kevinkdog 

Yikes! Never intended to make this a battle, just wanted to clarify some points. Sorry to all if it seemed like a battle. I probably should not have posted this full strategy in the first place. I won't be replying anymore in this thread.

Hello kevinkdog,

Thank you for showing everyone your algo.

Question please:

Do you have any review of Build Alpha trading system building software from your experience in the system building industry? Perhaps, one of your former students from your system building course tried it and shared with you their opinion.

I appreciate the response.

Thank you.

What is the contract size per trade ? CL or the mini ?
And what was the drawdown you experienced in live trading during this year ?

Thx for sharing the strategy !

pvetrader 

What is the contract size per trade ? CL or the mini ? And what was the drawdown you experienced in live trading during this year ? Thx for sharing the strategy !

I know I said I would not post in this thread any longer, but I will answer questions about this strategy, if there is any interest...

I assume you are talking about this strategy








Crazy simple stupid...

input: daysback(65),stopl(3000),fac(1.5);


if ADX(14)>10 and close-close[daysback] crosses above 0 then buy next bar at market;
if ADX(14)>10 and close-close[daysback] crosses below 0 then sellshort next bar at market;

SetStopPosition;
SetStopLoss(stopl);


SetProfitTarget(stopl*fac);




ANSWERS:

The results shown are for 1 contract of CL, and include slippage and commissions.

In 2022, it has had a max closed trade drawdown of $7245 and a max intraday drawdown of $11895.

Quoting 

daysback(65),stopl(3000),fac(1.5);

I take it that all of the above were determined via optimization, correct ?

kevinkdog 

A better backtest is not the end goal here. But feel free to share your trailing stop code if you want (I hope it is not setdollartrailing or setperecenttrailing).

Hi Kevin,

I am learning EL and would like to clarify something. When you said "(I hope it is not ...)"

Are there reasons for not using setdollartrailing or setpercenttrailing in strategies, because for example a fixed profit target works better generally in your experience?

Another question -- on using fixed dollar amount for stop loss -- CL price fluctuated with a wide price range over the past 10+ years. Why wouldn't you use a stop loss and profit target based on a % or multiple of price, or ATR?

Thanks in advance for your information.

Regards, Bill

syswizard 

I take it that all of the above were determined via optimization, correct ?

No optimization at all, just a guess based on my experience. I like longer lookback periods (65 bars is roughly 3 months), $3K is a big stop but smaller stops usually get hit too much, and I like the feeling of profits being bigger than losses, hence the 1.5.

All performance for the last 8 months is after development.

Not for everyone for sure (you mentioned earlier you were shocked at my lookback length and reward/risk).

Billiwon 

Hi Kevin, I am learning EL and would like to clarify something. When you said "(I hope it is not ...)" Are there reasons for not using setdollartrailing or setpercenttrailing in strategies, because for example a fixed profit target works better generally in your experience? Another question -- on using fixed dollar amount for stop loss -- CL price fluctuated with a wide price range over the past 10+ years. Why wouldn't you use a stop loss and profit target based on a % or multiple of price, or ATR? Thanks in advance for your information. Regards, Bill

1. SetDollartrailing and SetPercentTrailing are only accurate with tick data (as part of Look Inside Bar Backtesting), so if you run backtest for more than 6 months (the TS limit for tick data history), your results will be completely and over optimistically bogus.

2. I have done a ton of research on the subject, you can find it if you look (I can't self promote and tell you where exactly to find it, but it is out there). Everyone says ATRs are the way to go. I do use them sometimes.

But Dollar values work better many times. Why? My guess is many traders think in terms of "I can lose $1500 on this trade" and not "I can lose 2.3 ATRs on this trade."

That is just my theory though.

@kevinkdog

Thanks for the idea! Appreciate any useful info shared.

syswizard 

I take it that all of the above were determined via optimization, correct ?

I forgot I said this on Page 2 regarding optimization:

I did not do optimization of this, all values were a "best guess." Initial data was 02/2010 to 06/2019.

From June 2019- Jan 2022 (2.5 years was out of sample).

Jan 2022- present is all live, real time performance (as calculated by backtest engine).

Yes, I have been able to put together a small puzzle trading robot; however, it is for the ES instrument.

A trading robot is just simple math and if-then logic. The trading robot does its calculation in a nanosecond, without making a mistake, or getting distracted. The robot does exactly what I teach it to do and it can even make the trades.

However, humans are better at exercising judgment and understanding a situation of when and when not to trade.

An example of not trading would be an FOMC announcement when liquidity dries up on the DOM. Another example would be the pre-market reports at 07:30 central time. In other cases, the rational thing is to crank up the volume when things are working. The human's job is to set the parameters based on the market conditions and their judgment about how we should trade each day, each half hour, or perhaps each 5-minutes.

In my opinion, the best trading systems use a human and a robot(s).