Questions:
6:34 AM
Ovia
Following my post on The Grailtrading Blog i've been asked some questions. I'd rather post the answers here if I may because i want to leave that blog as a bit of history now :)
15 Aug 07, 02:31
jay: excellent lessons to have passed on soul - can you divulge any more of the legal stuff now? Cheers JKAfraid not JK, all sorted now anyways and all parties happy.
Sgguy: It's been a real pleasure following your blog Soul......here's wishing you the best in your future.
Thanks very much SG Guy, stuff will continue over in this blog
Now there is a great question, allow me to share my thoughts with you The answer to that is kinda No. the first tests we did had to be conducted on an MT3 platform and even today that gives us the most accurate results. Eventually we managed to get the MT4 backtester working well enough to trust the results, but it's only been that way since around November 2006. even then it had to be a specific broker (Mig) and specific backtest data (alpari data dump). this was because the alpari timeframe matched that of Mig and that Mig is the only broker i would trust to deliver accurate real time data. So many of the others have many missing bars and ticks - just look at a one minute chart on most brokers and see the missing bars - this affects indicators an amazing amount. Mig has the least number of missing bars consistently. Then the programming had to be perfect. im talking about us making sure each line of code was needed and worked right - it must have taken us 3 months just to get the expert working good enough to backtest and at that we had professional programmers on board. eventually yes, the MT4 backtester was reasonably accurate - well, but that i mean within about 5% of reality. of course we had to verify this by eye to confirm before we could do any simulation tests - we eye tested every single trade for four years of data and can confirm that mig ran with alpari 1 min back data is accurate to about 95% - but note, the 5% is quite a difference, and it's always 5% worse in real life, not better :) Suggestion to those looking into backtesting would be to buy the new book The Way of the Turtle: The Secret Methods That Turned Ordinary People into Legendary Traders  It has an excellent part about backtesting and the effect of randomness on a trading system - made me open my eyes i can tell you.The rest of the book is excellent also. |
19 Aug 07, 22:59 dan: soul its great the way you are trying to help people understand what went wrong with the grail are you working on anything new thats looks promising best regardsHi Dan, thanks very much for the comments, just hope it's useful to you. Yes, good things happening right now and very happy. Grail took some time to get over emotionally and it took about 6 months. Recently been working on a system which is really great, getting back into my trend following roots and doing really well. That strategy netted me over 4000 pips last week! That was of course not the norm, was because of the sub prime thing, but it was nice to break personal records of 1 week and 1 day closes. My previous best was in December 2004 when i closed 485 pips in a single day and did 1890 pips in a week. This time is closed over 3300 pips in a single day and almost 4500 in a single week. More than the grail system made over two years in a single week! I just wish that i had been in the position i was in 8 months ago where I was trading 125 lots of client money and £120 per pip of my own - now that would've been some party! The system is robust now and is producing better returns than grail at half the previous leverage and I've recently setup an agreement with a broker to handle managed accounts myself rather than through the company I worked with before. My first client with $70,000 will be depositing in the next 2 weeks so all looking good to go. I'm also trading on Zulutrade.com properley now so you can watch the performance over there - just look for the soultrader account. Thanks everyone for your thoughts, keep the comments coming |
0 Response to "Questions:"
Post a Comment