Discussion in 'Hook Up' started by Astrevea, Apr 2, 2022.
May you recommend a place to learn?
I would not create this topic if I didn't have many valid reasons for not doing it myself, the main ones being 1) i did some very basic programming in uni before and I know I am not very good at it and there are people there who can do a quicker and better job 2) I understand that programming is a great investment, but I don't have the time right now to learn it.
People who made some negative comments above, you do not know my exact life circumstances to be making such comments so please don't bother posting in this topic again, I won't bother reading.
I am not going to get into discussion about the pros and cons of backtesting/forward testing/live trading here. I have traded live indeed. Please keep it on topic guys.
Well, my current plan is to trade with IB, from my research this is the best broker in the world in most aspects. Is there a way to download historical data from IB going back many years, including small TFs, like for examole Dukascopy broker allows you to?
To those who have sent me a PM, I will reply in a bit, thanks
Decide on a back testing platform and then look on www.upwork.com or www.fiverr.com for a programmer who has experience of that platform. There are a lot of them.
It's possible there's some off-the-shelf software you can use to do this on your own, unless your rules are extremely complex.
Maybe Ami-Broker or Investor RT can do it. Ninjatrader can possibly to it, too.
Beyond that, I have good experience using Upwork for some smaller projects. The key is screening properly and doing some "interviewing" before awarding a job to someone.
If you have a long term perspective on this, maybe the best is to learn coding yourself. I regret not doing so my self and find it hard to find time to do so now with my current schedule.
Is there such thing as a standalone backtesting platform? Or are are you talking about a trading platform itself, i.e. MT4, MT5, ctrader, IB Trader Workstation, etc? Are there specific reasons why a programmer cannot just use raw historical data in Excel format to do the backtesting in which ever way they think is best? Thanks for your input and staying on topic guys by the way.
Is define by 5 parameters or 50?
Tons of data or some fromula of price and time?
It's not the same think in terms of work scale..
For simple cases back testing platforms like AMI broker can be good
Sure! I'm a member of both packtpub (so cheap it's almost free), oreilly ($39 a month). You can pick a book or video lesson, and start learning. Sometimes I buy udemy courses. If you are willing, you can even use youtube and ask questions on stackoverflow.
First you need to decide on a language, then proceed from there.
You need to learn three things to be able to backtest
1-) accessing / storing data (mysql, redis or any db knowledge is required)
3-) a list of rules to test
I mean, my backtester is full of ifs and elses, there's nothing complex about it.
I know this is probably an oversimplification and there's so much details to know, but I'd be happy even if it's a little useful to anyone here.
In terms of coding myself, I explained to some other people who replied why I am not in a position to do it now, that message is still being moderated. The rules are not extremely complex to be honest, the only complexity if you can even call it that is candle aggregation which I mentioned (i.e. having multiple variations of M5 candles being built from say M1 candles, such that you have one version opening at 10:00, one at 10:01 etc... this is to create more opportunities, I know that this is not something new to many experienced traders probably).And the other complexity is that we want to define rules for all candles of the "impulse" and for all candles of the "correction", this rules are basically the same for all impulse candles, and the same for all correction candles, with just a couple of exceptions. But the impulse may consist of a variable number of candles, same goes for the correction. This is where a bit of complexity is added.Do you reckon the software you mentioned above could be suitable having seen my rules? I will try Upwork, I tried Fiverr.com in the past and I know exactly what you mean about screening/interviewing
It's definitely not that complex, please see my last post once it gets moderated.
Thanks for this useful input about databases etc, this is exactly what I thought about this morning. You need some sort of database.
depends on how complex your model is. more complex you might need some light python. if its a simple strategy, you can honestly just test it in excel yourself. most of your work will be in formatting the data but the backtest itself will be trivial after that
outside of that, you could pay someone on upward, or find some dude on reddit's algotrading sub. plenty of programmers on there with 0 knowledge of trading and financial markets who are eager to partner with someone and do grunt work like this provided they feel that they are 'learning something'
id advise the excel approach for now. learn python once project complexity demands it and/or work requires it
Separate names with a comma.