To those who would interpret this as a covert attempt at reviving my nondescript journal, rest assured that it is merely a onetime update of my recent peregrinations. In fact the tone of this post will be serious and any references to humor will, as in the past, be purely coincidental. Not even a well-coordinated smear campaign by its city elders could thwart my arrival in Las Vegas. Upon hearing of my impending move, the city was put on high alert with my picture distributed to local airline, railway and bus officials. National Guard units manned checkpoints on all incoming highways and roads, with police choppers hovering threateningly overhead. All this notwithstanding, I still managed to sneak in using the subterfuge of a SARS mask and an accommodating camel. A despondent Mayor Oscar Goodman finally called a truce and upon receiving a signed copy of my latest book "The Fractalship of the SOMP", presented me with a key to the city. Visibly touched by this overflow of generosity, I told him a key to the Spearmint Rhino would have sufficed. The following ET members were instrumental in providing advice and assistance which helped turn a potentially chaotic life change into a mere logistical nightmare, and for this they deserve my thanks : Brutus, Magna, MrDinky, rs7 (Error 404), rtstrading, thetraderprofit, trader88 and zxcv1fu. I have utilized this recent break from trading to look into, dare I say, system trading. To this end, I have been regularly downloading the daily ES tick data from CME's website which I have been analyzing with the aid of software I write in Pascal. I also started reading the bible of system design, "Trading Systems and Methods" by Perry J. Kaufman but have lost confidence in the writer's authority upon detecting an apparent glaring error already on page 24 of the third edition. Had this been the first edition, I wouldn't have minded. From p. 24 : "Laws of Probability Two basic principles in probability are easily explained by using examples with playing cards. In a deck of 52 cards, there are 4 suits of 13 cards each. The probability of drawing a specific card on any one turn is 1/52. Similarly, the chances of drawing a particular suit or card number are 1/4 and 1/13, respectively. The probability of any one of these three possibilities occurring is the sum of their individual probabilities. This is known as the law of addition. The probability of success in choosing a numbered card, suit, or specific card is P = 1/13 + 1/4 + 1/52 = 18/52 = 35%." What was Perry thinking? And to think that I was relying on his later chapters on adaptive techniques and genetic algorithms to help transform my futile attempts at curve fitting into a robust system.
Hello Subman, Glad to here your alive and well and Las Vegas has opened its' arms to you. Blessings Make 'em pretty, Chris
Ack, what a gaff. Really inexcusable for someone passing himself off as an authority on trading system design. I suggest you read Smarter Trading instead. It's by the same author, but I haven't detected any any flaws of that magnitude in it.
Just curious about your reasoning behind moving away from SOMP? As, you seemed to have it down pretty good.
I am looking for something a little less stressful both in terms of responsibility (blame the system not me) and sheer quantity of trades (1, maybe 2, per day as opposed to 15). In "Fooled by Randomness", Taleb portrays a happily retired dentist who expects to earn 15% annually on his investments with a 10% error rate (or volatility) per annum. This "translates into a 93% probability of making money in any given year. But seen at a narrow time scale, this translates into a mere 50.02% probability of making money over any given second as shown" in the following table : Code: [b][u]Probability of making money at different scales[/b][/u] Scale Probability ----- ----------- 1 year 93% 1 quarter 77% 1 month 67% 1 day 54% 1 hour 51.3% 1 minute 50.17% 1 second 50.02% (From p.57, "Fooled by Randomness" by Nassim N. Taleb) Taleb continues : "Over the very narrow time increment, the observation will reveal close to nothing. Yet the dentist's heart will not tell him that. Being emotional, he feels a pang with every loss, as it shows in red on his screen... At the end of every day the dentist will be emotionally drained. A minute-by-minute examination of his performance means that each day (assuming eight hours per day) he will have 241 pleasurable minutes against 239 unpleasurable ones. These amount to 60,688 and 60,271, respectively, per year. Now realize that if the unpleasurable minute is worse in reverse pleasure than the pleasurable minute is in pleasure terms, then the dentist incurs a large deficit when examining his performance at a high frequency."
