Developed a very good algo... Now what?

[Adi Ben Israel]

Hi everyone. I'm looking for an advice:

1. I developed a very good algo (adiben83's Profile | Myfxbook).
2. It's running on a live account since June 2015, with relatively low levrerage, and has a pre-tax 68% return on 15% max draw-down.
3. I have a detailed presentation for potential investors & statements of course. Sent it to few hedge funs, but got no reply.
4. My algo system won't yield 500% per month ......... more likely 50% per years (on a long term average).
5. My account is ~$15,000, so I need funding. I don't know what's the best way to implement the potential of the system.
6. Any tips?

Thanks

 

[nishank gupta]

Check if it's scalable.

 

[Adi Ben Israel]

meaning...?

 

[daleholborow]

Why do you pick arbitrary year ranges when calculating and presenting performance? It seems that your backtest in particular, has several selected date ranges, but no obvious reason why they are broken up?

 

[Adi Ben Israel]

The reason is simple: I wanted to back-test the system as far back as I can go. I had tick-based data from 2004 (using tickstory) but could not run a back-test for such a long period because of MT4 limits (HFT files as you probably know...). So I could run a back-test for ~4 years at a time.
I had other tests on different time periods (again, 4 years for every test) and got similar results). I will gladly send you results for every time period you'd like.
Besides, I think statements from the live account are much more important than back-testing, and I think my statements speak for themselves. I'm attaching the files to this reply. Please take a look

 

[nishank gupta]

The reason is simple: I wanted to back-test the system as far back as I can go. I had tick-based data from 2004 (using tickstory) but could not run a back-test for such a long period because of MT4 limits (HFT files as you probably know...). So I could run a back-test for ~4 years at a time.
I had other tests on different time periods (again, 4 years for every test) and got similar results). I will gladly send you results for every time period you'd like.
Besides, I think statements from the live account are much more important than back-testing, and I think my statements speak for themselves. I'm attaching the files to this reply. Please take a look

Wow I did not know I could get tick data for free. Thanks for the source!.
By scalable I mean do you think the algorithm will do well when trading with say $100k, $1M, or $10M?
That said, I think your best chance would be to get funds from friends and family and start trading with a larger amount. Some trading strategies do not have a large half life.
Try elitetrader.com forums as well. Someone there might tell you how to proceed.

 

[Adi Ben Israel]

Tanks gupta! Yes, you can use tickstory for tickdata (it costs about $35, they get the data from Dukascopy).
I'm sure my algo can handle larger amounts. Most of my trades occure during "rush hours" of the trading day. It barely trades during asian time where volumes are low (you can see it in myfxbook's profile - solidP System by adiben83 | Myfxbook).
Unfortunately, family & friends will not bring me to a $1 million account... that's why I'm looking for other funding sources or collaborations. If you guys think of something please let me know.
TX

 

[tfors]

If you don't have (rich) friends now willing to invest, maybe start making friends, such as at meetup events maybe. Will not always work of course.

 

[yetanotherquant]

Well, first of all the (productive) backtest period is too small (as far as I know a rule of thumb is at least 3 year). It can be partially mitigated if your system trades frequently (by the law of large numbers it is then proven that your success is not an accident, however, this success may still depend on market regime).

Secondly, I had only a quick look at your presentation and it did not catch my eye.
You need more visualization.
Have a (critical) look how I analyzed and visualized the performance statistics of this giant of trading and of humble myself.

And thirdly, don't give up if several hedge funds don't reply to you.
I, myself, have little experience with hedge fund (there are not so many of them in Germany) but I have seen enough mutual funds: there are (too) many mediocres that merely cannot recognized good ideas. That's why you need, in a sense, to find a gem in a ton of bullshit, so to say. That's why try-try-try ... and not just send emails but seek for personal contacts.

 

[Adi Ben Israel]

Well, first of all the (productive) backtest period is too small (as far as I know a rule of thumb is at least 3 year). It can be partially mitigated if your system trades frequently (by the law of large numbers it is then proven that your success is not an accident, however, this success may still depend on market regime).

Secondly, I had only a quick look at your presentation and it did not catch my eye.
You need more visualization.
Have a (critical) look how I analyzed and visualized the performance statistics of this giant of trading and of humble myself.

And thirdly, don't give up if several hedge funds don't reply to you.
I, myself, have little experience with hedge fund (there are not so many of them in Germany) but I have seen enough mutual funds: there are (too) many mediocres that merely cannot recognized good ideas. That's why you need, in a sense, to find a gem in a ton of bullshit, so to say. That's why try-try-try ... and not just send emails but seek for personal contacts.

Thanks a lot!

 

[Andy Nguyen]

Quantnet is NOT the appropriate place to find funding for your trading.

Besides that, to avoid any legal issue with SEC, you are required to pass a Series 7 license when you solicit money from others. Even then, there are things that you are not allowed to do.

SEC approves FINRA rule requiring registration of algorithmic trading developers | Regulation and compliance | Bloomberg for Enterprise
Prohibited Activities - Series 7 | Investopedia

Good advice was given so it's time to close the discussion.