How to Install and ...
 

How to Install and Back-test ATS Programs  

  RSS

Rich B
(@copernicus)
Member Admin
Joined: 1 year ago
Posts: 208
09/10/2019 2:26 pm  

This video provides important information that you need to know to install and back-test ATS Price Following Programs using the Metatrader 4 platform which is a popular trading platform offered by many brokers that service the retail FX sector. We provide a step by step guide.

Trade Well and Prosper

The ATS mob

ATS Trading Banner

 

Quidquid latine dictum, altum videtur


Quote
burrup.lambert
(@burrup-lambert)
Eminent Member PREMIUM MEMBER
Joined: 2 months ago
Posts: 24
10/10/2019 4:14 pm  

Hi Rich. Are you using Pepperstone Data for this backtest? I can see you are connected to a Pepperstone Demo account and that the backtest takes it's first trade at the beginning of 2006. I've used all the tricks I know to obtain all my brokers data which is also Pepperstone but for AUDUSD H4 I can only go back as far as June 2009. If it makes any difference I am on the Demo01 server.


ReplyQuote
Rich B
(@copernicus)
Member Admin
Joined: 1 year ago
Posts: 208
10/10/2019 4:48 pm  

Hi B

We have tested across Pepperstone, Dukascopy tick data and also FX Open on this one described in the vid. The actual example used however is from Pepperstone Demo 01 account.

On AUDUSD H4 with Dukascopy free tick data you should be able to get back to 3/8/2003 from available tick data and construct H4 from that data. On Pepperstone we can only get back to 2006 with AUDUSD on H4.

Fred has dropped this H4 data into the Members Forum here. Have a crack at the back test using this data. If you do not have Tick Data Suite you will need to replace existing data history with this set by first deleting the data and then importing this data. Do this while you are offline as you may find that your broker tries to continuously override this data with its own.

There is very little material variation with the result between Dukascopy data and Pepperstone data over the same limited test range available from Pepperstone but Dukas gets you back further which is what we recommend.

It is well worth getting your hands on Tick Data Suite. Not for the tick data accuracy as Pepperstone, FX Open and Dukascopy are all good sources...but more for the long range data available across multiple broker sources (eg. Dukascopy, Darwinex, Alpari etc.). It now seamlessly integrates with MT4. It is an excellent bit of software mate and highly recommended. There is a monthly cost of about $30 per month for the supply of data and updates...but well worth it in my opinion.

Broker data variability can be an issue particularly for brokers who do not work off GMT +2 as our entry condition for our automated systems act off the open condition.

Cheers

Rich

PS Different accounts also possess a degree of variation. For example we trade live off the Pepperstone Edge Server (Razor accounts) which has a commission but tighter spreads unlike the standard Demo 01 account of Pepperstone. Once again however, the variation is not material to the overall result with these long term trend following systems where the majority of results stem from outlier trade impacts as opposed to minor variations arising from variances in the frictional costs of trading. Of course SWAP can be an issue with the longer holding periods, but with swings and roundabouts, it seems to all work out well with limited overall impact. It is a different kettle of fish however where accuracy is essential such as scalping and lower timeframe trading methods such as M1 up to M15.

 

PPS. When you test across your own brokers data, the key is to ensure the variation is not sufficiently materially different to backtest results from other sources. Slight variations are all ok. There will always be slight differences plus or minus between backtest and live trading results. You just want to ensure that the material variation is not sufficiently adverse to turn a strategy with positive expectancy into a negative one. We are only after return streams with positive expectancy over the very long term. The amount of profitability produced per strategy is not important at the individual return stream level. Provided it has positive expectancy (slight profit) across the long term is all we require before we get to the portfolio construction phase where we juice things up.

This post was modified 2 months ago 6 times by Rich B

Quidquid latine dictum, altum videtur


ReplyQuote
burrup.lambert
(@burrup-lambert)
Eminent Member PREMIUM MEMBER
Joined: 2 months ago
Posts: 24
10/10/2019 5:08 pm  
Posted by: @copernicus

Hi B

We have tested across Pepperstone, Dukascopy tick data and also FX Open on this one described in the vid.

On AUDUSD H4 with Dukascopy free tick data you should be able to get back to 3/8/2003 from available tick data and construct H4 from that data. On Pepperstone we can only get back to 2006 with AUDUSD on H4.

Fred has dropped this H4 data into the Members Forum here. Have a crack at the back test using this data. If you do not have Tick Data Suite you will need to replace existing data history with this set by first deleting the data and then importing this data. Do this while you are offline as you may find that your broker tries to continuously override this data with its own.

