Automating Closing Out Stock Position With Large Spread and High Volatility

C

ChadZ1

I'm trading small and mid cap stocks early in their trading days on days of elevated volume. As a result, their spreads are often large (0.5-2%), and I'm struggling to close out my positions in a way that even approaches a VWAP benchmark I used for backtesting that assumes a 5% participation rate until filled.

I'm trading on IB and have tried using IB's VWAP algorithm and Jefferies VWAP algorithm, but neither worked well. I have also tried IB's adaptive algorithm and a number of the other Jefferies algorithms with its volume participation algorithm (volpart) working best, but still far from the the VWAP benchmark. For some reason, with all of these algorithms, when I do get a fill, it is worse than those occurring around the same time the vast majority of the time. In particular, I mostly get fills on the far side of the spread and when I do get a fill on the near side of the spread it's because the stock price briefly moved through it such that the fill I get is actually poor compared to the others that also just occurred. I've noticed many of the best fills that I miss out on occur in dark pools, so perhaps it's unreasonable to think I could ever participate in that?

Any general insight on my problem? Is my VWAP benchmark simply unreasonable?

I was hoping to offload a lot of the work to a trading algorithm. Is there one (not necessarily from IB) that should do a better job? Baring that, what are some good ways to closeout in this scenario that keeps up with the volume? Would placing hidden orders at different levels throughout the order book (perhaps with pegged orders) and only crossing the spread when other trades are doing it be a good strategy? If so, are there any general heuristics to use on when to cross the spread and where to place the orders in the order book based on the other orders already in the book e.g. assume true volume is twice lit volume?
 
Thanks for the response! Can you elaborate in what sense getting the VWAP is unlikely? In your experience, is it just that there's such volatility that I will usually be pretty far from the VWAP but on average my performance could be about as good as the VWAP or should I expect to more consistently underperform such that my average performance will be significantly worse than the VWAP? The first scenario I can work with, the second I need to rethink things.
 
Not enough transactions you can participate on at every price level. Volatility is not the issue. Wide spread spreads and low volume. If AAPL goes nut and trades in a 5% range on 10mm shares, it is possible to get close to the VWAP as you can send child orders all day that will get filled. A stock that is $0.50 wide with less trades does not get you what you want. And, if the stock is OTC, not possible with a MM system.
 
Thanks Robert, I see what you're saying. I don't trade OTC, just NYSE and NASDAQ. Even if I can't consistently achieve the VWAP benchmark on a stock by stock basis, do you think I could get close to it On Average (sometimes underperforming and sometimes outperforming so that in the end I break roughly even)?

Truetype, thanks for the suggestion -- I will try a longer time frame.
 
Doing better than the VWAP, I have no Idea. I expect that requires a broker on the the trading floor of the NYSE where they get a match on ever trade and you don't have to wait your turn on the book. On stocks like you described, I'd rather work the order, accept the open or close or take my chances with the VWAP. Me, unless it is a lot of symbols, I would work the order. I'm assuming this is a large order.
 
I see, thanks for weighing in. I only trade about 0.2% of daily volume for these stocks, so seems like working the order way to go. Just not sure why the trading algorithms I've used have struggled so much on it.
 
Have you compared the VWAP vs the opening auction vs the closing auction to see if it is worth the time? You have to make the assumption you would not change the open or close, or they would change very little.
 
I have looked into that. I think on average that would probably work out better, but at the cost of a huge increase in risk as the standard deviation of my returns would greatly increase. The stocks I'm trading are pretty volatile.
 
You must log in or register to reply here.
Back
Top