I wrote a bot and gave it 25k to trade
- Thread starter _terminus_
- Start date
%%
Because he was asking for help.
The OP? Maybe I skim posts poorly....but I saw no ask in it. Merely a "this is what I've done".
Your question is in the red part of the circle.
(no control over the events e.g weather, trafic, cancer and sadly most people are stuck in here)
In the forum & life , we are in the blue(?) part (absolute control) & if you build yourself a moral authority, like some members of the forum has, you might get into the yellow part as well (partial control).
But still, that would be just a waste of time, questioning, human behaviour.
They're... Curious my guess ?
%%
Implied question;
quote '' so could have been just lucky'' unquote...………………………………………………………….
Great first post and nice result thus far. I hope that your trading system will continue to give you profits.
Are you considering to also use it with other futures, besides MNQ?
It is interesting to see that robots have become stronger and more productive so that traders can use them on a permanent basis, plus anyone who wants to build on this history can independently develop a system that will work in parallel with the main trading and provide additional income. Why am I talking about this particular format? Because I am fully convinced that trading needs personal attention and understanding of all the details of this issue. Although I can be wrong, because time does not stand still, and if it really helps to make money and there is no global risk to the capital, then you definitely need to try and implement it in your work.
Hi guys thank you for your answers.
@traider - Unfit, fit, well fit, best fit, overfitted
I don't think my system is overfitted, why do you? You could write a simple MA crossover producing very similar backtested results as mine... with less than 10 trades. That would be overfitting and I wouldn't trade it. The backtest above refers to roughly 1.400 trades, much more than the number of decision points and parameters in this system, about 25.
Besides, during the last three years the market has been quite colorful as others are pointing out.
I could die tomorrow, my system could break today, and buy and hold could be the ultimate overfitting strategy, trying to include every information that you think you have in just one trade.
By the way, it gained another 399$ (+0.41%) last night from closing a short trade left open on Friday evening. But it could be just luck. I'm serious. I'm being confident and skeptical at the same time.
@d08 - WL4
Yes, I still use it for rapid prototyping, backtesting and plotting. I replicated its programming environment in C++ using the same call names: InstallProfitTarget, BuyAtMarket etc. After the initial WL4 step I repeat the same tests in my environment, which is the one actually doing the trading.
Slippage
As reported by other posters, slippage in futures is generally low and has not been a concern so far to me. Actually I often get better fills than the backtest, that is positive slippage. But this may have to do with the small size of my positions and speed of execution.
@tommcginnis
Thank you for your encouragement. Yes, I try to remain agnostic on everything and yes, the algos used in this system are quite small. No ML, no neural nets, no fancy stuff, just basic math and statistics.
@Here4money
I wrote the algos and communication part in C++. The front end is coded as a Microsoft Access 2000 application.
@zoned_post_meridiem
No tick by tick, I use 1min bar data.
After testing many systems, I found that most of them give results very far from reality!
@southall
Yes, the system stops trading under certain conditions.
@Trader200K
Thank you. I use IB API via IB Gateway. The system runs continuously Monday to Friday. I use two computers, one for production and one for development. Nothing fancy, the production machine is an 8 yo box running Windows 7.
Using "old" technology such as Microsoft VS 2008 and Access 2000 allows me to benefit from greater reliability and speed of execution. I have always preferred older tools that stood up the test of time. Always been Lindy, even before knowing what that means.
It took me about one year to develop the first working version, albeit I started working on this project many years ago. I've been a freelance coder and IT consultant for 35+ years.
@HobbyTrading
Thank you. Not for now, I tried several versions of the system with other futures and it works pretty well unmodified with MYM and MES, but MNQ seems to offer the best combination of profit and recovery factor.
Asking for help
Well, I guess I'm already in a later phase than looking for help, I'm testing it with my own money, so if anything what I need more now is feedback. Sure, any suggestion will be greatly appreciated, but I'm presenting this work more in the spirit of "look at what I did" and "maybe it can be done" and serve as an inspiration for others, than anything else. Also because no amount of "help" would have been enough without the discipline and willingness to overcome the many problems that such a project offered. Concurrent programming, using the IB API properly, preventing the program from placing wrong orders at wrong times, partial fills just to name a few. And, finding a system that supposedly has some kind of edge.
