- Joined
- 13 September 2013
- Posts
- 988
- Reactions
- 531
Yeh you can't draw studies in RT or do much customization on the charts. it's more for viewing entries. can plot MA's etc but pretty basic.
eg.
RealTest - Portfolio Level backtesting software
- Thread starter Roller_1
- Start date
Yeh you can't draw studies in RT or do much customization on the charts. it's more for viewing entries. can plot MA's etc but pretty basic.
eg.
For charting, I would say definitely. The focus of RT is portfolio testing, IMO, so it isn't too focused on the charting part. You can view data/charts of a symbol but I think AB is superior in that respect. Not to say it wont be upgraded in the future, but I dont think it's the priority of RT. You will still own a copy of AB so you can still do both for checking individual charts/symbols.
MovingAverage
Just a retail hack
- Joined
- 23 January 2010
- Posts
- 1,315
- Reactions
- 2,565
I got the impression RT was like that. Not that I see that as a problem--I don't mind using both. I am super impressed at how easy it is in RT to develop and test a system.
Charting is about the only thing I now use AB for but RT does have some nice features.
You can toggle the axis between linear and logarithmic with a single keystroke.
You can keep the chart open and scroll through multiple symbols with a single keystroke.
You can add a Chart section to you RT script and display almost anything on the chart. The following adds the 20 and 200 day MA to the main pane and the 15 day RSI to the "Information" pane.
Because it's done in the script what you display can be tailored to the strategy.
DrBourse
If you don't Ask, you don't Get.
- Joined
- 14 January 2010
- Posts
- 888
- Reactions
- 2,090
Hi BoNeZ.....
Might as well throw my thoughts into this discussion.....
I would suggest that anyone using the Linear & Log Type Charts should make allowances for the conflict of interest between the different results….
I rely very heavily on Indicators, so I naturally use an Arithmetic Chart with Arithmetic Calculation Indicators for all my analysis…
I have seen many incorrect decisions derived from using Log or Linear Charts in conjunction with Arithmetic Indicators….
If and when the Mathematicians of this world come up with Linear and/or Log based Indicators then we would be able to match "Linear Charts with Linear Indicators, Log Charts with Log Indicators.
Each Chart Type produces slightly different outcomes when conducting Chart Analysis - I find it amazing, to say the least, that some people persist in using Indicators with Linear & Log Charts, because Indicators are an "Arithmetic Calculation" based on individual and complex Mathematical Formulas…
Some food for thought.
Cheers....
DrB
DrBourse
If you don't Ask, you don't Get.
- Joined
- 14 January 2010
- Posts
- 888
- Reactions
- 2,090
Never looked at RT or AB - so I am prob wrong in saying the following:-
Noticed a few posts stating that in is difficult to get RT results to align with AB's results - my above post may explain some of those diff results - Dunno.
- Joined
- 8 June 2008
- Posts
- 12,773
- Reactions
- 18,677
I have to agree on the last point
RT might be intuitive but not for a programmer..I was one
I was having serious troubles and even direct help from the creator did not help
A setting in the graphical interface seemed to have caused the issue
I find it much more difficult to develop than on AB but it is much faster and RT support is outstanding.
I had to stop my trial where I converted my AB systems to RT but will restart asap
It is also hard to compare AB vs RT results as AB is not flawless
My RealTest licence expired and I paid for an extension to continue getting updates.
They might be at different points in development but I've owned AB for years and the only change / fix / enchancment I've considered paying for was support for the Norgate NDU while most RT updates have something I end up using.
If one says buy stock XXX today the the other say buy it tomorrow I'm going to follow what I'm using (in my case RT). These small differences snowball over time with back test results from an AB and RT being similar but noticably different.
They might be at different points in development but I've owned AB for years and the only change / fix / enchancment I've considered paying for was support for the Norgate NDU while most RT updates have something I end up using.
I gave up trying. Once I had the system and database settings in sync the two were closer then had to decide which was "right". When asking a question in the RT forum I get a answer (often from the owner Marsten) while any answer in the AB forum is usually "read the manual" followed by "you don't know how to code".
If one says buy stock XXX today the the other say buy it tomorrow I'm going to follow what I'm using (in my case RT). These small differences snowball over time with back test results from an AB and RT being similar but noticably different.
