So MW, what would you say the role of strategy P/L has to do with whether or not this book can achieve its intended purpose?
If not talking about profitability and just setting things up as some book says with examples, I’d say under a month with no prior experience and about a week for a programmer. Assuming working on that close to fulltime. Python is just yet another tool to learn. It doesn’t magically solve anything and has plenty of rabbit holes people gets sucked into, that are completely irrelevant to trading. If I’d start from scratch I’d stick to specialized backtesting and execution software (eg RealTest) and only go beyond if it absolutely can’t support something and I’m absolutely certain that something is critical for my trading. If you already are an experienced trader and have your strategy, the key for automation would be having rules a machine can follow. It’s not that hard to automate to the point when (a) you just need to run something once a day and have it generate orders for you. Way more tricky to have (b) orders dynamically managed during a day with a lot of input real time data points. Next level of complexity is (c) full automation when you run and forget for like a week or two, and only rely on status reports. Most people like the sense of control “a” gives. Some strategies require “b” to be viable. No one needs “c” really and it’s very time consuming to develop. I’ve done it, but wouldn’t recommend it. More of a vanity thing.
MW you still with us buddy? How bout a comment on the pdrop some more wisdom on us MW you still with us buddy? How bout a comment on the post above? Sure is quiet in here back at the topic zone for this thread, which is not as it turns out anything having to do with P/L, but the OP is talking about "help me get setup", "the basics", "implement trading". I don't see anything about lets act superior and have a pissing match about credibility based on a nonrelated criteria, which is what you did with your derogatory screed about P/L. Who asked ya pal? Now, how bout you drop some more wisdom on us dealing with my question to you in that previous post to you M.W.? ...This one right here... Your Go big guy...
Thank you ValeryN. Good stuff. Sounds like a non-coder python noob is looking at a couple hundred hours to get a foot in the door. I've seen you talk RealTest and why. I will guess a couple hundred more hours to get a foot in the door with real test, so there's 400 hours to get standing up with no movement. Roughly 400 hours of work to get to the door. The Toolbuilder's Dilemma.
Certainly no point in doing stuff that gives no movement. Typical reasons for automation are Freeing up trading hours in the years to come for something else that is either more enjoyable or makes extra money. Making money thru running strategies that wouldn’t be feasible without automation Making more money buy doing more of the same thing that you know is already working and scales to more instruments Reducing human errors including emotional once. Somewhat offset by programming bugs
The 1 star reviews on amazon.com give me the impression it might be possible to get the code to work only for someone who is already a python expert with lots of spare time. But Andreas Clenow wrote his updated code should work with a newer, maintained version of the underlying zipline backtester. So it's hard to say how long it would take for someone without programming experience to get it working. You might want to look at some videos (part 1, part 2, part 3, part 4) about installing and using zipline first.
I said everything there is to say for me in a post on the first page of this thread. Not sure what else you are looking for. I also made my stance clear about using third party backtest engines in other threads. I believe most people will fail trying to use third party software without any knowledge how it was written. Sure, one can iterate over data points and play around like a kid in a sandbox but that's it. Or have you ever seen someone who has no clue how to code successfully operate an algorithmic trading framework? Every off the shelf platform is limited by default, some more than others. And anyone who is serious about this work writes their harness more or less on their own or builds on a few foundational blocks. That is my experience with everyone I converse and work with in this area.
No. It confirms what I believe. People who are in the business of writing books, speaking at conferences, and marketing themselves rarely make enough trading to earn a living. If they did why jump through all those hoops and inconveniences. And the book teaches nothing that one can't learn from the software documentation. It is marketed towards those who look for shortcuts in life. Rarely do those individuals get what they hope for.