The Case for Price Improvement Reform

header image

How trading execution prices can be improved using better pricing models and additional financial market data.

As discussion on the ‘payment for order flow’ model rages amongst market participants, regulators, and financial media, a subtle yet powerful change in trading disclosures has the capacity to deliver a range of positive outcomes without the implementation of disruptive market reforms. At the core of the debate is the price improvement (PI) metric, currently calculated by broker/dealers as the difference between the execution price and the best quoted market price. In the US, where off-exchange trading is permitted, the improvement is measured against NBBO reported by consolidated tape operators. The liquidity hidden between the best quotes is significant enough to allow brokers to not only fulfill optimal execution requirements but also route customer orders to liquidity providers that pay the highest rebates. Compared to the best quote, PI is rarely a a negative number, essentially misleading investors by failing to report on the built-in trading costs. For retail investors, a small change in the formula - specifically, to replace the best bid or offer with a mid-quote - is likely to result in negative price improvements raising awareness on hidden commissions and rebates. For institutional investors, a change in the formula is unlikely to given that non-retail orders are typically not routed to liquidity providers. Moreover, institutional investors utilize a variety of techniques to control trading costs and achieve best execution, using so called Transaction Cost Analysis (TCA). TCA offers a way to detect inefficient orders by benchmarking execution prices versus other market participant prices, taking into account order size, liquidity depth, market direction, and meta-order parameters.

Chart 1

Inefficient sell order

Example of an inefficient sell order, which not only crossed the spread but broke through multiple levels of the limit order book, followed by a quick recovery.

Execution Analysis

In a basic implementation, one needs to compare one’s own trade prices with the VWAP of other participants over a predefined time interval or over a predefined volume after each trade. This requires access to trade reports in raw format, as well as to a public trade data feed log with exact timestamps and trade numbers. SIP trade feed would be sufficient to accomplish this task. To eliminate market impact, one’s own trades are excluded from the public history to calculate competitor VWAP over a pre- defined time window. As an alternative to a defined time window, competitor trades can be included for the same or greater volume as own trades adjusted by the volume participation parameter. In a more advanced implementation, competitor VWAP is calculated only for orders of the same direction. This enables a user to compare own buy trades with buy trades of other participants, correspondingly. For this type of analysis, a full order log that relates orders to trades is required. Full order logs such as NYSE Integrated Data Feed are specific to each exchange and are not available from SIP. In the instance that a public trade log is not available, it is possible to perform approximate calculations using OHLC bars of up to 1 minute as a proxy for competitor VWAP. If the bar history provides VWAP parameter, one can subtract total value of own trades and re-calculate competitor VWAP to improve comparison.

Table 1


Differences between execution prices and competitor VWAPs can be aggregated to report cumulative results for each order and meta-order.

Note: Trade analysis applies only to filled or partially filled orders. Cancelled orders are out of scope.

Placement Analysis

The placement analysis looks at order parameters (price, quantity, type, condition, venue, timing) on one hand, and characteristics of visible liquidity on the other. The purpose of this exercise is to detect routing latencies and order misconfigurations, such as oversize market orders or limit orders that break multiple price levels. Rules for optimal order placement can be codified in a rule book, with each rule assigned a numeric score.

Sample rules:

  1. Market order size too large.
  2. Limit order size too large (information leak).
  3. Limit order price too aggressive (spread crossed or filled at multiple prices).
  4. Ice order filled where basic limit order was sufficient (extra fees).
  5. Cross order cancellation (extra fees).
  6. Better price visible at another venue.

For basic placement analysis a SIP quote feed is sufficient. For some rules a proprietary Level 2 data feed or even full order log is required to reconstruct order book at the time of order arrival.

Audit and Reporting

Both types of analysis can be used to produce reports on execution quality, on a regular basis. By ascribing metadata to each order (such as algorithm type and version, account and broker references, etc.) such reports can locate underperforming trading strategies and reduce trading costs on a daily basis, especially for high-turnover portfolios.

> For buy-side trading desks, TCA is key to optimizing execution in today's rapidly changing capital markets. - Sergei Rodionov, CEO, Axibase

To accomplish this objective in a way that works across all OMS integrations and asset classes, customers often have to embark on a build-your-own solution path. With the addition of TCA functionality to Axibase Time-Series Database, customers can now avoid ongoing costs and rely on an industrial solution that incorporates best practices for order placement, provides integrations with various quote/trade formats, and includes a scheduled reporting engine with support for SQL.

Related Articles