There is very little material variation with the result between Dukascopy data and Pepperstone data over the same limited test range available from Pepperstone but Dukas gets you back further which is what we recommend.

It is well worth getting your hands on Tick Data Suite. Not for the tick data accuracy as Pepperstone, FX Open and Dukascopy are all good sources...but more for the long range data available across multiple broker sources (eg. Dukascopy, Darwinex, Alpari etc.). It now seamlessly integrates with MT4. It is an excellent bit of software mate and highly recommended. There is a monthly cost of about $30 per month for the supply of data and updates...but well worth it in my opinion.

Broker data variability can be an issue particularly for brokers who do not work off GMT +2 as our entry condition for our automated systems act off the open condition.

Cheers

Rich

PS Different accounts also possess a degree of variation. For example we trade live off the Pepperstone Edge Server (Razor accounts) which has a commission but tighter spreads unlike the standard Demo 01 account of Pepperstone. Once again however, the variation is not material to the overall result with these long term trend following systems where the majority of results stem from outlier trade impacts as opposed to minor variations arising from variances in the frictional costs of trading. Of course SWAP can be an issue with the longer holding periods, but with swings and roundabouts, it seems to all work out well with limited overall impact. It is a different kettle of fish however where accuracy is essential such as scalping and lower timeframe trading methods such as M1 up to M15.

Thanks Rich. I have my own copy of TDS and as you said it is a great piece of software and integrates well with MT4.

I should have been a little more specific with my post sorry. I was wondering how you obtained the Pepperstone data from 2006 - 2009? Using the Page Down (or Home) button method in MT4 I could load data with Pepperstone until 2011. Using the Tools > History Centre "add a bar form 1970" method I could further obtain until 2009. This might be a question for Pepperstone support...

As you mentioned, there should be little difference between Pepperstone/Dukascopy backtests however I would like to validate your testing using the same data set and time periods. I also assume this is why all ATS strategies and "the beast" have been tested from 2006 onwards and not earlier so that you can say with confidence that the whole date range has been testing on both brokers as Pepperstone only goes back to 2006.


Rich B liked
ReplyQuote
Rich B
(@copernicus)
Member Admin
Joined: 1 year ago
Posts: 208
10/10/2019 5:26 pm  

Yep mate....the 2006 and on backtest is used given we trade live off Pepperstone and use this live account as a method to validate the backtest in parrallel testing which we talk about in the vid. We still go further back in our tests using Dukascopy where possible to ensure robustness across a longer range of market condition....but for direct tracking purposes we present our backtests in Pepperstone terms.

Mate if you get yourself a VPS (we use ForexVPS with Pepperstone), then it is useful to simply use a live Pepperstone account that you will be live trading off for data collection purposes. You simply open up your desired universe of charts on M1 and keep them loaded at all times. It slowly builds your M1 history over time allowing you to construct longer range tests from M1 and upwards. 

I think we may have a longer data range because we have been on Pepperstone for many years. I believe there is an initial limit in history provided when you sign up.... so the longer you stay with them the more data you get sort of thing. You might be able to request more data from Pepperstone support....but not sure.

I am just preparing an example between Dukascopy Tick by Tick and Pepperstone Demo 01 Control Points for the same date range to demonstrate what we feel is a lack of any material variation. Many might feel that they are materially different particularly if they are scalpers....but this is not so in longer term trend following world. The slightest puff of a breeze causes a variation in the return stream. Unless the variability is obvious....we don't call it material.

This hopefully will give traders a bit of confidence with these longer term trend following systems and the limited variation that exists in data variability (provided they are GMT +2) and in frictional costs such as variations in spread, SWAP etc.

Below I have attached the HTMP report for Pepperstone Demo 01 and Dukascopy Tick by Tick for the same common date range for ATS1000 AUDUSD H4. 

There is a slight difference (refer below)....but in terms of what we do it is not material to the end result at the portfolio level. They both have sufficient positive expectancy to progress to the portfolio compilation phase of the exercise. We can find that the tick by tick is sometimes better than the Pepperstone Demo 01 result...so there are swings and roundabouts.

Trade Result Comparison

Trade Result Comparison Equity Curve

I hope this helps B

Cheers

Rich

StrategyTester  Pepperstone Demo 01

StrategyTester  Dukascopy Tick by Tick

 

This post was modified 2 months ago 4 times by Rich B

Quidquid latine dictum, altum videtur


ReplyQuote
Share:
%d bloggers like this:

Please Login or Register