I don't consider myself an experienced trader. I could never profit from trading by hand. I tried many times and barely broke even, so I guess this is the last chance I'm giving myself in this business. When this system will fail, I'll try to readjust its parameters. Should that not work I'll probably end it.
Most importantly, I could have never completed this software without listening to high volume dub techno all day long.
@traider - Unfit, fit, well fit, best fit, overfitted
I don't think my system is overfitted, why do you? You could write a simple MA crossover producing very similar backtested results as mine... with less than 10 trades. That would be overfitting and I wouldn't trade it. The backtest above refers to roughly 1.400 trades, much more than the number of decision points and parameters in this system, about 25.
Besides, during the last three years the market has been quite colorful as others are pointing out.
I could die tomorrow, my system could break today, and buy and hold could be the ultimate overfitting strategy, trying to include every information that you think you have in just one trade.
By the way, it gained another 399$ (+0.41%) last night from closing a short trade left open on Friday evening. But it could be just luck. I'm serious. I'm being confident and skeptical at the same time.
@d08 - WL4
Yes, I still use it for rapid prototyping, backtesting and plotting. I replicated its programming environment in C++ using the same call names: InstallProfitTarget, BuyAtMarket etc. After the initial WL4 step I repeat the same tests in my environment, which is the one actually doing the trading.
Slippage
As reported by other posters, slippage in futures is generally low and has not been a concern so far to me. Actually I often get better fills than the backtest, that is positive slippage. But this may have to do with the small size of my positions and speed of execution.
@tommcginnis
Thank you for your encouragement. Yes, I try to remain agnostic on everything and yes, the algos used in this system are quite small. No ML, no neural nets, no fancy stuff, just basic math and statistics.
@Here4money
I wrote the algos and communication part in C++. The front end is coded as a Microsoft Access 2000 application.
@zoned_post_meridiem
No tick by tick, I use 1min bar data.
After testing many systems, I found that most of them give results very far from reality!
@southall
Yes, the system stops trading under certain conditions.
@Trader200K
Thank you. I use IB API via IB Gateway. The system runs continuously Monday to Friday. I use two computers, one for production and one for development. Nothing fancy, the production machine is an 8 yo box running Windows 7.
Using "old" technology such as Microsoft VS 2008 and Access 2000 allows me to benefit from greater reliability and speed of execution. I have always preferred older tools that stood up the test of time. Always been Lindy, even before knowing what that means.
It took me about one year to develop the first working version, albeit I started working on this project many years ago. I've been a freelance coder and IT consultant for 35+ years.
@HobbyTrading
Thank you. Not for now, I tried several versions of the system with other futures and it works pretty well unmodified with MYM and MES, but MNQ seems to offer the best combination of profit and recovery factor.
Asking for help
Well, I guess I'm already in a later phase than looking for help, I'm testing it with my own money, so if anything what I need more now is feedback. Sure, any suggestion will be greatly appreciated, but I'm presenting this work more in the spirit of "look at what I did" and "maybe it can be done" and serve as an inspiration for others, than anything else. Also because no amount of "help" would have been enough without the discipline and willingness to overcome the many problems that such a project offered. Concurrent programming, using the IB API properly, preventing the program from placing wrong orders at wrong times, partial fills just to name a few. And, finding a system that supposedly has some kind of edge.
I don't consider myself an experienced trader. I could never profit from trading by hand. I tried many times and barely broke even, so I guess this is the last chance I'm giving myself in this business. When this system will fail, I'll try to readjust its parameters. Should that not work I'll probably end it.
Most importantly, I could have never completed this software without listening to high volume dub techno all day long.
Last edited:
I wasn't only referring to other US equity futures. But other asset classes as well, such as forex, commodities, bonds and so on.
Yes, I meant that I tried several versions of the system with other futures (commodities) not only index futures. And that an unmodified version of the system works pretty well with MYM and MES. Despite that MNQ produced the best results.