Turok RE The problem for testing is that you add/change rules *after* the testing shows poor results What prompted the adding of the trailing stop abogdan was an AMGN long positon on the model I was running. There was not a situation of poor results in this case, unless you were referring to something else. Amgn open 63.99 High 64.40 low 63.60 Up target 64.40 low target 63.35 Close 64.07 This was not a bad result situation. As a matter of fact I would have suggested trailing stops in profitable situations if abogdan did not add it in further clarification. Also the other 6 stocks all made there 1% targets both up and down.
>What prompted the adding of the trailing stop >abogdan was an AMGN long position on the >model I was running. Roxy, I know nothing about your knowledge in the area of system testing so don't take what I say here a patronizing. I just have to state a very basic principle for the record. Also, this next comment assumes that your statement above is true and the reason a trailing stop was "added" was because of the behavior of a particular position you were in at a particular time. If you expect to be successful in the system side, and you decide to "add" a rule because of one particular trade or position, then you must back up and see how that rule effects the system overall. In summary, the *reason* the multiple rules were added is not important to the result, but the situation must differentiated or the rule applied across the board and the result reanalyzed. And I would expect the 6 stock to reach their 1% target over and over and over and over and over and over and OVER... don't be fooled into thinking that in itself makes it a tradable system. I ask you personally Roxy, do you have any idea what this system does on the bad days? -1x, 2x, 5, 10x of the good day? Do you know the average gain, loss, profit distribution, win ratio, profit factor, sharpe ratio, equity curve, average shares etc? If you don't, my polite recommendation is that you get to work in defense of your trading account. JB
Abogdan: >Please let me know if I was clear Well, that depends if I am supposed to glean information from other previous posts or if I am to use only this post as the *complete* system. I don't see any algorithm for determining the new share size at each flip. As I noted in a previous post this multiplier can be fixed or can be a formula taking into account the delta and other predictables. You previously suggested a 1.1 multiplier but I'm not sure this is what you wish me to test or if you will later say that you have a different formula for this. Since share sizing is so integral to this system, without an answer on this there is no way I can proceed and hope to even approximate your test results in any way. JB
Sorry, I was not clear. The multiplier is determined by the average flip loss. I just used 1.1 as an example assuming that the average loss was 0.1% and the target was 1%. obviously, if you have different ratio then the multiplier will be different too. Also, you don't have to close positions everyday if you did not reach the target. You can carry it over and use higher target. This is also is very much optimizable. Regards,
i just basically opened up a spreadsheet to find some good numbers. it really helps you understand what's going on. play around with it, abogdan has provided us the framework.
>just basically opened up a spreadsheet to >find some good numbers. it really helps you >understand what's going on. play around >with it, abogdan has provided us the framework. I understand quite well what's going on. I've got the specialized backtesting software to "play around" far more than can be done in a spreadsheet I'm attempting to produce solid evidence which is the only thing that one should trade on. Would you mind posting your variables and your spreadsheet results for the following days?...(all 2003) 3/11, 5/9, 7/10, 9/19, 9/29, 9/23, 10/14, 10/20, 10/21, 10/29. Number of flips, share totals, share mutiplier, delta, profit target, etc. That would be a big help to those of us attempting to confirm tests on this system. Also, may I ask what spreadsheet you are using (I assume excel) and how many days of Bid/Ask data did you import into your spreadsheet? Thanks JB
>Multiplier is the simple ratio between >your target and the average flip loss I figured that would be the case. I just needed it confirmed from you for the purposes of the test. Thanks JB