Iceberg Right Ahead!
An important part of what QB does in delivering value for our clients, is to dig deep into the details of market microstructure. One aspect of this is determining the sizing of child orders, which is itself influenced by the liquidity already available on the order book. A particularly fascinating topic is the hidden liquidity which can be advantageous to find, and influences order placement decisions when discovered. In the past we wrote a paper on this topic which you can read here*. In this analysis we concluded that hidden liquidity is larger in illiquid products:
1. Between classes, Agricultural and Energy futures have more hidden liquidity.
2. Between different products in the same class:
– RBOB Gasoline (RB), NY Harbor ULSD (HO), and Natural Gas (NG) have more hidden liquidity than Crude Oil (CL);
– Platinum (PL) and Palladium (PA) have more hidden liquidity than Gold (GC), Silver (SI), and Copper (HG);
– The two year note (ZT) has less hidden liquidity than other rates products.
We are always on the look out for changes that are impactful to our clients’ execution. One example was just over a year ago, when the CME completed their phased roll out of the “Market By Order” (MBO) market data changes. (December 2016 – April 2017). You may not have heard of this before – if so you are not alone, awareness is still rather low amongst traders. (This was picked up in an article by the FT at the time. See here).
The main feature of MBO is the ability to see individual order granularity in the order book market data. For example, you can see there are 10 orders of 100 lots each, whereas before you would only see a total of 1000 lots at a price level. The other less understood detail about this change was that the order ID started being published in the market data. This means that others can see the repeat of the order ID as the iceberg tranches are filled, confirming in real-time the existence of an iceberg. Previously this was opt in only (the default was anonymity) and no doubt a rather unpopular option presumably not taken up by most participants. But now for over a year, it is published and there is no opt out. (The details: Tag 1091 “Pre-Trade Anonymity” was previously supported whereby you had to send “N” to turn off the default anonymity and allow the order ID to be published. As of March 26, 2017 this was no longer supported resulting in all tag 37 order IDs being published).
There was already a big gap between human and machine traders participating on exchange order books. MBO made this gap a giant chasm since more information is available to sophisticated participants to identify the trading footprint of others. Manual trading became even more disadvantaged. Generally, more information in publicly available market data is a good thing, but there is a disparity that emerges if only some participants can practically consume and utilize the data. QB empowers its clients with algorithmic strategies to bridge this gap, and be very competitive with all other market participants.
For this reason, we strongly recommend against using native iceberg order types on any exchange, and highly recommend using QB’s Bolt strategy instead. Synthetic icebergs are better than the native ones – these are where the child order refresh is done by the provider, away from the exchange, so no parent order is rested at the exchange. However, while these circumvent the order ID disclosure, they can still leave a detectable footprint. Iceberg order types also only ever fill at one level and can therefore hold up a market that would otherwise move through that level, and will likely do so after the order is complete. Bolt can significantly improve execution relative to icebergs, by avoiding the order type altogether, providing a hidden footprint and allowing for price improvement beyond the specified price level.
Icebergs are a great find for traders seeking liquidity and minimized market impact. They are just not so great for those that choose to use them as an order type. Contrary to Titanic, in trading you want to collide with icebergs – just don’t be the iceberg!
*We recently cross-checked our earlier paper on hidden liquidity using the MBO data we now have available. We found that essentially the conclusions are the same. This was done by sampling multi-level sweeps and using MBO order level detail to compare the pre-trade displayed liquidity with the executed quantity to determine the amount attributable to direct hidden liquidity (icebergs).
Quantitative Brokers
New York
May 31, 2018