When the trading is restricted to one active trade at a time, the algorithm becomes extremely starting point sensitive. Basically you'll select which shares to buy simply based on when you set the starting date for the algorithm. I think that basically makes it random. Would be great to see a video where you code up a way to take an algorithm like this and programmatically check starting point sensitivity.
Thanks a lot for sharing your thoughts Andre. I have already covered to tackle that "problem" by e.g. splitting in some previous videos but will also cover that in future videos.
Simple and effective as usual. Bravo! A humble proposal : what about to use a correlated couple of futures, such as ES and NQ and to trade their spread ? Instead of a single equity?
Thanks for sharing! Nice job! I like the way you do experiments. Could you make videos on risk management of assets when, for example, they cointegrate? Or other types when we need to care about risks. Thank you and good luck!
Mean reversion and trend following are two opposing trading methodologies. You need to separate the trending market from the range bound market. Only then will you be able to make this better.
He might have shifted to the second day close to account for a position draw down. There were buy signal clusters and if you look at the close of first to second you have a large draw down. Look at first to third signal you have an even larger draw down. You would be stopped out with a major loss entering the first two signals
I couldn't understand the part where the position variable set equal to False and in the loop after that we use "if not position" argument which actually turns the argument to true. So in the beginning of the loop we say "if we are in a position and signal is buy....".
Sure! You get access to the Code by becoming a Tier-2 channel member here: th-cam.com/channels/87aeHqMrlR6ED0w2SVi5nw.htmljoin Looking forward to welcome you! :-)
@@Algovibes Well, I was using the Pinbar rejection indicator from TradingView, which is almost an identical plot as the Bollinger Bands middle line, and it catches that big downside move on the chart. Tanks.
Why don't you just make the stop losses part of the selling conditions? I'm having trouble understanding why you would put them in the loop? I could only understand that in my mind if they were trailin stop losses. Your thoughts are appreciated.
Is there any way to set the symbol with a variable so that i can just change the symbol on the top once and it hut run and runs everything for that new symbol
When using the panda.ta (instead of ta) for the RSI, there are only 4 buy dates instead of 10+ as in the video. I am not sure why there is the discrephencies. Any idea?
All strategies perform nice to very nice when there's a bull market. Unless you can start accounting for/predicting the market bull-bear-wise, all strategies will always fail.
![Backtesting OVER 500 STOCKS with a Trading Strategy using Object Oriented Programming [OOP]](http://i.ytimg.com/vi/R0FLsIsR4Ic/mqdefault.jpg)
![Backtesting OVER 500 STOCKS with a Trading Strategy using Object Oriented Programming [OOP]](/img/tr.png)
![Backtesting a Trading Strategy using Object Oriented Programming [OOP]](http://i.ytimg.com/vi/yTupVd6D9m8/mqdefault.jpg)
When the trading is restricted to one active trade at a time, the algorithm becomes extremely starting point sensitive. Basically you'll select which shares to buy simply based on when you set the starting date for the algorithm. I think that basically makes it random.
Would be great to see a video where you code up a way to take an algorithm like this and programmatically check starting point sensitivity.
Thanks a lot for sharing your thoughts Andre. I have already covered to tackle that "problem" by e.g. splitting in some previous videos but will also cover that in future videos.
Very elegant way to create the 'signal' column. 👍
Glad that you think so :) thanks for watching!
I think if you add sma20 like above in buy and below in sell the result will be better as it will follow trend . Thanks for sharing knowledge 🙏🙏🙏 .
whats really helpful, and tricky, is taking your videos and turning them into Classes. so tricky sometimes.
Facts. Surely a bunch of commenters start a git if his videos with python classes lol
Will do so in the next video! :-)
Awesome analysis and great to see this all done in Python, I've only really done things in Tradingview Pine Script :)
Thanks mate, happy to read :-) Be invited to check out my other stuff!
As usual... great info and very well explained... I really love your channel, it's extremely helpful
Thanks a lot Marc. Appreciate your kindness!
I swear I love this channel 😍
Thank you so much! Channel loves you too ❤️
nice, lets test it on crypto!!
Simple and effective as usual. Bravo! A humble proposal : what about to use a correlated couple of futures, such as ES and NQ and to trade their spread ? Instead of a single equity?
Thanks a ton! Also great suggestion, thank you
Great Video. You The Man!!❤🔥
Thx boss
Thanks for sharing! Nice job! I like the way you do experiments. Could you make videos on risk management of assets when, for example, they cointegrate? Or other types when we need to care about risks. Thank you and good luck!
Thanks a lot Mansur! Did some videos on that already (e.g. VaR). Is that what you are looking for?
Good vdo.
thanks buddy. appreciate your comment!
thanks for the video mate, interesting stuff!
Glad you think so Gianluca, thanks a ton!
Seems difficult to test properly and avoid overfitting though given these strategies don't do many trades
Great vid thank you...Would be cool to plot the cumulative sum of the returns to see how stable returns are and drawdowns....
Thanks mate. You can just take the cumulative prod (cumprod()) instead of the prod and then plot it.
Mean reversion and trend following are two opposing trading methodologies. You need to separate the trending market from the range bound market. Only then will you be able to make this better.
He might have shifted to the second day close to account for a position draw down. There were buy signal clusters and if you look at the close of first to second you have a large draw down. Look at first to third signal you have an even larger draw down. You would be stopped out with a major loss entering the first two signals
great video, thank you very much
Very welcome! Thanks a lot for watching and leaving a comment :-)
Very interesting as always :)
Thanks a lot Dennis 🙂
I couldn't understand the part where the position variable set equal to False and in the loop after that we use "if not position" argument which actually turns the argument to true. So in the beginning of the loop we say "if we are in a position and signal is buy....".
Hey Cem,
no in the beginning we say "if we are NOT in a position AND we have a buying signal" -> Buy!
Shared code
Sure! You get access to the Code by becoming a Tier-2 channel member here:
th-cam.com/channels/87aeHqMrlR6ED0w2SVi5nw.htmljoin
Looking forward to welcome you! :-)
Nice
I am surprised to see GME yielding a 126% return if you start the strategy from beginning of this year. I used a 5% stop loss
Yea, that's quite interesting indeed.
Ty good video
Thanks for watching mate
Maybe an MA or something to filter the larger fall at the end?
Probably not a bad idea! Thx for watching :-)
@@Algovibes Well, I was using the Pinbar rejection indicator from TradingView, which is almost an identical plot as the Bollinger Bands middle line, and it catches that big downside move on the chart. Tanks.
Thanks for this. I am using finta TA and the RSI differs from yours - which one are you using please ?
Welcome mate! Using the RSI from the library here. Has a default lookback of 14. Cheers
Why don't you just make the stop losses part of the selling conditions? I'm having trouble understanding why you would put them in the loop? I could only understand that in my mind if they were trailin stop losses.
Your thoughts are appreciated.
Very interesting 😍😍😊😘😘🤩💕💕❤️
thanks buddy
@@Algovibes💕💕❤️❤️❤️❤️🤩
Is there any way to set the symbol with a variable so that i can just change the symbol on the top once and it hut run and runs everything for that new symbol
Hey Arman,
Yes! Will exactly show how to do that in the next video :-) stay tuned
please how do you take profit
You mean like instead of the Stop Loss add a Target Profit?
Is there a python package which can read the raw data of ninjiatrader?
When using the panda.ta (instead of ta) for the RSI, there are only 4 buy dates instead of 10+ as in the video. I am not sure why there is the discrephencies. Any idea?
All strategies perform nice to very nice when there's a bull market. Unless you can start accounting for/predicting the market bull-bear-wise, all strategies will always fail.