Best "risk-off" assets?

In looking at the behavior of bond ETF's around the 2008 downturn I see EDV pops up and performs well (along with TIP and TLT). I don't know what it is or why, but here's the notebook.

It's another zero-coupon fund, like ZROZ.

Paul Perry

Yes, EDV and ZROZ move exactly the same except for this EDV $11.11 dividend on Dec 22, 2009. So I think ZROZ it is!
Another thought is to dynamically detect what is moving counter trend at the time ...

EDV is pretty low liquidity. Yesterdays volume was under 35K shares. At this moment, EDV has a spread of 0.11 and is displaying one lot on each side of the spread. ZROZ is displaying a 0.16 spread now and looks to average about 20K shares per day. Wouldn't you expect a lot of slippage with any size on either of these? I wouldn't want to do too many round trips taking liquidity on these.

Yes that's actually exactly why I decided to go with TLT for now.

Seems ripe for market making though eh? Must be super correlated with TLT at a beta of 2 or something.

Simon,

I agree that it looks good for market making. That's gotten a lot tougher these days than it used to be. Payment for order flow and other market structure issues which result in legalized front running by fractional cents are making it really tough to be a liquidity provider as a retail trader. This has actually widened spreads considerably on low liquidity stocks as retail liquidity providers can't get filled. As a result, I'm here to explore other strategies.

Interesting! I remember reading http://www.amazon.com/Day-Trade-Online-Wiley-Trading-ebook/dp/B001OWDXPW years ago, and I've always had that idea in the back of my mind - automated market making of illiquid ETFs on NYSE alongside the specialists, but I never did it. It sounds like this is what you have been actually doing for a while? I'd love to hear more about it.

BIL

I used to love BIL... years ago. These days it doesn't pay a dividend and very slowly depreciates due to near zero short term interest rates, operating costs, along with fear of rising rates. Cash is a better choice.

Simon,

I actually have been doing my trading manually... several thousand small executions per year. It used to work, but doesn't any more. Profits are lower while risks have climbed. I've watched other retail traders leave the game. I've watched as their "footprints" on the order books have evaporated over the years. Although spreads have widened significantly as a result, they can't be captured by retail traders. If you try to narrow the spread a little... or a lot, non-displayed liquidity will jump in front of you by one thousanth of one cent and capture the trade. The displayed retail trader will generally only get an execution if the market is ready to move well past their price level. Thus risk is higher than it used to be and fills are far less frequent. So, you say, why not go the non-displayed route. Well, then you are hiding invisibly on one exchange and aren't protected by the NBBO. So, when an order comes through, you won't get executed anyway. Like I said, I'm well past due for a new strategy.

Hmm have you tried posting the limit orders on one of the inverted taker-maker exchanges? (forget which ones they are, I think one of the BATS is like that).

I was also thinking, if one were using a fast enough platform, even if you got adversely selected in ZROZ, say, and traded through, there might still be time to hedge in TLT. If the spread is $0.13 and beta is 2, that means you've got $0.03 of leeway in TLT to get an aggressive fill and get hedged. That's one of my future projects. Totally impossible with Quantopian of course.

I use IB's unbundled commission structure and exchange rebates to significantly reduce my transaction costs. I have tried to use the inverted cost structure exchanges on Bats BYX and Boston both before and after the market structure changes. However, I found that the use of those exchanges only increased my costs. If 95% or more of the liquidity is in the form of non-displayed front running, you gain little by being first in line for execution when that liquidity is exhausted. Even before the front running, I was pretty good at positioning myself on the right exchanges before the crowd and gained little by being one of the few who were on Boston (BEX) and later BYX.

As a market maker, you are always adversely selected. It generally used to be mostly by a few cents even on illiquid securities. Sometimes you would get filled at multiple price levels when a single large order hits a or from a flurry of activity, but the price would usually bounce right back. Now, all front runners get their fill and the rubber band stretches tightly before you get taken out. These days, they assume that you have a large amount of non-displayed liquidity to go with that single lot that you might be displaying. When your one lot or otherwise small order is no longer displayed, the displayed spread might become huge as all the algos remove the remainder of the displayed liquidity instantly.

I understand your thinking and it appears sound... If you carry it through, I'd really like to know how you do. IBPY, Matlab, or a faster (longer development time) direct interface through the IB API, etc. would be more appropriate platform choices. As you stated, Quantopian is not geared for that type of trading. I was actually looking for methods to automate my market making as you are suggesting when I found Quantopian. As a result of the rampant front running, I've decided to explore alternative strategies on Quantopian before expending additional resources to compete in a game where the structure of the marketplace has become such a disadvantage. Even if I find that I wish to continue market making, I believe that I'll get a lot of relevant knowledge and experience here.