30 year data

Hello!

Does anyone know, where we could download / buy 30 year good quality data for strategy generating and backtesting?

Brg, Csaba

Hello,

I doubt there can be found any free data of such long range. We have 17 years of forex data from Dukascopy in QuantDataManager and StrategyQuantX available which is not that bad but if you need more years I suppose it needs to be purchased somewhere

I have 34 years of pretty much hole-free M1 data (1987 -> today) for 21 symbols, which I use to create my strategies in SQ. PM me if interested.

I don’t understand what 30/40 years of data is for … since since 2008/2010 has been the past in the telematics market … also because ‘volatile’ dynamics and monetary policies have totally changed. with such old data you risk modeling strategies that are no longer current. in my opinion’ and algo professional traders that I know 10/12 years are more than enough

just out of curiosity I can check further back… just out of curiosity

thanks gianfranco

Not everyone has to follow this concept, but I prefer to include all the data for a pair since it´s “inception”, which in turn creates rock-stable systems (at least for me that is in the last years) that make money, compared to if “just” using 10 or 15 years (which I did before as well).

@geektrader 

I wrote you PM. 🙂

@all

I would need long M1 data, because I do not like the monte carlo tests. These are just probabilistic game for me.
I like test method which are “REAL”. So for me a monte carlo gives not a real, clear result, because there it depends on, which parameter will be randomly modified.

But the tests like another timeframe, another market, higher-lower precision are better, because there is no probability. These gives exact results.

Or what are your opinion?

you are talking about Asirikuy data, which i think is from OANDA broker…they are UTC1

yes you can find strategies which are working for the whole history, but mostly only on EURUSD and does this long future will grant you some future results?

could be YES, but also NOT 🙂

i am not using those data, only for interest

dukascopy data are by my opinion better in quality

I have never used tick-data to create my strategies, I am using this as an extra measure of stability. If a strategy only works if using intrabar (1M or tick-data) but fails on the select time-frame only method (or vice versa), I would not trade is. Here is one that I found today, similar to yours, just that OOS was fine (cannot be seen in this picture as this is a re-optimized result). Ret/DD is in the 40er range, which is pretty much standard indeed if using 34 years of data EURUSD H1 timeframe. I never had much success with building strategies on “just a few years”, while the ones created on 34 years of data all do well going forward for the last years. But everyone has their own approach, so whatever works for you, is good 🙂

Since I´ve been asked about this many times, I´ve decided to make this data available for everyone at a pretty fair price (took me years to compile) and I´ve posted it over here:

https://www.forexfactory.com/showthread.php?t=986459

those must be still the data from Asirikuy Community which are 160 USD/year – are you only reselling or what took you years?

It was geektrader that made it for Asirikuy i remember reading on their website.

good job then

ALl thought working on data does not mean you own it and can re sell it. And since geektrader did not respond it is probably not his data to sell.  I mean i have worked on it aswell so then it is mine ? In that case I be happy to give it away for free.

i would say that guy from mechanical forex would be not happy 🙂 2cents