holy grail system?

[shooly76]

it has taken over 4 years to build a highly profitable (potentially) system. whether it be discretionary, mechanical, discretionary-mechanical, or automated. so this is what I finally came up with..and it completely blew my mind when I saw the results. Ive never seen anything like these numbers-anywhere! I consider this to be a black swan event..well at least for me it is.

this is an automated intraday HFT strategy..so no overnight positions. it makes 214.64 trades/ day (avg)..so obviously low commission rates are a must. there is actually another version of this system that netted $1,310,830 in ONLY 21 months, 1 contract! but it makes 499 trades per day! so it is almost impractical due to the high commission costs. I also ran it on 17 instruments..the only instrument that lost money was ES. top performers GC, CL, TF, NG. that netted $22,177,848!!! I will trade it if I can find total commissions and fees under $3 or less round trip.

okay..now back to this system that trades TF only..

backtest period: 21 months
# of trades: 72007
walk forward optimization period: 28 days
best # results:10
# of contracts:1
instrument: TF
max DD: $617.03
percent profitable: 52.52%
profit factor: 1.99
avg W/L ratio: 1.80
total net profit prior to optimization: $695,350
total net profit after walk forward optimization: $757,350

I will sim this for one week before trading it live. I know this seems meaningless but due to the high trade # 72007, and backtesting, optimizing and walk forward testing that it should perform very closely to the results produced from testing. not to mention it works on 16 instruments (tho, much better for faster heavy volume markets). and when I ran backtest on that version of the system, total # of trades was 8,457,599!! Van Tharp only suggests only 10,000 trades and good backtest results to get a solid idea of future system performance. therefore I dont think sim trading this strategy for an extended period of time would be much different from the extensive backtest results.

I would post an actual summary of results, but I cant save the summary screenshot in strategy analyzer..and I really dont want to disclose all of the info on the summary.

[sledge]

How does it fair in market replay?

[shooly76]

thanks..I knew I was forgetting something important..I have not run on MR. but now that you mention it, I need to do that. I will post back here after testing on MR..

[sledge]

thanks..I knew I was forgetting something important..I have not run on MR. but now that you mention it, I need to do that. I will post back here after testing on MR..

Depending on your RAM and CPU speed, market replay might bog down massively with that many trades per day. You might have to run a month at a time or something.

Same with SIM or LIVE. You'll probably be restarting everyday, which isn't a problem with your system parameters of no overnight.

[shooly76]

thanks sledge- when you say restart strategy daily..you simply mean 'enable/disable', correct? just want to make sure we are on same page.

I still have not run this on replay..would like to do that today at some point. but like you said.. tons of trades to go through. I have 8GB memory and 2.70 processor (would like to have more!)..but still should be suitable at 1-2 months at a time.

I ran another backtest and monte carlo simulation. the above posted results were actually post- walk forward testing so it is actually 22 months.

so here are some different results from another 21 month backtest prior to walk forward:

total net profit: $689,680
max DD: $1,040
#of trades: 66142
percent profitable: %52.52
101.44 trades/ day (avg)

[shooly76]

forgot to add above..that image is of the monte carlo simulation

[Harry]

What bar type (minute, tick, volume, etc.) did you use for your backtest?

Did you account for slippage?

[shooly76]

after 1 tick for slippage and a few other 'minor' parameter changes.. strategy still came out nicely.. $105,000 avg for 21 months for 2000 monte carlo sims (at 50% axis) ..I knew that it would be way lower net profit after slippage..but still nice system..just wish there was a way to calculate historical and potential forward slippage..if there is..please let me know how..I have CQG data feed..never noticed too much while using chart trader..but way more trades per day..different sport altogether..edit..meant automated way more trades/day..

[shooly76]

actually..the above results are after a major adjustment..much less net profit, but way more tradeable for my small account.

today, TF was the absolute worst day to trade an HFT system like above..Im seeing HFT is only feasible w certain variables..like..very large capital!! haha..

even still much fine tuning is still needed..to acount for slippage..I tested the 'same' above strat w different candles.. just saying I did notuse charts that had a TIME increment on them..those only yield bad results. but I already knew that from long ago...just saying..minute charts= negative net profit for this style of system..

also.. abig reason for fine tuning this strat today is to 'teach' it to avoid FLAT choppy TF markets like today..

[shooly76]

also..seeing now that slippage is more of a concern that commissions..I thought that would play a huge part..but still not what I was expecting..

but this system is still nice even after 1 tick slippage..I (hopefully) doubt w CQG there will be any more than that.

Im thinking it may work even more efficiently than TF ...on other instruments such as GC and NG

may not make as much..but drawdown may be lower..need to test those markets again..

Im looking for efficiency..and lower trade numbers..and clearly daily DD is important in this build..

very exciting tho..I feel like my work and time is starting to pay off somewhat..THANK YOU NINJA STAFF!!

[sledge]

also.. abig reason for fine tuning this strat today is to 'teach' it to avoid FLAT choppy TF markets like today..

So it did not score big after 1:30-2:00pm EST?

That was NOT choppy anymore!

[shooly76]

I was thinking of running 20,000 monte carlo sims?

....but I figured 2000 is plenty for the trade #

any input much appreciated..

[shooly76]

yeah..I know..it put on points after I disabled it..but it lost $3600 today in sim mode..I can see why from looking at charts today..

but that is why Im making changes to it now..but the changes I made will hopefully (edit) help to avoid the flatness of today..

[shooly76]

but its weird..the only reason I say 'teach' it is because of the crazy equity curve up..evn after 1 tick for slippage.

its like the equity line keeps going higher..almost as if the system likes going higher..like its adapting to the historical data and codes..??

[shooly76]

actually..it was still pretty flat all day..there were only like 2 40+ tick moves that my strat caught today..but I disabled it at around 3pm EST