In between trading, long sessions at the poker tables and mandatory community service, I still manage to find time to backtest ideas I concoct during these aforementioned activities. Rigid application of the principles I have gleaned from Kaufman and Malkiel, together with a seeming lack of imagination, have resulted in most of these "systems" not making it past the back-testing stage. As all disillusioned system testers know, when in doubt it is best to err on the side of safety - I have a nagging suspicion that adherence to this precept may have been the reason for my rejecting a potentially profitable system, or two. Let me illustrate. Using CME ES tick data, if the last sale is currently at 1038.75 and I have a limit order to buy at 1038.25, then this order is deemed to be executed (for testing purposes) if and only if there is at least one print at 1038.00 (or lower). In real life, the price could "ping-pong" between 1038.25 and 1038.50 just long enough for the buy order to be executed with the price continuing higher, not even once touching 1038.00 (for the moment at least). And herein lies my dilemma - assuming that this scenario is repeated many times a day, instead of blanketly ignoring these trades, what percentage, if any, should I allocate as executed? If this is what turns a losing system into a profitable one, is there any reason to feel inadequate or insecure? The following 27 seconds in the life of yesterday's (October 10) ES December contract taken from CME's FTP site, shows the initial price at 1038.75, the "ping-pong" between 1038.25 and 1038.50, and the subsequent return to 1038.75 : Code: Time Price Quantity Bid Ask ------ ------- -------- ----- ----- 083600 01038.75 11 1038.50 1038.75 083602 01038.50 28 1038.50 1038.75 083602 01038.75 5 1038.50 1038.75 083603 01038.50 425 1038.50 1038.75 083604 01038.75 25 1038.50 1038.75 083604 01038.50 288 1038.50 1038.75 083608 01038.25 2 1038.25 1038.50 083608 01038.50 10 1038.25 1038.50 083608 01038.25 6 1038.25 1038.50 083608 01038.50 5 1038.25 1038.50 083608 01038.25 25 1038.25 1038.50 083609 01038.50 35 1038.25 1038.50 083609 01038.25 19 1038.25 1038.50 083611 01038.50 1 1038.25 1038.50 083611 01038.25 1 1038.25 1038.50 083611 01038.50 4 1038.25 1038.50 083612 01038.25 27 1038.25 1038.50 083613 01038.50 25 1038.25 1038.50 083615 01038.25 5 1038.25 1038.50 083616 01038.50 2 1038.25 1038.50 083617 01038.25 5 1038.25 1038.50 083618 01038.50 25 1038.25 1038.50 083618 01038.25 9 1038.25 1038.50 083621 01038.50 3 1038.25 1038.50 083621 01038.25 3 1038.25 1038.50 083621 01038.50 2 1038.25 1038.50 083622 01038.25 11 1038.25 1038.50 083623 01038.50 23 1038.25 1038.50 083623 01038.25 50 1038.25 1038.50 083623 01038.50 87 1038.25 1038.50 083624 01038.25 1 1038.25 1038.50 083624 01038.50 21 1038.25 1038.50 083624 01038.25 3 1038.25 1038.50 083624 01038.50 348 1038.25 1038.50 083625 01038.25 20 1038.25 1038.50 083625 01038.50 206 1038.25 1038.50 083625 01038.75 1 1038.50 1038.75 083626 01038.50 11 1038.50 1038.75 083626 01038.75 3 1038.50 1038.75 While I don't have access to historical market depth, it can be readily seen that there were 187 contracts traded at 1038.25, meaning that my buy order could have been executed. When navigating the realm of numerical estimation, I'm as innovative as the next guy, so I'm not really looking for methods of implementing this. What I would welcome, however, is critique or feedback from people who have encountered this very problem and whether they chose to deal with it.
That doesn't look like a risk free quarter or half a point to me. I like looking at a longer time frame to trade than 27 seconds...that looks too much like a train wreck.
Nowhere do I speak of profit target, time frame or risk. I bring the CME data merely to illustrate my point, and that is the treatment of limit orders in back-testing. Assume I've been in the trade for 27 minutes, short at 1040.25 with a 2 point target (limit order to buy at 1038.25). From what you can see, is my trade closed out in those 27 seconds? Traditional back-testing says no. I say maybe.