Discussion in 'Hook Up' started by leochen2321, Nov 13, 2017.
What's the Sharpe? Does the system hit the bid/lift the offer or is it based on placing orders?
I think there are a lot of questions and answers needed for this particular system. In terms of position sizing, did you test it with a % of initial equity or a specific amount? How much is the slippage and commission? Is there a consecutive draw down? Did you check to see if it executes the trades as your designed algo to do so? What is the purpose of your system, breakout or pull back? Swing trade? Average number of days of holding position? Any robustness test? no curve fitting? Testing in different market environment?
There are a lot of work to do with this system before saying it is ready for going live in my opinion. 5 months of walk forward out of sample test is not much to consider.
There is a firm where you can make money from your algo (you keep the algo -it is your and private)
I am NOT endorsing them and NOT NOT endorsing them - it's just a resource for you to evaluate and if you can profit from it - "then bob's you uncle".
I think you get half of the 20% they charge the client (ie 10% of the returns)
Quantiacs hosts the biggest algorithmic trading competitions with investments of $2,250,000.
This is a great way to build your track record as a quant and to make money with your trading ideas.
The best three trading algorithms get $1,000,000, $750,000, and $500,000.
You pocket half of the performance fees as long your algo performs.
Similar thing at Quantopian. No reason you wouldn't use one of these services to at least show you weren't completely fabricating your data.
Also for the OP, keep in mind that it's not too hard to data mine for 5 strategies, then put $1000 in all 5 and just show us the one that works after 6 months. Not saying you did that, but that's why you're getting the skepticism.
Couldn't you at least have added a personalized note for ET, instead of just copy/pasting your FB post from a week before?
Very good point that often gets overlooked in these backtests. I could be a paper backtest millionaire doing intraday mean reversion on the 2 Year Note Futures. That equity curve goes negative when I factor in the fills.
-40% DD? That is a lot you know. So you are making 40% annualised return for a risk of 40%, that is not very good Return vs Risk ratio. In my opinion the ratio should at least be 3 to 1 for a good system.
You're comparing the largest DD over 17 years to average annual return. If you can maintain 3avgannualreturn:1maxDD over 17 years you do not have just have a "good" system, you will be very close to holy grail. I haven't seen anything like it over that long a period. If you do please consider managing money and let me in.
The characteristics you talking about will have a spectacular blow up at some future date
Robustness has its limitations
I would have closer look at system with 1:1 ratio (max 40% drawdown) and made money every year, 40% every year with number of trades above certain amount.
Software companies with their backtesting products will tell you about 3:1 ratios so you can be their customer much longer searching for that holy grail.This is pseudo science
System that made money every year with 1:1 ratio between average annual return to max drawdown of the system can have robust logic behind it,the difference is made money every year.This is very interesting proposition and i would say come in,let's talk.
Separate names with a comma.