Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
Hey Treydog I think I might know why you are unable to recreate the walk forward results with time stamp flags. Multicharts WFO breaks up the data into insample and out of sample data segments. When it runs the strategy on the out of sample segment it does not have the data from the insample data set. Ex. you use the 20 day moving average your strategy will need to wait 20 days before it can start trading in the out of sample data set. When you use time stamp flags your strategy has access to all the data so it does not need to wait 20 days. I mentioned this flaw design to Multicharts on their forums.
Yes this is part of the problem I have been addressing with multicharts. However if give it the entire data set and use the time and date stamps of the trades in that period since you have the lead time it should be OK. i also have done it out of sample session by session and you are able to replicate the correct number of trades if you give it the extra lead time. However if you use commissions and slippage its still incorrect. So if you go window by window + lead time you can get very close to the WFO answer, though not exactly. I thought if you use the date flags, on the entire set the preload time would be built in so it should be ok. Thats not exactly the case but I haven't investigated it further until i can replicate the WFO results exactly using commissions slippage window by window. Honestly i work on this almost every day with these guys at support. I just want to get back to strategy development, its a huge hinderance and annoyance but what else can i do? Anyway they were supposed to get back to me about that yesterday, maybe i will get something today. when i do i will update.
So I got an update from Andrew. Basically its confirmed that adding commissions and slippage does not allow you to replicate the WFO. Although we had some minor success with no commissions and slippage. He told me their engineers are working the new beta and that they don't have time for my problem. Which is really not just my problem but every multicharts users problem. I am really getting frustrated with the amount of time I have to go back and forth with them to prove that I found some severe flaw. Not to mention the fact that they keep sending me fixes that really don't fix the entire problem. So now I am on the back burner again, I have no idea when they will get back to me with an answer. This is really holding up my development process. The worst part is they asked me to take down this thread, or say that the problem was fixed.
XXXXXXX, I saw you opened a respective thread on Big Mike’s Forum. Since we discussed the Walk-Forward Optimization and steps to reproduce it manually in MC, would you be able to comment that there is no issues with that? Thank you.
Best Regards,
Andrew Shuvlenov
Customer Care Specialist
1-888-340-6572
Ashuvlenov@MultiCharts.com
6:00 am - 4:00 pm EST
No the problem is not fixed, you are able to reproduce it on your end so no I will not comment that there are no issues when there blatantly are issues.
IMHO this is very important issue. I'm on the step to WFO my strat which uses timestamps. What should i do ? take a break for a couple of monts/years ?
This is absolutely a NO GO. We have to force the support of MC to resolve this issue. In my opinion the new beta is not such important like WFO problem on an existing version.
Actually, the new beta is probably the most important beta/version released in the last couple of years.
MC 8.7 introduces millisecond data and this forced the database to suffer a major upgrade.
So you basically have now a complete new millisecond database that needs to have backward compatibility with the previous time keywords/functions and new ones were built to take advantage of the millisecond data.
Also, a couple of other tools will see significant changes due to the new database which will enrich MC's experience.
I'm sure this was not an easy task and most of MC resources must have been on this the last few months.
If I become half a percent smarter each year, I'll be a genius by the time I die
I am not denying that an upgrade with great features is going to be good. But why not fix a feature that has been claimed to be working for several releases? I mean its a bedrock feature they advertise on their website. Basically everyone who owns multicharts has a broken piece of software, or at least one that is falsely advertising it has the capability to do something it really can't. I may be emotionally charged here because its holding up my own progress, as well as how annoying it is to work with them. So I am just going to stop here. Either way, I want this thing fixed.
That sounds a little bit bold, why do you think that? (also related to my question below)
Where did you find that? Don't get me wrong, I'm not saying that I don't believe you , but I like to know what things we can expect and why this will be the most important version in the last couple of years. If I look in the Project Management I see four features and one bug fix for MC 8.7 Beta 1.
I agree with Treydog999 here, since I don't consider it a very valid reason that they were busy with new features. Of course, new features are great and I look forward to these as much as any other user, but introducing new features can not be (in my view) an excuse to not fix the existing product.
Fixing bugs should always take precedence since we can easily wait two more months before a new feature is introduced, but, as others have already said in this thread, we can not wait two months before we can perform a reliable WFO.
Since HFT and algorithmic trading is the hottest thing around these days, I thought "algo" guys would love the idea of having milliseconds available to them, giving the extra granularity to program their strategies, hence - the most important version release in the last years.
Sorry for that, apparently my assumption was wrong
A good friend of mine always told me that my assumptions will be the death of me
Please don't misinterpret me, bugs should have always precedence but there are bugs, and I have no idea if this is such a case, but I assume (again, I'm assuming) it's not since several fixes have been made available and none solved the issue, that will be easily fixed with some new feature that will be released in the next version.
So if some bug will be easily fix when that new feature becomes available, why lose time trying to fix something now if in a couple of months it might stop working again since with that new feature, the way that tool works will change again?
I share treydog pain, heck, we all share his pain, we all had our share of working with something that was only half working and had to wait for the next version to see the problem solved but the only thing I could do was nagging support about it, so they could move faster in solving the problem.
I hope I'm not putting words in his mouth but I believe treydog said he was losing his time testing the fixes MC sends to him.
He discovered the bug. The tool is not working accordingly to his settings and knowledge.
Apparently, developers thought the tool was working fine, accordingly to their knowledge.
Assuming treydog is more knowledged on the matter, since he's a trader, he's the most qualified to be testing this, telling developers that the tool continues to be faulty, which he is.
Helping MC developers with this treydog is helping himself on the long haul because he will have a tool set accordingly to his settings.
Personally I consider this to be a very useful and friendly relationship between developer and final user. Both helping each other in building a better platform.
If I become half a percent smarter each year, I'll be a genius by the time I die
My biggest complaint with Multicharts is their half-finished portfolio backtester. I already have given up on their walk forward optimization. Single symbol walk forward optimization is pointless for me since I backtest on my strategies on many symbols to ensure robustness. Besides no walk forward optimization there is no bar magnifier making the execution model less accurate. Worst you can’t even trade the portfolio strategy without having open up an individual chart for each symbol. This is very time consuming and greatly limits the number of symbols and strategies you can trade.
Each new version of Multicharts introduces both bug fixes and new bugs. I feel like you’re stuck between a rock and a hard place. Do you have to continue to suffer from existing bugs or upgrade and deal with the new bugs?
At this point, I’m waiting for either Multicharts 9.0 with the feature to trade multiple instruments from chart or Deltix DCS Trading. I thought version 8.5 would be the last release before 9.0. Now there is a Multicharts 8.7 beta…
(also related to my question below)
