RealTest - Portfolio Level backtesting software

[Roller_1]

RealTest is a new portfolio level backtesting tool that has been developed by Market Wizard Marsten Parker. His website can be found here RealTest (mhptrading.com)

There is plenty of info on the site about it features and some tips and tricks, plus a few youtube clips (He's doing more soon).

The software has been released this year and in my opinion blows Amibroker out the water in many aspects, although i still use AB for some strategies. The language is easy to learn and if you have used AB or tradestation it is very quick to pickup.

The powerful thing about RealTest (RT) imo is the control you can get over the strategies you build. Its like being proficient at AB CBT (which i'm not) but in just a few lines. Another powerful feature is the ability to test multiple strategies in one script and accurately control how funds are allocated between them, position size, Multiple time frames etc. It can use Norgate Data so tests are accurate, can easily use yahoo data too if you're happy with that or just beginning.

If you trade shorter EOD time frames then i think RT is more powerful than AB, plus MUCH faster. I have been using AB since 2016 and have moved all my short term systems to RT while adding some more.

There's a free trail and it would be good to have some guys check it out and ask some questions, i've been using it for a few months and wont be looking back.

I can post some charts if anyone wants. There's too many features to list straight up!

 

[Warr87]

looks good mate. it is interesting. i really like the multi portfolio analysis which AB just doesn't do. i see it also exports for the chartist api, but what about exporting and placing orders for IB? if it did orders for IB then i'd be in without a second guess.

 

[Roller_1]

looks good mate. it is interesting. i really like the multi portfolio analysis which AB just doesn't do. i see it also exports for the chartist api, but what about exporting and placing orders for IB? if it did orders for IB then i'd be in without a second guess.

You can do explorations the same as AB they're just called scans, fully customizable . There is actually a few different ways to do it. Is that for basket ttrader?

You can't auto trade from it like you can with AB but you can produce .csv in whatever format you like

 

[captain black]

Looks very interesting @Roller_1 , thanks for posting.

I checked out the Youtube videos. I like the way he's taken the best bits from other software and created his own platform. The code syntax looks similar to Amibroker so should be easy for an Amibroker user to get up and running quickly.

I remember the early days of Amibroker and how it's developed a strong user base so it will be interesting to see how the user base for Realtest grows over time.

 

[Warr87]

You can do explorations the same as AB they're just called scans, fully customizable . There is actually a few different ways to do it. Is that for basket ttrader?

You can't auto trade from it like you can with AB but you can produce .csv in whatever format you like

i had to google basket trader myself as I haven't used that function from IB. that's just a dynamic list in IB from what I can tell. The missing auto trading function is what I really want from AB. Exporting to a csv for radge's api isn't really a big deal .... guess what AB can do? haha. if they had an auto trading function then I dare say they will get a lot more switching over (no need to being an advanced python programmer to create a custom interface into IB).

It does look like a good program and I may get it in the future. At the moment I would still have to buy something like Radge's API to do semi-auto trading whether I buy it or not. I'm not seeing a whole lot of incentive to switch besides the the curiosity of multi strategy testing. Something I really really wish AB had. Not enough to take my money right now though.

Edit: I agree with @captain black that it is interesting and has a lot of potential! I hope it does take off.

 

[stasisbr]

Just testing the software at the moment. Can confirm its very fast for backtesting in comparison to AB.

Couple of other initial impressions:
1 - Charting options appear to be limited in this initial released
2 - Data Source configuration is a bit different compared to AB

1. The Data Source/universe is defined in the individual script file, this includes any specific watch lists/symbols etc
2. Need to click import before running the script - this makes RT import the selected data from Norgate into named db file

I think I'll port one of my active strategies over to RT for comparison purposes.

 

[Roller_1]

i had to google basket trader myself as I haven't used that function from IB. that's just a dynamic list in IB from what I can tell. The missing auto trading function is what I really want from AB. Exporting to a csv for radge's api isn't really a big deal .... guess what AB can do? haha. if they had an auto trading function then I dare say they will get a lot more switching over (no need to being an advanced python programmer to create a custom interface into IB).

It does look like a good program and I may get it in the future. At the moment I would still have to buy something like Radge's API to do semi-auto trading whether I buy it or not. I'm not seeing a whole lot of incentive to switch besides the the curiosity of multi strategy testing. Something I really really wish AB had. Not enough to take my money right now though.

Edit: I agree with @captain black that it is interesting and has a lot of potential! I hope it does take off.

No worries, AB is more customizable and is better for charting and things like that. Depending on your style of trading RT may not suit but for short term multi strat testing it's good.

tbh some of the example scripts he could probably charge for lol

 

[Roller_1]

Just testing the software at the moment. Can confirm its very fast for backtesting in comparison to AB.

Couple of other initial impressions:
1 - Charting options appear to be limited in this initial released
2 - Data Source configuration is a bit different compared to AB

1. The Data Source/universe is defined in the individual script file, this includes any specific watch lists/symbols etc
2. Need to click import before running the script - this makes RT import the selected data from Norgate into named db file

I think I'll port one of my active strategies over to RT for comparison purposes.

Nice one,

Just to confrm. You only need to import a datafile once to populate that file for the date range (once a day if you want the recent data). The data file is then stored in memory for quick access. So you don't need to import for every run

 

[Warr87]

No worries, AB is more customizable and is better for charting and things like that. Depending on your style of trading RT may not suit but for short term multi strat testing it's good.

tbh some of the example scripts he could probably charge for lol

i'm not too worried about charting tbh. I mostly work off what numbers and stats I have. Charts can help to clarify a few things (as I can be visual).

My comments on the product aren't that it isn't good, just that it could have gone a little further especially if it wants to take customers from AB who aren't getting all the features they want (even though AB is a very powerful program in itself).

 

[Roller_1]

i'm not too worried about charting tbh. I mostly work off what numbers and stats I have. Charts can help to clarify a few things (as I can be visual).

My comments on the product aren't that it isn't good, just that it could have gone a little further especially if it wants to take customers from AB who aren't getting all the features they want (even though AB is a very powerful program in itself).

Yeah that's fair. It will be interesting where it ends up because until a few months ago it was designed just for the way he trades and was shared amongst his friends i think. Since i started testing with it in Oct i think it was it has changed a lot to be a more rounded piece of software.

are you trading US stocks or Aus (or both)?

 

[Warr87]

Yeah that's fair. It will be interesting where it ends up because until a few months ago it was designed just for the way he trades and was shared amongst his friends i think. Since i started testing with it in Oct i think it was it has changed a lot to be a more rounded piece of software.

are you trading US stocks or Aus (or both)?

Just Aus stocks right now. But I will eventually move to US stocks too. Ideally I'll have it all automated too, as that is my goal. Unless you are a python programmer or hire someone you need to rig together a bunch of different softwares to automate it. It's a shame.

I have tried to do some multi strategy testing to see what would be the best combination of my strategies. RT would have been perfect for that.

 

[Linus van Pelt]

I'm currently using AB but am also researching Python for technical trading, mainly due to its better capabilities over AB for additional models other than decision tree.

Any comments on how RT might fit in with using Python to develop technical/algorithmic trading?

Does RT do optimization and walkforward like AB?

 

[Roller_1]

I'm currently using AB but am also researching Python for technical trading, mainly due to its better capabilities over AB for additional models other than decision tree.

Any comments on how RT might fit in with using Python to develop technical/algorithmic trading?

Does RT do optimization and walkforward like AB?

i'm not really sure what you mean by 'fit in' with python sorry but yes, optimization and walk forward testing are possible like AB.

See simple example below of doing a optimization of a MA varying the length from 10 to 100 bars

 

[Roller_1]

This is a basic sample script that's included.

i just added the parameters section so i can optimize the short MA period.

i have never downloaded or used Yahoo data before fyi.

