Interview with Patrick Nouvion
How did you come to trade and when? As far as I know, you have developed a multitude of technical indicators and EAs for MetaStock and MetaTrader 3/4. Tell me, please, from what you have began? Which of your indicators and expert advisors do you consider to be the best ones? You have to understand that I believe everyone should create their own system so that they fully understand it and can adjust it... I think even if I had created the "holy grail" and sent it to people they would come back and tell me: It sucks. So my goal is more about providing tools, training and support to people and not about writing profitable experts. You know both development environments: MetaStock and MetaTrader 4. Which of them would you call the best one? I think it is a great idea. Most brokers had their in-house contests, but this really makes it the "World Championship", so it is more appealing. Not just because of the prize money, but also for the recognition one would get by winning the Championship. Also it would help to advertise the automated trading feature present in MetaTrader. I was in Vegas for the Forex Traders Expo last week or so, and still a lot of people came to me and asked "What is automated trading?" and once I told them they still could not believe it was fully automated. A lot of gurus insist on execution and how it has to be fast, mechanical. They insist on how you have to respect your own rules, well an automated expert is the perfect solution for all this, but not enough people realize it is available yet. If you were not a staff member of the Championship Sponsor, would you participate? Of course, I would. Like I said before, beyond the money, there is also the aspect of recognition. I don't know whether I can write a better expert than the next guy, I don't know who else out there writes experts. To be able to tell myself "well, you finished in the top 10" would be a great validation. On the opposite side, if I finished last, then I would just reconsider my career path. On what principle would you create your expert advisor? Or, say it in another way, what expert advisor will win, in your opinion? If I actually had a useful answer I doubt I would tell people, but in my opinion it will be about the market. Is the market trending or not? Does your system look for that and adjust accordingly? For example: if(Market is in Trend) { Use MACD system } else { use RSI system }So your system would be classical trend/flat EA?Well, classical ... the tools are the same for everyone, how you manage your money, how you size your positions, how you filter your trades (Days, Time Range etc.) is what makes the expert good or not. To me, spending time figuring out if you need to use an RSI of 14 periods versus an RSI of 20 periods is a waste of time. So, yes, I guess a classical trend/flat EA that is adaptively combined with some other simple filters. What would you wish the Participants of the Championship? I wish everyone good luck on the Championship and in their trading activities. Thank you Patrick. Created: 2006.09.22 Author: MetaQuotes
Pages:
1 I get different results depending on when I run the backtest, on the same computer
with the same data and the same server.
I have asked this question before and received no answer, so here it goes again. .. Does the tester use the live spread?, if not, how is spread incoperated into backtesting? These seem like critical questions to me when backtesting on small timeframes. 2006.09.28 00:50
TR1 wrote: Hi TR1, thanks for volunteering your help. I can be contacted at trohoang@yahoo.com. I am pretty sure that we are not the only ones with profitable EAs in backtesting;
hence, there are many others with similar issue. Like yourself, I use live forward
testing with backtesting to determine whether or not the EAs are coded properly;
however, backtesting is essential to determine whether or not the strategies work
in bigger time frame. One can't use limited forward testing to trade and ignore
backtesting because the backtesting functionality is not working properly. But your results are surprising, and I would very much like to work with you on determining the cause, as if it is a problem with recent builds then I would really like to confirm that. If you're interested, please PM or email me via my strategybuilderfx. com profile if you're a member there, or post here if there's another way to contact you. Renat wrote:
No need to show simple tables. Just open a new thread in forum and publish all details
(charts, final stats, etc.) about your results. I have started a thread in the forum here:http://www.metatrader4.com/forum/2381/ Trohoang, I've found MT's backtesting to be quite accurate provided one works within
the limitations as well as the bugs, which are much fewer in MT4. I really only
use forward testing with backtest comparison to confirm that I can trust the backtests.
And if an expert can't be backtested (due to certain indicators not working in
backtest, or being sensitive to intra-1M-bar price movements), I just don't bother
with it, as too many people
see great results in a limited forward test only to go live and discover that the great results were only a good period for an overall losing system! You might be interested in my messy backtesting analysis at http://www.strategybuilderfx.com/forums/showthread.php?t=15467&page=8&pp=8 and http://www.strategybuilderfx.com/forums/showthread.php?p=146009#post146009. But your results are surprising, and I would very much like to work with you on determining the cause, as if it is a problem with recent builds then I would really like to confirm that. If you're interested, please PM or email me via my strategybuilderfx. com profile if you're a member there, or post here if there's another way to contact you. trohoang wrote: No need to show simple tables. Just open a new thread in forum and publish all details
(charts, final stats, etc.) about your results.stringo wrote: trohoang, start tester in visual mode and see difference. There is only one explanation - different data give different results Stringo, I am quite confidence in the data as I have exported the data onto a 4th computer and got the following results: 2006.09.26 13:33
stringo wrote: trohoang, start tester in visual mode and see difference. There is only one explanation - different data give different results Stringo, I am quite confidence in the data as I have exported the data onto a 4th computer and got the following results: Computer A Computer B Computer C Computer D EA1 very profitable No good very profitable Very Profitable EA2 very profitable very profitable No good Very Profitable EA3 very profitable No good No good Very Profitable EA4 very profitable very profitable No good Very profitable EA5 very profitable very profitable No good Very profitable EA6 very profitable No good very profitable No good From the above tests, 4 out of 6 EAs are profitable on 3 computers and only 2 EAs are only profitable on 2 computers. I am now certain that Strategic Tester is not stable. Have you guys ran across any profitable EAs that you can test on multiple computers? BTW, most of my EAs are on the 1hour chart; hence, the 1min data should make the strategic tester quite stable. trohoang, start tester in visual mode and see difference. There is only one explanation
- different data give different results
2006.09.26 10:12
Lenar wrote: Hi Lenar,Hi Trohoang, you have to provide absolutely equal data. If you get it from one server,
it doesnt mean that its the same data. Please read: 'Testing of Expert Advisors in the MetaTrader 4 Client Terminal: An Outward Glance' Hope it will clear some of your questions. Thank you for your reply. All three computers were imported with the same historical data 1M, 5M, 15M, 30M, 1Hr, 4Hr; hence, they should give similar results. Note that I am only testing it from April 04/05 to July 28/06 (not the present data); hence, the server should not come into play. Am I missing something as the strategy tester is not giving the same result? Patrick wrote:
I don't know for sure, I would have Metaquote's people answer this question, but to me it seems to me that the strategy tester access a different "temporary" data file depending on wether or not you are connected to your broker's server. Therefore you get one or the other but the results on each data set remains the same even though it is a different result for both data file. I have not had time, nor did I really ever care to research this any further sorry :) Hi Trohoang, you have to provide absolutely equal data. If you get it from one server, it doesnt mean that its the same data. Please read: 'Testing of Expert Advisors in the MetaTrader 4 Client Terminal: An Outward Glance' Hope it will clear some of your questions. 2006.09.25 12:14
|
|





I get different results depending on when I run the backtest, on the same computer with the same data and the same server.
I have asked this question before and received no answer, so here it goes again. ..
Does the tester use the live spread?, if not, how is spread incoperated into backtesting?
These seem like critical questions to me when backtesting on small timeframes.
Backtesting seems to be a big issue and eveyone is having the same problem. It seems that it is not even on Metaquotes' agenda to address the problem. I am at the stage where I am beginning to develop my own backtesting platform. I have a couple of theories why backtesting issues are not being addressed but I won't go there...