Hi Simon, from our monitoring all IB integrations are running smoothly. If you're seeing an issue in the account, could you send us a private note to [email protected] with permission to look at the dashboard? Then we can take a closer look at the orders.

Okay I am seeing them in IB, but not in Quantopian's paper trading, or else I have a bug somewhere that only shows up in Quantopian paper trading but not IB.

I'll email you folks.

It was my bug, possible related to something in Quantopian/zipline which is not obvious, so I'll explain it for posterity:

Periodically I calculate my signals and rebalance my portfolio, and then for some of the assets, issue stop-loss orders at 10% below last price, trailing (ie: for longs, stops never go down, only get tighter). Every handle_data, I monitor the stops, and if they are for less than my current position, perhaps because of late fills, I cancel and replace them.

The key is that to place the stop losses, I use:

orderid = order_target(sid, 0.0, style=StopOrder(stop_price))  

Now, if I place that order immediately after placing my position-resize order, on startup, at least in Quantopian paper trading, my position size is still 0, and here's the key point: that orderid will be valid, but if you call get_order on it, you get a null/invalid order

So, in my handle_data, I was checking all my stop_loss orders, but to prevent the above from crashing my code, I was checking to make sure the order was not null. However, this left a gap in my logic; what if I iterate all the orderids I have, but not all of the orders are valid. I therefore don't update my stops to my new position sizes (from 0 to 1000 shares for instance), so for an entire day, I have no working stop loss orders, and that's why they weren't showing up in the blotter.

The solution is, if any my orderids correspond null/invalid orders, do over the entire stop-setting process, to place brand new stop orders. By this time, since it's a new minute bar, perhaps some of my positional orders have filled, and order_target won't be generating these bogus orders.

Tada!

And there's still a bug - I have to do a do-over if any of the stop orders are bogus. Sigh!