IB orders, commissions, and paper trading

[nijshar28]

Hi. I am trying out different execution algos with Interactive Brokers (paper trading mode).

First, I noticed that it allows me to place VWAP and TWAP orders. However, they start showing up as overfilled in the live orders tab on TWS (e.g. Quantity 2200/2000). I know this might be something very specific to the IB system, but I was wondering if anyone else tried using these orders in their paper trading account. Do they actually sort of work, or is it best to stay clear?

Second, I see that my simulated commissions for the TWAP order are taken out for each time slice. I mostly intend to trade mixed lots (e.g. 297 shares) over the course of the entire day. It looks like TWAP slices it into 10-20 smaller orders with a commission minimum applied to each. Would I pay a lot less in commissions if I just try to fill that 297 share lot as single Market-on-Close order?

On a more philosophical note, what order type do you think is best for my trading style? I have a systematic strategy, approx. 300 positions (L/S cash-neutral) with daily rebalancing and 3-week portfolio turnover. Importantly, most names I trade are small and micro cap stocks. The simulated account size is $1M. Right now I am debating between a simple Market-on-Close vs. entire day TWAP / VWAP with an attempt at price improvement. For reference, today I tried a TWAP order from Noon to 3.30pm and ended up paying ~$0.01 / share in commissions for a marketable order. Instead of doing these marketable TWAP / VWAP orders, I should at least try to offset what I assume are increased commissions by attempting a midpoint, or last close buy, right? And if the TWAP order doesn't fully fill with those restrictions, I can top it off with a Market-on-Close order towards the end of the day (?)

Not sure if this is the wrong forum to ask these questions as some of it is pretty specific to IB and their trading platform. But I feel that any advice from you guys would be super helpful. Thank you.

 

[Shadetree42]

I've never been overfilled on any IB algo order.

Comms are a bit confusing, but seem to come out correctly. E.g., you'll see 80 shares filled with a $0.23 comm and then the remaining 20 shares filled later with a -$0.03 comm.

In my experience, VWAP does better than TWAP, and it's usually best to set it to be passive (attempt never to take liquidity). An exception would be a low-price, high-volume stock, where you're better off placing at order pegged to mid.

 

[nijshar28]

Hey thank you so much for your reply. If I set the Passive mode and No Liquidity options in VWAP, should I expect price improvement even for a market order?

As far as limit orders go, I am trying to figure out how to approach that for my strategy. I don't have an intelligent price target for my orders. I just know how many shares I want on any given day. Intuitively, I think I should try to do better than a market order, but I am not sure what that "better" is in this case. Midpoint? Last close? Something else?

 

[HobbyTrading]

On a more philosophical note, what order type do you think is best for my trading style? I have a systematic strategy, approx. 300 positions (L/S cash-neutral) with daily rebalancing and 3-week portfolio turnover. Importantly, most names I trade are small and micro cap stocks. The simulated account size is $1M. Right now I am debating between a simple Market-on-Close vs. entire day TWAP / VWAP with an attempt at price improvement. For reference, today I tried a TWAP order from Noon to 3.30pm and ended up paying ~$0.01 / share in commissions for a marketable order. Instead of doing these marketable TWAP / VWAP orders, I should at least try to offset what I assume are increased commissions by attempting a midpoint, or last close buy, right? And if the TWAP order doesn't fully fill with those restrictions, I can top it off with a Market-on-Close order towards the end of the day (?)

How much do you pay daily for this rebalancing in the tests you have run so far? I don't have any figures myself, but it does sound rather expensive.

 

[Shadetree42]

Hey thank you so much for your reply. If I set the Passive mode and No Liquidity options in VWAP, should I expect price improvement even for a market order?

As far as limit orders go, I am trying to figure out how to approach that for my strategy. I don't have an intelligent price target for my orders. I just know how many shares I want on any given day. Intuitively, I think I should try to do better than a market order, but I am not sure what that "better" is in this case. Midpoint? Last close? Something else?

The algos can't really trade anything low-priced with heavy volume, as they will not usually give you any fills significantly inside the bid/ask. I'll usually do a manual IEX pegged to mid limit order.

In general, IB VWAP will place and adjust regular limit orders on one of the large exchanges

 

[nijshar28]

The algos can't really trade anything low-priced with heavy volume, as they will not usually give you any fills significantly inside the bid/ask. I'll usually do a manual IEX pegged to mid limit order.

In general, IB VWAP will place and adjust regular limit orders on one of the large exchanges

Hey thank you for the pointers.

Pardon my ignorance but what is IEX?

When you say "IB VWAP will place and adjust regular limit orders"? Do you mean a pegged-to-mid order? It has to be pegged to something within bid/ask right? Thanks.

 

[nijshar28]

How much do you pay daily for this rebalancing in the tests you have run so far? I don't have any figures myself, but it does sound rather expensive.

You make a good point. In the paper trading account I paid about $1800 to open $1M worth of stock positions, so that's roughly 0.18% of trade value. It also amounts to about $0.01 / share in my case. Assuming I do 10% of that volume daily (I need to check if that's reasonable), I am looking at a daily loss of 0.018% from commissions. Assuming 253 trading days/year, that's a 4.6% annual loss! Sounds like I should explore weekly, or monthly rebalancing... Thanks!

 

[Shadetree42]

Hey that you for the pointers.

Pardon my ignorance but what is IEX?

When you say "IB VWAP will place and adjust regular limit orders"? Do you mean a pegged-to-mid order? It has to be pegged to something within bid/ask right? Thanks.

IB algos simply place regular limit orders at full-tick prices -- except for occasionally filling with their own dark pools. You need to submit manual orders to send them at mid.

 

[HobbyTrading]

You make a good point. In the paper trading account I paid about $1800 to open $1M worth of stock positions, so that's roughly 0.18% of trade value. It also amounts to about $0.01 / share in my case. Assuming I do 10% of that volume daily (I need to check if that's reasonable), I am looking at a daily loss of 0.018% from commissions. Assuming 253 trading days/year, that's a 4.6% annual loss! Sounds like I should explore weekly, or monthly rebalancing... Thanks!

That is quite a number. And if you think about it, probably a stock will oscillate a bit: one day a bit less, the next day a bit more, and so on. So you'll be trading a lot but basically the stock size does not really change.
Maybe it is better to only rebalance a position if the deviation between the current size and the desired size exceeds a certain threshold. You would have to calculate or simulate to see the impact of such an approach on your strategy.

 

[nijshar28]

That is quite a number. And if you think about it, probably a stock will oscillate a bit: one day a bit less, the next day a bit more, and so on. So you'll be trading a lot but basically the stock size does not really change.
Maybe it is better to only rebalance a position if the deviation between the current size and the desired size exceeds a certain threshold. You would have to calculate or simulate to see the impact of such an approach on your strategy.

Hey. Again your comments are really helpful. I feel that's a great way to think about this. I am already seeing ways to statistically test for a change in stock size. I can visualize individual stock signals as a population of time-series, characterize the noise and use it to detect signal changepoints on individual stocks. I can also apply smoothing to the signal (e.g. take a moving average).