eSignal wish list : the dedicated thread

The Doc's are a bit light, but getting better. And examples to abound.

WRT slowness, I have found that most formula's load fairly quickly, but I did have a few that loaded REALLY slowly. I started
a dialog with Matt and found out some details of formula's calling other formula's that I didn't understand before. As it ends up, everytime your formula calls another one, the ES Engine interates all bars on the called formula and stores the results. So, I had
formulas that did the equivilant of:

for (i=0;i< 10;i++){
call ("/formula/path/form.efs", i);
}

Which would cause 10 different formulas to spawn each which would interate all bars on the chart, and then store that information for access latter. Needless to say if you were doing what I was and really calling the formula each time with unique paramter then this would get totally out of hand. Accually now that I know this, I am amazed they loaded as fast as they do.

During our conversation, Matt and I came up with two ideas on how to do what I wanted to do (have modular .efs files that I can call from many other formulas and not cause this huge overhead). Called formula's will continue to work as they do now, because in many cases this is the best way to do it, but there will be new ways.

Anyways, this is a long winded way of saying - watch out for formula's that call other formula's. This is the only thing I have seen so far that causes any real delays in loading.

Quotesheets - I'm not at liberty to say, but just wait. It will be very cool.

DDE - I agree. I know they have thought of this. Since you can call your own .dll's you can do this now if you code it yourself, but I too would like to see this as a standard part of ES. Not sure if or when...

Auto upgrade checking is nice if done correctly...not sure if the ES code can support the concept of just updating modules which is what would really make this a valuable feature. If it has to be an entire update, this isn't as important.

G
 
Hello Garth, I trust Matt for doing a pretty good job regarding Quotesheets. Indeed we have all seen what he can do with Ravenquote so I expect the same or even better in future eSignal.
 
1) 2 years of historical 1-minute data
2) indicator and custom -based alerts ( with ability to specify sound file as an alert )
 
to see formulas or studies highlighted or colorcoded in the quotesheet for each stock would be great. Hopefully it will be like in stockwatch pro or better. And i agree with others regarding hotlists. They really need to be improved, would be nice to have hotlists like in qcharts and updating time of 5 minutes is to much, 1 or 2 minutes would be better.
 
Greetings everyone.

This is Matt Gundersen of eSignal. It is very exciting to see this thread created by eSignal users for eSignal users. I have subscribed to this thread so I can listen to your requests (and occassionally respond) and learn how we can continue to improve eSignal for your trading.

I've browsed through this forum (all 4 pages right now). I can only imagine it's going to get bigger... Thought I would provide a general overview of what's coming down the pipeline...

In Advanced Charting, we have some new features coming out. Data Export is one of them. This will allow you to export the OHLC data and any study data from your advanced chart. If you have a chart with Price, Stochastic, and MACD, all the data is exported. Additionally, if you choose to export in HTML format, or to Excel, the color coding is retained. Tab Separated Ascii file is also available for export.

For those of you familiar with Advanced GET's Stack/Tile studies feature, this has been carried over into advanced charting.

Did someone say Tick Charts???? Yes, Tick Charts are coming. We'll have raw tick mode, N-Tick bars (you want 15 ticks per bar, you got it!), And N-Second-Tick bars. You want 15 second bars, no problem. These are schedule for July. There, I said it, now I'm committed to it.... :eek:

BackTesting is coming. It will also be available in July. We've taken what I think is an interesting approach to backtesting. The output of the backtester is XML. This means, for those of you that want to fiddle around (maybe if XL supports importing XML), you can bring the output of the backtester into a spreadsheet and play around, do all kinds of analyis (perhaps including Monte-Carlo in XL).

However, we realize there will be those that don't want to import a XML file and just want the facts! Tell me how profitable was my backtest, how many winners? Loosers? We will have a very nice reporting engine/viewer. The report will be generated in HTML. This means if you wanted to post a backtest report to a website to show your friends or educate others, no work for you. Just upload the report and it's webready.

We're writing some docs up over the next week on the backtester. I'll try to post a word doc to this forum so everyone can get a pre-release peak. I'll also try to post a sample of the HTML backtesting report.

On the subject of trading. You guys cranked up the heat and we listened. We've got two Broker's scheduled for integrated trading. They will be available in July. CyberCorp and IB.

There was a big ol' post by Garth on this forum where he mentioned somethings about the EFS engine. In response to that: (NOT AVAILABLE YET), a new function was added to EFS and is in testing. callFunction("Filename.efs", "FunctionName", [arguments]); Think of this as a cross between #include and call (without the overhead of "call"). We're working out the final details now and will document this.

We're almost done. One more thing, EFS performance stuff...

One of the things we heard about EFS was the OHLC formulas (the ones that show Today's OHL or the Previous OHLC), Pivots, and the MSH-MSL formulas were slow. Yes, they WERE! They will be faster in this next release (think I saw Chuck Thompson post that it will be out this week). :eek:

Techie Talk: What we discovered... All those formulas did a lot of work with dates. Which mean "Date objects" were getting created/instantiated/constructed. Date objects have a lot of performance overhead. We added a thing of two to EFS to improve the speed and re-wrote the OHLC, Pivot, and MSH-MSL formulas. They will be included in the next release and run MUCH FASTER.

That's it for now.
Matt Gundersen.
 
Hotlists is a great idea, and one could imagine that is "quotesheets" are done right, hotlists would be easy to not only create youself, but if esigal could easily provide prebuilt ones for common lists.

And I totally agree with more historical data for intraday data. There are two issues with historical data from the user point of view. One, there has to be enough of it to provide some sort of statistical reliability for back testing trading systems, really a few hundred bars if I am testing accross multiple timeframes and symbols (so several thousand bars in total). Right now at higher timeframes (like 60 minutes), there just isn't that kind of data.

The second aspect of historical intraday data is to provide enough data that you can test against different market types. This is harder, as a bull or bear market can easily last multiple years.

Since MANY people seem to feel that more historical data is important, I would suggest that someone here start a thread on the esignal yahoo group requesting this once again, and asking other members who are interested in seeing more data respond to the post affirming this need.

Maybe they will listen if they see how important this is.

G
 
Hey Matt,

>There was a big ol' post by Garth on this forum

You saying I run off at the mouth (uh...keyboard)?

Not that I would disagree, but just wanted to check ;-)

G
 
It could be interesting to have the option to write our plug-in or add-in for eSignal as for RavenQuote and see the result of them inside a window of eSignal as new graph style (@D - 3D) that we could write for our purpose.

Also, I love the plugin style that you have wrote for quotesheet in RavenQuote, hope to see them in the future!
 
Garth,

Scott J from eSignal told me they had been collecting datas beyond their 60 days actual limit since Jan 2002. As he said in his mail from end of May, they do need to program some changes in their tick servers to index the additional data. They hope to do it by the end of summer at which point we will have 9 to 10 months of intraday data and growing...Their target is to provide at least 1 year of intraday data.

Does someone know how long is the qfeed intraday history ? I guess more than 2 years...?
 
I think qcharts keeps it back to 85 or so.

One year of data is enough to help traders with issue #1 in my list, and is what I would consider a bare minimum to support real backtesting for intraday data.

I think 5-8 years is the minumum that is needed for #2.

And of course, more is always better.

G
 
You must log in or register to reply here.
Back
Top