it's nothing new. It's been written about, blogged about, threads on ET, ad infinitum. Trouble is that it's downright hard to get setup to really try it out and get your mind around it.

    In contrast, simple long, short, flat positions are so much easier to understand but conversely so much harder to make consistent with them.

    My guess is that's why the academics keep saying the markets are random because they think of trading only as short, flat, and long. Just a thought.

    Briefly, scaling means you buy one unit (whatever size you can afford to do multiple times) at a certain price, then buy another of the same or less size and so on.

    The key here that's different from pyramiding is that you keep buying the same size or less. Pyramiding usually means buy 1, then buy 2, then buy 4 or some other increasing number. Pyramiding can be very dangerous due to the exponentially increasing risk but doable, I guess.

    My personal Eureka in this area was grasping the value of the average of the entry prices after adding to and removing from a position.

    To clarify that some: when you partially scale out but not completely then you have to keep averaging your entry price when you turn around and continue adding to a position.

    Basically, there's real power in being more fluid or fuzzy logic, if you will, about a position. Rather than just long / short / flat, you can be flat / a little bit long / more long / a lot more long / a little bit short/ more short / or a lot more short.

    That means, for example, that you can then go from very long to less long and back to very long. So there's tons more potential "positions" available to move between.

    That can often work out as more profitable than simply moving between flat, short, and long.

    But what software platforms make this type of scaling easy to track and test while you build your skill at this?

    Also, it seems rather challenging to handle this in a discretionary way. Perhaps certain ways of trading only become practical with automation.

    It would be interesting to learn if anyone does this by hand.
     
  5. Google sub- and super-martingale.

    You have a probability set, and you place your bets according to the "statistical edge" you have.
     
  6. travis

    travis

    Premise: if you are tired of my questions or don't have time, feel free to ignore what follows.

    No, I wasn't talking about Martingale, nor were the two I mentioned.

    "The above refers to simple scaling into or out of a position.": that is exactly what Livermore and Balsara talk about, and that is why your ideas sound reasonable to me, even though I had no idea about how to implement them practically.

    "Trouble is that it's downright hard to get setup to really try it out and get your mind around it.": right, that is exactly my problem. On top of it, I am a poor programmer.

    "In contrast, simple long, short, flat positions are so much easier to understand but conversely so much harder to make consistent with them.": that is exactly my problem - I like to keep things simple at all times, so I can understand everything I am doing, but this scaling into or out of a position interferes very much with keeping things simple.

    "Briefly, scaling means you buy one unit (whatever size you can afford to do multiple times) at a certain price, then buy another of the same or less size and so on.": that is exactly what I thought, but my logic then says "why should I not buy and sell all contracts at the moment I appraise as best and not worry about the rest?".

    "The key here that's different from pyramiding is that you keep buying the same size or less. Pyramiding usually means buy 1, then buy 2, then buy 4 or some other increasing number. Pyramiding can be very dangerous due to the exponentially increasing risk but doable, I guess." Wait, sorry the laziness in not going to open up those two books again, but I don't think that everyone means Martingale by "pyramiding", and especially neither Livermore nor Balsara. Sorry for daring to reply after my premise that I don't know/understand anything about this, but there were a few things that seemed clear.

    Actually you see, I remember clearly that Livermore said "if a position works, add more to it" and "if a position doesn't work, there is no reason why you should add more to it". So he didn't mean Martingale, which when something doesn't work, doubles up the investment. But the problem is that my logic never even understood what Livermore said. I can't grasp it, once again.

    "Basically, there's real power in being more fluid or fuzzy logic, if you will, about a position. Rather than just long / short / flat, you can be flat / a little bit long / more long / a lot more long / a little bit short/ more short / or a lot more short.": very clear, once again. Very clear what you guys mean (you and the other two), but I'm gonna try now to put into words why my logic doesn't get it. I don't get it because... let us say my system says buy when so and so happens and sell when so so and so happens, because it has worked in the past. Now let us say that once the position got opened, two hours later was making money. Now I do not have any statistical data backing up the (Livermore) concept that all trades that work after two hours are more likely to work until the end. Or do I? All that my testing says is that if you open your trade with x situation and close it with y situation, it is likely to make money. It says nothing about whatever happens between x and y, so, trading based on the assumption that a trade that is doing well between x and y, will be even more likely to do well by the time it reaches y, is equivalent to trading based on just my assumptions and not backtested formulas (which is something I am not supposed to do).

    Yes, of course, Livermore would say that if your trade starts doing poorly, you should reduce once again your positions (and eventually close them), but livermore wasn't first of all backtesting any of his methods and second of all how do you account for all the fees you're paying, and after all, if it isn't true that a trade that does well at the start is more likely to do well through the end (and certainly it is not tested), then you're just going to end up with a system that adds commissions and slippage to the original system, without adding any profits.

    "But what software platforms make this type of scaling easy to track and test while you build your skill at this?": this is great because you answered exactly my previous question: "How do you test all this to prove that it works?".

    So, what software platforms make this type of scaling easy to track and test?

    (Not that I will buy the software, because this all seems too complicated for my limited skills, but let's at least tell the others who may be reading. As I said at the start, feel free to ignore my entire post, because I know I just have too many questions and doubts, and also too much ignorance and stupidity, if compared to the average trader on this automated trading section of the ET forum).
     
