You should find the balance in the amount of information you want, and can handle. Each extra filter should be a real acquisition with a decent positive impact on your trading.
Need system back-tester?
- Thread starter Steve Ladd
- Start date
You should find the balance in the amount of information you want, and can handle. Each extra filter should be a real acquisition with a decent positive impact on your trading.
I've been trading this system discretionarily for a year and have done well. (So has the IRS, unfortunately.) I'm willing to invest in it. In this forum I have tried to discuss it but it is too different for other people to be interested in. It is complicated due to the number and disparity of assets I follow and because for each asset I have seven measures. It's low-frequency. The automation would only be to back-test and optimize, not run the trades.
If you have done well.. then stop trolling these boards and PROTECT YOUR IP.
If your edge is truly replicable and valuable, email me and I can set up up with a blind template w/ generic triggers. It will cost you some in investing in a b'testing platform ~$350+ data, 1-2 months of learning to program on your own... book me for coding help after NYSE hours, or other Amibroker consultants in this space... once u are on your own, plug in your "edge" ..
If your edge is truly replicable and valuable, email me and I can set up up with a blind template w/ generic triggers. It will cost you some in investing in a b'testing platform ~$350+ data, 1-2 months of learning to program on your own... book me for coding help after NYSE hours, or other Amibroker consultants in this space... once u are on your own, plug in your "edge" ..
%%
Most likely will not get into trouble with complexity due to number of assets, Steve.I've gotten an idea from the guy that wrote the book Rocket Science For Traders. but i never bought the book or bought the idea , because trading is not rocket science. The low frequency part sounds like a big advantage, it is for me anyway.
But the complex part most like to get you in trouble/blow up., is your ''position size based on feel '', think you said. [Dont take this as a negative insult, but if you keep on position size based on feel, you may never have to even file with IRS, or state tax/any tax again if applicable] LOL
And this is not really negative on small caps[IBD trades/invests plenty of them] ; but complexity of closing a position of 2000 stocks can have huge slippage compared to 100.Wisdom is profitable to direct.
I don't exceed certain position sizing rules I've come across in the literature but I haven't settled upon a mathematical way to determine the exact amount. I need to change that. But back to my present purpose, would anyone recommend a trading code writer suitable to my needs? I understand they may be expensive.
This cannot be programmed because it's discretionary
Which makes it difcult to backtest in a reliable way. Each new backtest will give at least slightly different results as discretionary will never take for 100% the same decision at the same time.
If you have a system that has done well and the reasons are well-understood, congratulations for finding something which means something to you. The low frequency part sounds like part of the edge, and few here would argue with that, though it can also provide "lucky breaks" which lessens statistical significance.
What you have though is a discretionary system, which so far seems to work for you, but also not possible to backtest. I suspect that if you get to backtest it, you'll be disappointed, 'cause that's what good backtests do. They show you what the real results would be under real circumstances (up to a certain level of detail ofc), over as many years and market conditions as possible. Backtesting discretionary will just be too much of a hassle and biased to be of much value.
If you really want to backtest, which can bring many positive rewards and learning experiences as well, you will need to study how to become a mechanical trader. What makes you focus on each column/value in Excel, which stocks and setups to take? If you can't develop your own rules for this on paper, then it seems you're stuck, or you need an AI/ML algo to learn from you, something much much harder than just a regular backtest with regular trading rules.
Hiring a programmer to automate this won't get you very far unless that programmer is a better trader than you, and then again, it'll not be your system, but that programmer's. So it's imperative you figure out what makes your system tick, why you decide as you do and become clear on that, or risk lots of spent effort, time and pain to become mechanical without the aptitude for it.
Mechanical systems are dumb, they just do what the rules say, so it's different from selecting based on whatever you "feel" or "see" at the moment, which can work for those who prefer those methods, but ain't mechanical enough. So how much disappointment this process may bring can not be overstated enough. On the plus side, mechanical backtests should be fully repeatable, so makes it easier to go over the tests again and compare (which comes with a whole set of more pitfalls, see below).
Without more details or "juice", it's hard to offer more help, though for those of us who spend this time and energy we've more than enough with our own ideas and attempts at implementations.
