Three Innovative Ways to Reduce Adverse Selection
Adverse selection in the context of limit orders is often refer to it as “being picked off” or “getting run over”, as the market moves through a trader’s limit price. One way to avoid this risk is to predict when prices are about to move and then quickly get out of the way by canceling or repricing orders. This is effectively how high frequency market makers manage adverse selection risk. But this approach is generally prohibitively costly for most buyside traders since it requires not only sophisticated analytics but also cutting edge technology.
Another approach to address this is by slowing the execution process to allow prices time to adjust and pegged limit prices to be updated. This second approach is something that must be done at the trading venue level since it involves taking actions on other traders’ orders, either explicitly via delays or implicitly by delaying the execution process.
Recently (and not so recently), various venues have introduced tools that follow one or more of these approaches to reduce adverse selection.
In this post, we focus on three of the more innovative approaches taken at the venue level to address adverse selection:
* The IEX D-Peg and D-Limit
* IntelligentCross Midpoint and ASPEN
* The Nasdaq Midpoint Extended Life Order (M-ELO)
IEX D-Peg and D-Limit
The IEX -Peg and the recently approved IEX D-Limit order effectively use both approaches in tandem to reduce adverse selection. These order types use IEX’s “Crumbling Quote Indicator” (CQI) to predict when the quote is about to move adversely and reprices these orders immediately. Any incoming orders attempting to “pick off” these orders, however, must pass through its famous 350-microsecond speed bump, which is applied to all incoming order messages, but not to the pricing updates applied to pegged orders. The latter basically gives the IEX D-Limit and D-Peg orders something akin to a 350-microsecond head start when getting out of the way. Therefore, if the IEX CQI is effective – and the data the IEX has made public suggests that it is – the IEX D-Limit has the potential to significantly reduce adverse selection on lit orders in the same way it protects dark D-Peg orders.
It is hard to overstate how innovative the D-Limit order type is. Employing HFT-type intelligence to adjust prices is novel in that it gives the buyside some of this logic without the heavy quantitative and technology investment required to do this in-house. And this logic comes with a 350-microsecond head start baked in. In addition, the approval of the D-Limit order type could have far reaching implications should other venues decide to deploy similar re-pricing logic or perhaps to take other novel approaches to improve performance.
IntelligentCross Midpoint and ASPEN
A second tool that has been gaining traction since its rollout two years ago is IntelligentCross. IntelligentCross provides both a dark midpoint match (“Midpoint’) and a lit book (“Adverse Selection Protection Engine” or ASPEN). IntelligentCross does periodic matching, where orders are effectively held until the time of the next match, at which point matching orders are paired off at the price prevailing at the time of the cross. While incoming orders aren’t explicitly slowed down as with IEX, the execution itself is typically delayed for most orders as they wait for the matching period.
For “uninformed” traders, the wait is generally inconsequential, as the duration is brief (milliseconds) and prices are unlikely to move in such a short period on average. But this is not the case for predatory traders looking to “pick off” orders just prior to a price change. They are typically submitting orders just before an imminent price change to pick off about-to-be stale orders. So, to them, any delay can be quite costly, as it provides additional time for prices to move and crossing price references (e.g., midpoint) to update, thereby eliminating the profitable trading opportunity.
I should point out, though, that while the approach taken by IntelligentCross is conceptually simple, their implementation is quite sophisticated. For example, IntelligentCross uses sophisticated modeling to determine the optimal time between matches to not only reduce adverse selection risk, but to do so without millisecond level delays, meaning the potential number of matches are extremely high. So, it is not really accurate to describe IntelligentCross as a simple point-in-time matching engine. Rather, a better way to think of it is as a “nearly continuous” market, where the “discontinuities” in trading are extremely short and exist primarily to reduce adverse selection and enhance performance.
A third tool, the Nasdaq’s M-ELO order type, takes yet another approach to adverse selection reduction. Since Phil Mackintosh has written extensively on this topic, I will simply refer you to his work. But in a nutshell, M-ELO aims to reduce adverse selection by matching orders that pre-commit to rest for at least 10 milliseconds. After two offsetting M-ELO orders have rested the requisite 10 milliseconds, they cross at the prevailing midpoint (i.e., the quote after the rest period has ended). By forcing both sides to commit to a delay, the M-ELO order type reduces the profitability of very short-term trading strategies, e.g., those with signal horizons of less than 10 milliseconds.
From a buyside perspective, all three tools above could lead to significant reductions in adverse selection costs for resting orders and are clearly worth a look, especially when using algos with a significant passive component, e.g., VWAP/TWAP, dark aggregation, etc. And with most brokers, incorporating these venues could involve a simple change on the broker’s side, without requiring any significant effort by the buyside trader herself. With that said, the efficacy of these tools in the context of an algorithm hinges on exactly how these tools are incorporated into the algorithm. Therefore, having a discussion with the broker or an execution consultant (either broker-provided or an independent consultant like The Bacidore Group) is definitely worthwhile to ensure these tools are being utilized most effectively.
The author is the Founder and President of The Bacidore Group, LLC. For more information on how the Bacidore Group can help improve trading performance as well as measure that performance, please feel free to contact us at email@example.com or via our webpage www.bacidore.com.
Please check out our new book Algorithmic Trading: A Practitioner's Guide, available on now on Kindle and paperback. Click Here for more details.
For an overview of our other services, please click HERE.
And please check out our other blog posts available HERE.
And if you'd like to receive notification of new posts, please join our mailing list below.
Copyright 2020, The Bacidore Group, LLC. All Rights Reserved.