It's their version of a price improvement. If you submit an adaptive market order, it'll wait to see if it can improve on price. The results of the improvement are shown in the activity reports on IB. Of course, sometimes you get a worse price too, but on balance, it's better to submit adaptive vs market.
@isotope1 This sounds like you run it "on demand" so that you can supervise each cycle and the intermediate steps before orders are submitted. I am not good at reading Python code. It is my impression that the instruments, currency exchange rates, and other parameters are hardcoded in the program. So you can only change those by stopping the program, modifying the code and then running the code again. This is why I asked my previous question as this implies that you can't change settings while the system is running.
Ah yes- that's right. However, as it trades only once a day, this is the best approach in my opinion.
Nice, I hope to contribute. What about feeding the execution stuff into GATs, e.g. creating an abstract project which executes against IB. Then both frameworks could use that library... I guess that could be a nice pull request
It exists! https://github.com/erdewit/ib_insync - I only wish it had been there while I was writing this. If anybody wants to strip out my IB stuff and replace it with this, go for it!
Yeah I'm aware if that project, the code is really nice. I was meaning more of a ib_insync++ which had execution logic and simple execution strategies.
Do they report separate stats for orders submitted with the adaptive algo only? I'm sure it's highly dependent on what you're trading, but my results were far worse after some testing on less liquid equities.
Yeah, there's a column for the improvement. I know a thing about illiquid equities. For that, either try to match things on the book and submit several small orders, or (much better), submit a bunch of limit orders in a 'ladder' with GTC orders and be patient.
Are you referring to the 605 reports? If so, they do report price improvement but the adaptive algo orders are lumped into big buckets with various other order types. Or has IB studied the adaptive algo separately and provided results somewhere? If so, would you mind posting the link?
Are there additional fees for using adaptive algos or any other algos? I remember things like VWAP have an additional charge for equities