- Joined
- 28 May 2020
- Posts
- 127
- Reactions
- 229
The doc is available online and you don't have to be a customer.
Here are the Tiingo instructions: https://mhptrading.com/docs/topics/idh-topic517.htm
- Joined
- 28 May 2020
- Posts
- 127
- Reactions
- 229
Trying to get AB and RT results to match exactly is difficult for many reasons:
- AB is buggy
- AB and RT handle Padding and Adjustment differently
- RT always uses as-traded prices, and back-adjusts (i.e. *after* the split) on the fly as required. There are subtle look-ahead leaks in AB depending on Adjustment, esp. Adjustment: TotalReturn.
- RT handles dividends differently than AB. The only way to include dividends in AB is Adjustment: TotalReturn, but then that distorts the historic time series by adjusting the prices for future dividends.
- The derivation of some metrics is different. For example, AB always uses compounded derivation, even if the position sizing model was non-compounding.
- I once investigated a difference between AB and RT where a new stock got listed, I was using weekly bars, and AB formed the complete bar (i.e. ROC(C,5)) one week early. Once you start getting trades differences between the two you'll never get exact results.
- There are some precision differences between AB and RT, so a position score of eg. 1.234567890 and 1.234567891 might result in symbol XYZ chosen in RT and symbol ABC (alphabetical) chosen in AB.
- Joined
- 28 May 2020
- Posts
- 127
- Reactions
- 229
I agree there isn't as much eye candy in RT than AB. However, one thing I love about RT charting over AB is, when I open a chart from the trades list, it horizontally scrolls to the trade (the buy--sell line above). In AB even when I RMB "Show current trade arrows", I have to scroll to find the trade. AB may have prettier charts, but I can view many charts much faster in RT (just press up/down arrow to scroll through the charts in the trades list). And if you sort the trades list the chart stays in sync with the currently selected trade. I usually sort the trades on PctGain ascending and descending to quickly review my best and worst trades.
- Joined
- 28 April 2020
- Posts
- 93
- Reactions
- 151
I purchased RT about a year ago roughly. Have to say i LOVE it!! Way easier to code and get ideas tested (in minutes) than AB and the RT forum is actually friendly and helpful (total opposite of AB forum imo).
I have developed 2 excellent systems with RT that i now trade live. Scans are quick and easy. As mentioned above the charts dont look as pretty and imo dont have the functionality of AB, so i still use AB to eyeball stocks
I have developed 2 excellent systems with RT that i now trade live. Scans are quick and easy. As mentioned above the charts dont look as pretty and imo dont have the functionality of AB, so i still use AB to eyeball stocks
Hi Roller_1,
I have a few questions about RealTest:
1. Can the software perform intra-day back testing?
2. If intra-day back testing is possible, what would be a good cheap datasource for ASX FPO intra-day data?
3. Can intra-day time frames be applied for back testing periods? For example, if I wanted to test a trading strategy that only enters new positions between 10am and 12pm.
Thanks
Realtest doesn't support intra-day testing. In the guide docs, it's suggested that it may in the future.
Amibroker is probably a better option for intra-day. Realtest is great for portfolio testing, not intra-day type strategies.
As for #3, I would think you'd need a more quant heavy type software. A lot of these engines are built in python. Finding a 'cheap' source of these intra-day bars is going to be tricky as well.
- Joined
- 13 September 2013
- Posts
- 988
- Reactions
- 531
Basically, the answer is see what @Warr87 posted above. Some people have found ways around it I believe, but to what extent they are using RT for I'm unsure. If you go to the forum there is some discussions posted in there. But there is no elegant solution
For that kind of specific trading, I think something like quantrocket would be better. There are others, but I can't remember what they're called. If you don't know how to code in python, look up '100 Days of Code: The Complete Python Pro Bootcamp' on Udemy and try to get to day 45, at the very least. Then you should be able to follow along tutorials on trading in python, including using APIs to retrieve tick data, then clean it for backtesting. Collecting your own tick data will be your best bet, IMO. If you're comfortable with headless environments, you could even setup a LXC or docker container, or a lightweight VM. A beelink, or n100, '1litre computer' will be a low power draw but allow you to run things like a python trading program. (I haven't done this myself but I am looking into setting up my trading computer on a small 1lt computer.)
Similar threads