One little tidbit I recently learned is that when I simplified to a much quicker backtest/programming-cycle, I ended up with heavily curve-fitted systems, though I like curve-fitting functions and long-term, they haven't made 1 real trade yet so I suspect the process has lead to fishy rules in this case. Ironically, I made a better/more complete system when full backtests took good portion of a day.. But this may also have to do with making everything from scratch, so I suspect is part of that process too. So yet again I re-realized something that I had realized before, thanks to backtesting, though not "profitable to direct" yet
.
Spending years fleshing out rules that satisfy both past and future is different than calling all the shots while chilling on the beach. So also see if you have aptitude for this kind of trading. In essence, what we're searching for is the holy grail, which doesn't really exist, but often dependent on changing market conditions. If someone can be consistently profitable on random data, then maybe they're there already...
One more thing, what I've used alot is writing down ideas, and lately, trying to structure my thoughts along the lines of "what do I know that I know" and "what do I know that I don't know". Methods like these and writing down thoughts helps to organize them, and prioritize based on criterias that should help you make progress toward your goals. After all, organizing thoughts and ideas is mostly what programming is really about, and the lack of it, will often result in bad systems not solving real-world problems.
What you have though is a discretionary system, which so far seems to work for you, but also not possible to backtest. I suspect that if you get to backtest it, you'll be disappointed, 'cause that's what good backtests do. They show you what the real results would be under real circumstances (up to a certain level of detail ofc), over as many years and market conditions as possible. Backtesting discretionary will just be too much of a hassle and biased to be of much value.
If you really want to backtest, which can bring many positive rewards and learning experiences as well, you will need to study how to become a mechanical trader. What makes you focus on each column/value in Excel, which stocks and setups to take? If you can't develop your own rules for this on paper, then it seems you're stuck, or you need an AI/ML algo to learn from you, something much much harder than just a regular backtest with regular trading rules.
Hiring a programmer to automate this won't get you very far unless that programmer is a better trader than you, and then again, it'll not be your system, but that programmer's. So it's imperative you figure out what makes your system tick, why you decide as you do and become clear on that, or risk lots of spent effort, time and pain to become mechanical without the aptitude for it.
Mechanical systems are dumb, they just do what the rules say, so it's different from selecting based on whatever you "feel" or "see" at the moment, which can work for those who prefer those methods, but ain't mechanical enough. So how much disappointment this process may bring can not be overstated enough. On the plus side, mechanical backtests should be fully repeatable, so makes it easier to go over the tests again and compare (which comes with a whole set of more pitfalls, see below).
Without more details or "juice", it's hard to offer more help, though for those of us who spend this time and energy we've more than enough with our own ideas and attempts at implementations.
One little tidbit I recently learned is that when I simplified to a much quicker backtest/programming-cycle, I ended up with heavily curve-fitted systems, though I like curve-fitting functions and long-term, they haven't made 1 real trade yet so I suspect the process has lead to fishy rules in this case. Ironically, I made a better/more complete system when full backtests took good portion of a day.. But this may also have to do with making everything from scratch, so I suspect is part of that process too. So yet again I re-realized something that I had realized before, thanks to backtesting, though not "profitable to direct" yet
.Spending years fleshing out rules that satisfy both past and future is different than calling all the shots while chilling on the beach. So also see if you have aptitude for this kind of trading. In essence, what we're searching for is the holy grail, which doesn't really exist, but often dependent on changing market conditions. If someone can be consistently profitable on random data, then maybe they're there already...
One more thing, what I've used alot is writing down ideas, and lately, trying to structure my thoughts along the lines of "what do I know that I know" and "what do I know that I don't know". Methods like these and writing down thoughts helps to organize them, and prioritize based on criterias that should help you make progress toward your goals. After all, organizing thoughts and ideas is mostly what programming is really about, and the lack of it, will often result in bad systems not solving real-world problems.
Last edited:
Simple's post should be in some ET section for all to see...! 
Bullshit.
Indians are stealing code from other websites.
Indian traders? LMAO!
Indians - majority are copy&paste clowns.