I bought the ebook algorithmic trading success from this website. I have been exploring shifting from R to python. This website has awesome blogs and very well written posts and tutorials so I gave it a shot. Honestly buying material from here was pretty much a buyers remorse type of thing over all. really 1/5 stars and i will not buy another thing from there again. Here are Pros and cons
Ebook is well laid out with Easy to read code with highlighted code blocks
90% of the material in the book is the same as blog posts some of it is verbatim the same - so i just paid for free content
Errors in Python code examples
- NONE of the trading examples work due to errors.
Poor customer service when emailed about said errors
- never got a reply
Majority of the information in there is really basic. Most of the people here on futures.io (formerly BMT) will already know of these methods.
Does not cover futures or options examples.
Bottom Line: Don't buy anything from here. Just read the interesting blog posts and walk away.
Side note (yes its a bit of a rant): Also this really blemished my experience with Python. I was really considering moving to python but R is just so much better and the information that is free and open sourced is much better documentation wise. I would never have considered having to pay anything for learning how to use quantstrat or performance analyitics. Yes python may be faster, but R and other packages work way better together and have a much better community. not one that is going to rip you off.
The following user says Thank You to treydog999 for this post:
I just want to update on the situation. So after continuing to work through the code examples in the book to see if i could figure out what the problem is and at least get some value. Its becoming more apparent that the author had implemented some of these before and basically just cleaned up his original version to put in the book. For example his event driven back tester and Moving average cross over examples. Pretty simple stuff, but he makes function calls that don't exist, imports library that are never used, and tries to instantiate objects that were never defined in the books text. Making it pretty much impossible to debug or work out how to turn these code examples into something useful that you can learn from.
Also on the customer service side its been several days with no reply. I have tried to make another attempt at emailing and still nothing. So basically my money went into thin air and there is no one home. Not even an automated reply that they received anything. Just no contact what so ever. If i eventually do get a reply then I will update this review, for now I am just out in the cold.
Hi - Was looking at the website. Did you buy the software package or just the eBook? Was wondering if the software package perhaps has more accurate code snips and tutorials.
You're right about the MA model - kind of ridiculous because every platform includes an MA strategy by default. Not rocket science! And you don't need Python to write a robust MA strategy. Ninja and TradeStation both have excellent platforms for development. The TS function library is very extensive. That said, if you're planning to go up against HFTs or something exotic like that, you probably need more sophisticated coding capability as well as co-location to test it out.
Have you ever looked at the Quantopian community? It's a free platform where you can write, learn and share algos in Python. Here's the link: https://www.quantopian.com/posts You can get all the details in FAQs. There's some interesting stuff in there if you take time to go through the posts.
Ernie Chan's blog also has some good information: Quantitative Trading He mentioned IPython in his November 14th post - apparently a new research environment in Quantopian.
Ninja and TS work well for me for the kind of strategies I develop. But I've always been interested in the quants and how they program exotic algorithms, particularly the microsecond executions.
Thanks again for the post. Interesting stuff.
The following user says Thank You to phantomtrader for this post:
I bought the ebook package instead of the code + ebook package. Because I was not sure if the vendor was any good so i wanted to spend the least amount of money to find out. Assuming the code in the book and the code provided is exactly the same then there would be no difference. I don't think there are additional tutorials in the code package.
Regarding the rest of the post I will answer, but it is mostly off topic. So lets stay on topic after this post.
Not to sound cocky, but the level I am developing at TS, Ninja or MC will just not cut it. They are much to limited, even their c# code is really just wrapped proprietary code that is not platform agnostic. Also .NET is not as secure as c++ or C vanilla due to the Microsoft architecture. Also because of the ease of use of these languages or built in functions, it introduces a lot of limitations. Also I have used MC and TS before and found many bugs in their backtesting engines that have been patched. Due to the proprietary nature of their backtesting engine, it is really opaque as how it is calculating certain rules and fills because they do not expose the code. I am building algos that are much more sophisticated that can not work in these frameworks and prefer to know exactly what is happening. I am currently using R with some custom packages i built but considering moving over to python due to certain tech stack advantages.
Quantopian is still under heavy development and is only for equities. They do not support futures, options, FX or spreads. Also they do not support multi currency portfolios (again problems for fx or cross borders trading). I have gone through Zipline the engine that Quantopian is based on and found it quite lacking for my needs.
Ipython is totally independent of quantopian, it is part of the scipy and anaconda packages. It is used for generating a ipython server and using ipython notebooks to interactively work with python and create reproducible research.
If you are interested in exotic algos and microsecond executions Ninja and TS will never allow you to do that because of the limitations above. Not to mention TS tied to a broker, which I just plain don't like. Millisecond executions sub 100ms and lower will probably be completely written in C++/C on a custom execution platform or expensive enterprise trading solution (Deltix, RTS, etc). In the microsecond space say sub 100us and lower your probably running as close to the metal as possible, using FPGAs and GPUs for concurrent calculations and modeling (MCMC), kernal optimizations, low latency messaging services and cards, custom garbage collection and TCP / network bypass.
The following 2 users say Thank You to treydog999 for this post:
Yes - agree with what you said. I'm interested in the exotic algos but not really interested to get into development. My strategies run on Ninja and I'm using Ninja as my broker. I use TS only because I like EasyLanguage and their function library.
The type of setup you describe is expensive and knowledge-intensive. I'm not a trained programmer - simply self-taught - my background is in chemistry and physics. But as I said, I'm interested in these exotic algos - thanks for the input.
I have finally received word back from the author. He said he has changed and added the the correct code to the ebook. Took about 8 days to get a reply back. I haven't had time to go over it and see if its all good or not. But at least something has been done. I will check the code and book over the weekend when I have some more time. Hopefully everything will go smoothly.
The following 3 users say Thank You to treydog999 for this post:
The code in the book is the same but the code in the source code add on pack has been changed. Basically you need to get the source code pack if you want to see whats really going on. I think its a bit unfair to steer you that way. But it is a business i suppose.
The following user says Thank You to treydog999 for this post: