Best place for even more backtesting-data?

Forums › ProRealTime English forum › General trading discussions › Best place for even more backtesting-data?

Viewing 12 posts - 1 through 12 (of 12 total)
  • 03/16/2018 at 8:54 AM #65401
    jebus89

    Anyone know of a similar platform that could provide even more backtest data?

    Metatrader maybe? Anyone used MT4/5? how much backtest data do u get on indicies?

    In PRT on the hourly+ chart on eur/usd u can get data going back to 199x, but on indicies its just 8-10 years on most… Does anyone know where and how i can get data for say 20 years on S&P or something similar?

    03/16/2018 at 11:11 AM #65413
    Despair

    Because I was too cheap to pay a lot of money for data I coded a database that takes price data from various free sources. My database is for example parsing a lot of data from google finance (great for Stocks an ETFs) and what it doesn’t find at google finance it downloads from Quandl. Quandl gives a lot of data even with a free account. For example for the S&P 500 I have daily bars since 1982. For several commodities they have data all the way back to 1959!

    Then you will of course need a platform that allows you to import data from an external source.

    03/16/2018 at 7:00 PM #65467
    Vonasi

    I would question what use data from 1959 is? The markets today are very different animals from what they were even 20 years ago so data from 1959 is pretty irrelevant. In fact anything before about 2004 is pretty pointless as that is when automatic algo trading started and the way markets behave changed considerably. What happened yesterday has far more relevance than what happened when dinosaurs were roaming the earth!

    03/16/2018 at 8:46 PM #65473
    Despair

    I agree i general but then it depends on what kind of strategy you are looking at. For example for seasonal strategies I had no good results when using “only” 20 years of daily data. When I developed them instead with 50+ years I got much better results. But this are strategies that trade maybe 8 times per year on average so this is not so strange that more data was better.

    03/16/2018 at 10:05 PM #65475
    Vonasi

    That is a good point about seasonality. Where seasonality is concerned more is definitely better when it comes to data. A truly seasonal commodity (commodities are where I feel the best seasonality is to be found) will hopefully be truly seasonal over a very long period and the longer the period the better confidence can be had that that seasonality will carry on into the future. However if a run of the mill indicator is applied to a 1959 to the present day chart then it is highly unlikely that it will return similar usable results at the start of the period compared to yesterday.

    03/17/2018 at 11:12 AM #65485
    Despair

    I run two seasonal strategies live both developed on data reaching back to 1970 (with PRT; I have so much data for some commodities on daily bars) that perform well. I also tried to develop similar strategies for assets with less data but they failed pretty quickly.

    03/18/2018 at 1:26 PM #65570
    Despair

    LOL, talking of “too” old data. I just got my WRDS access up and running and requested daily data for the Dow Jones and was really surprised about the amount of data they over. The Dow Jones data starts in 1896!! 😀

    03/18/2018 at 3:48 PM #65583
    Vonasi
    Despair wrote:

    The Dow Jones data starts in 1896!!

    I’m guessing that won’t be tick by tick!

    03/18/2018 at 4:55 PM #65586
    Despair

    Daily data is rarely tick-data. 😉

    03/18/2018 at 5:18 PM #65591
    Vonasi
    Despair wrote:

    Daily data is rarely tick-data. 😉

    I was being ironic but I am not very good at it 🙂

    I bet it would have been very easy to write a profitable strategy that worked well in 1896. Finding a computer to run it on may have been a little more difficult.

    03/18/2018 at 5:38 PM #65594
    Nicolas

    Back in the ’50s-60s, slope of a simple 3 periods MA was a really profitable way to generate profits 😉

    03/18/2018 at 6:32 PM #65607
    Vonasi
    Nicolas wrote:

    Back in the ’50s-60s, slope of a simple 3 periods MA was a really profitable way to generate profits 😉

    …and that was probably the only indicator that they had and it took all day to draw each individual one on each individual paper chart with a pencil.

  • Author
    Posts
Viewing 12 posts - 1 through 12 (of 12 total)

Find exclusive trading pro-tools on ProRealCode Marketplace

Similar topics:
automatictrading best indicators
best indicators/strategies to exit from a long position
how to find best performing algo this week?
prt v12 the best ever done…….
select best signals of a strategy to adapt it to the spread
stoploss not put in place
the best day for prorealtime (not) – all fails

Create your free account now and post your request to benefit from the help of the community
Register or Login