I zipped through the trades and I annotated the segments as black (long) and red (short). I am pleased to see that we can do trading suppression using a signal genrator (the volume 50%). The great thing that shows up is the chop stuff I mentioned before. These are the adjacent bar trades that give up the volatility of the bar involved. For past indicator traders, you will recognize how doing icebergs on the STOCH14 worked to eliminate the whipsaw. There is a good example of four o'clock drift on low volume (see the slow downward price with low volatility bars with high overlap. We will stick with indicator and volume coding and NOT do any staight PA coding for the time being. The annotation names you use on the chart are terrific. As simple developmental analytical technique is to use a matrix. there are two axis and for some reason the typewriter dominance got put into PC's rather than the science of time based analysis. So I vote for letting time show as a vertical dimension going down and placing the variables in columns. Since we are binary there need be only a few null hypothesis columns for each indicator relative to their locations (chosen signal settings) as go/no goes. Most logic can be done with "AND's, OR's, Inverters and clocks that count either way (the bi-stable multi-vibrator). Luckily coding has so many different instructions, we can bypass the simple stuff with instructions and just go to the suppression circuitry which so far does not have simple code named instructions. One note. If you invert the positive output of an AND or an OR, you do NOT get the opposite logic output of the negative out put of an AND or OR. Since we have shown two five legged AND's on the first demo post, this is a consideration for keeping out of induction.
Great trading log. It is set up as a typewriter orientation. We can run a logic diagnostic along side it on the same excel sheet. I wish to refrain from doing a bar by bar diagnostic because bar-by-bar clouds the issues as the day unfolds. We have raw price and volume coming into the coding (software) Could we put in columns for each indicator to say on a go/no go basis the status of each indicator output? Use the PACE columns for volume and add columns for PACE shift and volume direction bar-to-bar. The best of all possible worlds is a row for each successive (collectively speaking) delta of the system. For now the finest delta is 5 minutes where one column could register change. Often there would be simultaneous change in several columns. Each column represents a binary go/no go signal. And the columns are individually the toatality of the signals in the system. This keeps us in a non probabilistic "certainty" setting at the end of each sweep of the code. Thus we always operate in NOW and there is zero prediction. This system will always use market orders and thus give up the "volatility" from close to close. Later we can ease on down to faster rep rates and squeeze out "giving up any more than the liquidity spread which in ES is a tic). Let me know about all the columns. Work up from the bottom of the chart indicator by indicator. We can chunk all of the volume columns to the very right. By doing this we get to look at what happens to the rows that form the nonstationary window at any time and, further, we get to see when the left side (meaning top of in this matrix orientation) of "end effect pane" begins to appear. With this done we can roll through the code versions to get to the effectiveness, efficiency, and optimumization issues. We have the chop, drift, flapper and initial suppression on the table now.
Here is my chart with the segments on it. Wins are black upward slope and red downward slope. Losses are opposite. Look at the A condition of MACD and the rattling around (B) of STOCHS. Do an If A, then suppress B for version 3.0 of the code. I also want to pick off the first rocket of the day. we can deal with that later. For past viewers this is called the 30 minute trade (one contract (DJ as I remember) and 30 min a day trading to make 75,000 a year for early birds who have jobs.)
Let me know how we missed the nice short (in version 2.0) after the long that you started day with. We caught it in version 1.0, I believe.
ScottD: Did you also know that Jack has a long internet history? Here's a "mentoring" example. Some guys from Harvard came up empty handed after trying to decipher Jack's "teachings" for more than a year, starting in 1999. Look at posts # 1 and 2 here: http://groups.google.com/group/misc..._frm/thread/e21428336d0962ed/4088dacd0b5143e7 Now see how they felt over a year later: ...I don't understand why it's not possible to verify such a successful futures trading system if it is based on objective rules. My own attempts to automate the system have been frustrated at every turn by an enormous problem that is apparently only resolvable by gut instinct... I have spent a great deal of time respectfully reading and thinking about what you have written, but mostly what I feel now is frustration. http://groups.google.com/group/misc..._frm/thread/ccc785d53a418f8a/ad6210f80809669d Here Jack tried to obfuscate his way out of his sinking stock picks that lost $22,000: http://groups.google.com/group/misc...d/thread/981513b3a003f259/2db2e88d1407af42?hl Here he LOST 24% in a trading contest and unsuccessfully tried to obfuscate his way out of it: http://groups.google.com/group/misc...a927db8ba840?hl=en&lnk=st&q=hershey+loses+24#
What exactly is A? I would like to see how the coder figure that one out. Seems overly vague, and difficult to program to me at least.
All morons. Profit for that test was -912.50, or at $50 per point -912.5/50=-18.25 points approximately, and I guess includes commissions.
No, these may actually be the same people, because there is no logical connection in this statement. In fact, it's highly suspicious because it is so vague. These types of conversations with regard to developer to coder are usually very specific, and there's nothing to suggest that they aren't working together already. Also, if I wrote a program that complex that turned out to be unprofitable, it would be scrapped at this point in a heart beat. The logic actually isn't that complex, and for that reason it's also not profitable.
Attached is an updated trade log. Contains the values for all indicators at the end of each bar. Most of the columns are raw data from the logic engine. The volume pace values in the far right hand columns are calculate by the spreadsheet.