My algo trading daily performance

DSM

Well-Known Member
#51
Thanks for your post MechTrader. Just my observation - Your algo will make money when the market is trending, and likely loose in a rangebound market. (Could you add some filter? e.g Trade only if the market moves x points and holds that? And trade lightly if the marked does not move x points in either direction?)

As the markets tend to trend 30% of time and remain sideways 70%, the wins of the days when the market tends should compensate for and make more than the days the market is rangebound. Maybe your backtest would have proved this to you. You could also check for performance after adding the filter.

Good luck.



Five continuous losses, gave back alot of profits. Hopefully things will change soon in the month of January.

Thanks,
MT
 

mechtrader

Well-Known Member
#52
Hi DSM,

Thanks for your concern and advice.
==>Yes the algo makes most money in trending markets and most likely suffers a loss in rangebound conditions.
==> I have backtested a version in which the winning percentage is more and also the number of trades is reduced to 1/3. It takes most trending trades and removes rangebound. The profit is also reduced to half.
==> I will apply that version along with this baseline system when I am ready to do multiple positions. As of now I am just running a simple baseline system with no position sizing whatsoever.
==> The reason for running this baseline version is , it alone without any filter for market regime offers a good sharpe ratio and avg profit per trade for over 5 years. Also the baseline system has passed the montecarlo simulation and out of sample backtesting.
==> Hence it makes sense to run this for a while and add the filtered version when it has given 50% ROC.

I hope I have justified your concerns.

Thanks,
MT
 
#56
sir, i have 5 questions.

1.What is one time cost, and recurring/annual cost
2.which broker do you use?
3.I have heard that if strategy is coded by symphony then Its not possible to form a strategy based on candle stick charts. I am also using amibroker and want to automate with my strategy in which i use 5 min. candle stick chart. so is it possible?
4. Do they charge live trading fee per month which is different from brokerage/commission?
5.Finally, are you using a strategy for intraday only?

Thanks in advance..
 

mechtrader

Well-Known Member
#57
sir, i have 5 questions.

1.What is one time cost, and recurring/annual cost
2.which broker do you use?
3.I have heard that if strategy is coded by symphony then Its not possible to form a strategy based on candle stick charts. I am also using amibroker and want to automate with my strategy in which i use 5 min. candle stick chart. so is it possible?
4. Do they charge live trading fee per month which is different from brokerage/commission?
5.Finally, are you using a strategy for intraday only?

Thanks in advance..
H there,
1. there is no one time cost (if you code ur strategy yourself), symphony charges are 25k for a year + vps charges 12k per year+ datafeed charges
2. Composite edge
3. Symphony codes teh strategy in presto and in that strategy one cant use historical data. I myself trigger trades from amibroker. I run amibroker on the Vps.
4. No extrafees, kust brokerage of rs 15 per executed order.
5. Yes as of now i m doing intraday only.

Thanks,
MT
 

bpr

Well-Known Member
#58
H there,
1. there is no one time cost (if you code ur strategy yourself), symphony charges are 25k for a year + vps charges 12k per year+ datafeed charges
2. Composite edge
3. Symphony codes teh strategy in presto and in that strategy one cant use historical data. I myself trigger trades from amibroker. I run amibroker on the Vps.
4. No extrafees, kust brokerage of rs 15 per executed order.
5. Yes as of now i m doing intraday only.

Thanks,
MT
Hi MT,

What is the timeframe you use? 5min/3min??

you said you use histroical data? How far back you look back before to generate a signal?