So i opened the script > pressed import to import the 'SPY' symbol from yahoo from 92- current which took less than 1 second.

i then ran an optimization on the MA1 parameter which there are 19 combinations. This took, 1.75 seconds ( you can see in the status bar at the bottom)

from here i can double click on any of the results line and view graphs for profit, DD, DD bars, rolling profit windows, etc. you can the scroll through the windows using the up and down keys to easily compare between MA lengths. Or view the trades on charts or individual trade stats.

 

[Linus van Pelt]

i'm not really sure what you mean by 'fit in' with python sorry

Although I know Python I'm just learning what may be possible with its statistical modelling / Data Science capabilities such as Logistic Regression, Neural Networks, etc.

Suffice it to say I create an algorithmic trading system using Python, which gives buy/sell signals, price, position size, etc. etc.

But I would still need to backtest it. Which is something baked in to AB. There are some backtesting packages for Python, but AFAIK they are slow and not as full featured as AB.

So I was wondering if RT would support backtesting a Python model? Perhaps so, if it generated a list of trades say a CSV that could feed into RT?

No worries if you can't comment on the above. I appreciate you informing me/us about this product, and it looks like I need to Do My Own Research to see how it might meet my needs. I'll certainly check it out!

 

[Roller_1]

Although I know Python I'm just learning what may be possible with its statistical modelling / Data Science capabilities such as Logistic Regression, Neural Networks, etc.

Suffice it to say I create an algorithmic trading system using Python, which gives buy/sell signals, price, position size, etc. etc.

But I would still need to backtest it. Which is something baked in to AB. There are some backtesting packages for Python, but AFAIK they are slow and not as full featured as AB.

So I was wondering if RT would support backtesting a Python model? Perhaps so, if it generated a list of trades say a CSV that could feed into RT?

No worries if you can't comment on the above. I appreciate you informing me/us about this product, and it looks like I need to Do My Own Research to see how it might meet my needs. I'll certainly check it out!

I just saw on the RT forum that in the next release it has the option to upload a exported AB trade list into RT to view stats or combine with other strategies to compare.

If you email Marsten he will answer your questions. If a csv is in the correct format it will work

 

[Roller_1]

I recently worked out that RT is going to be as good as anything if not better for tracking trades for both individual strategies and the portfolio.

For US systems it is very easy if you use IB you can just download a Flex report and upload it into RT to view live trades. The ASX is a bit more of a pain because all trades are recorded in US time zone for record dates and RT bar dates don't line up.

But a simple CSV is all that is needed. These are some weekly trades from one of my systems (QTY adjusted)



This is the graph of my current live systems i'm running. My MR has been running for years but i only have stats on it in the right format for the year so far. Each system has currently 1/3 of the portfolio with the Long/Short sharing equity. The shaded grey area is the combined portfolio profit.

I'm really liking this feature because I needed to get better at stat recording at this is making it easier.



Talk about launching a system straight into a start drawdown in the the WTT! 2 weeks later and i would be up around 35% possibly

 

[Warr87]

I am warming up to RT.

And nice graphs btw. Looks good. nice way to track your systems.

 

[Roller_1]

I am warming up to RT.

And nice graphs btw. Looks good. nice way to track your systems.

I was using STT but I don't really like it, not very user friendly for higher frequency. Hopefully this will will solve those issues.

It ewould be good for daily discretionary traders too, you would have solid stats, could easily flick between charts to review trades. You could even test different exits on your entries from a trade list. ie. instead of a discretionary exit use a 5% trail stop of C < C[4] for example

Gee, i'm sounding like a used car salesman

 

[Roller_1]

This is the equity of a sample script included with RT showing a way to implement a MR system.

Shows how much some shorts are getting absolutely smoked at the moment. GME leading the way.



The bottom test is 2010-2021 the second one includes Jan 21. Crazy



Where the damage is done. Tomorrow will be worse considering GME will probably open above $200