  7. The answer is simple but not easy and depends on whether you're following or fading the trend.

    Following the trend then if you start with 10 units and the market goes against you right away, you lose 10X your stop loss amount. That will happen for the vast majority of the trades you put on when trying to catch trends.

    When you truly do discover a trend with a fixed position size, the profits can sometimes compensate for all the losses but never sufficiently to warrant the risk of all those losing trades.

    In contrast, if you start with only 1 unit and get stopped out, you lose 10X less money on all the stopped out trades.

    Now, since you started with 1 unit and gradually increase by 1 at a time, then when do find a strong runaway trend, you multiply your profits fantastically over the losses you incurred.

    In this scenario, even if you set a simplistic (newbie level) trailing stop from the highs then that means you have zero risk and guaranteed locked in profits that continue to grow and expand in the case of that wonderful run away trend.

    Clearly, taking all the losses with only 1 unit but extracting the profits at many multiples of that definitely compensates fabulously for the risk of all the losing trades. Do you see?

    Another painful aspect of trends with a fixed position size is the danger of exiting too early. Why? It's absolutely necessary to follow a trend all the way till the end in order to compensate for all the losing trades which are the cost of doing business in trend trading.

    If you exit from a trend with a fixed position and later see the trend suddenly continue, you're in huge trouble. It's because the psychological urge to reenter the trend becomes overwhelming. It's truly terrible to be on the sidelines of a huge trend. Even worse if you took a lot of small losses to get this this point. Sadly, if you make the mistake of reentering you almost invariable enter at the wrong time and take a major loss. Possibly multiple times. The odds are against reentering a trend correctly.

    If instead of exiting entirely, you scale out gradually on retracements and scale back into the trade on continuation, then you avoid all those excruciating problems and guarantee that you ride the trend all the way to the very end even if the trend suddenly bursts into action again.

    The point, is, you can NEVER predict when that trend will end. And with scaling positions you never have to. Your position size adapts to the strength of the trend automatically according to your mathematically derived plan.

    Of course, then there's fading the trend. With the right tools and skills, you can play the opposite direction of that trend at the same time. The real purpose of this is to profit during all the sideways markets as a reversion to mean type of system.

    In that case of fading a trend you look to profit from when the trend does completely terminate.

    There again you have a major advantage of scaling your positions.

    If you were to a fixed 10 units into a trend (that means you're playing counter trend), that sets your sell price at X but if the trend continues going strong rather than retrace right away, then you lose money at 10 X that rate and when do you exit?

    Worse yet. if you manage to sell into the very beginning of a major trend you might not see prices get back to a break even or profitable point for YEARS and possibly never at all.

    However, if you sell one contract at price P1 and then another at P2 then you have 2 contracts. But now your entry price for break even rose to the average of them or (P1+P2)/2.

    So, in effect, you just magically raised the entry price of your first contract. Viewed another way, if you enter too early, this is like taking an eraser and fixing your mistake. It feels wonderful psychologically to be able to fix your mistakes.

    If you continue adding to your position evenly, then you effectively continue raising your original entry price.

    Markets ALWAYS, ALWAYS retrace to some degree. And eventually they retrace 50%. So if you scaled appropriately in order to profit from a retrace to X% level then you're guaranteed to eventually exit with a profit.

    90% of the time this only takes hours or days but, of course, sometimes for a strong trend it can take weeks or months to retrace enough to take profit--meanwhile your risk continues to grow.

    In Conclusion:

    None of these methods are tenable alone, in my opinion. Few people can endure the long string of small losses required to find a trend because they want their trading to create a steady cash cow that profits every month.

    On the other hand, the trend fading type systems generate large amounts of cash like clockwork during sideways or slow trends. But unfortunately, it creates a lot of risk and ties up a lot of assets during major runaway trends.

    So, the real power comes from merging these two.

    Then the profits from one more than compensate for the risk from the other since they show reverse correlation so that you attain a nearly straight diagonal equity growth curve.

    It's truly lovely to see on a chart.

    In short, if the trader quits the game of playing probabilities and looks only for certainties based on position sizing, then the markets become a money machine.

    Actually, doing this takes all the fun out of trading. But who needs fun from the markets? This is about money. There's much more exiting things to do with time than trading.
     
  8. Hey, that was my question. I never found one at retail prices. Maybe institutions have something that costs 5 or 6 figures to purchase.

    I personally wrote my own. But even though it's now available as a product, I'm not at liberty to divulge where and how to get it.

    Besides, that's not my point.

    The point is that to point people in the right direction as to their goals. Once someone starts looking in the right direction, they will find or create the tools they need.

    Posts on ET pointed me in the right direction. It took a long time to sink into my thick skull.

    I'm just "paying it forward" here for those who are willing to work to find the solution.

    Either find the tools or create them yourself like I did.

    Wayne
     
  9. jjw

    jjw ET Sponsor

    it the system is available as a product, why can you not identify it?
     
  10. Pippi436

    Pippi436

    Of course there are applications available to retail where you can code your own position sizing methods(including scale in/out, pyramiding, mgale or what have you). Smartquant, Neoticker, and i'm sure quite a few others. You are pretty much only limited by your trading and coding skills.
     
    #10     Apr 19, 2009