Copyright © 2022 Pragma – Confidential pragmatrading.com
FX Passive Order
Analysis
V E R S I O N : 1 . 0
D A T E : 7 / 2 7 / 2 2
FX Passive Order
Analysis
V E R S I O N : 1 . 0
D A T E : 7 / 2 7 / 2 2
Copyright © 2022 Pragma – Confidential 2
* C.A. #215
Introduction
Venue selection is an important part of trading in FX markets where a variety of ECNs and liquidity providers
compete for client flow. An FX algo strategy that attempts to capture spread by placing passive orders must
consider the quality of liquidity available at a particular venue and how much quantity to commit to that venue in
order to maximize the probability of getting a fill at a favorable price. The strategy must also balance trading at the
“primary” ECNs, EBS and Reuters, that impose 1M order minimums and secondary ECNs that allow trading in
smaller sizes.
In this study we analyze over 500,000 passive order chains placed by clients via FX TWAP and VWAP strategies for a
variety of directly traded currency pairs over a period of two years from February 2020 to January 2022. By
measuring differences in arrival slippage, we observe that trading in smaller sizes results in price improvement on
the order of 7% of spread over 1M orders. For the clients that prefer trading in larger sizes, we also note that for
1M orders, trading on secondary ECNs outperforms primary ECNs for both EBS and Reuters currency pairs.
Methodology
An FX strategy that follows a pre-determined trajectory such as TWAP or VWAP may attempt to execute passively
by placing a passive order at the reference near-touch price on an ECN. If the market moves away, the passive
order will re-peg by modifying the limit price or canceling and sending an updated order until it either gets filled or
the strategy has to cancel it and cross the spread in order to keep up with the trajectory. We consider the full
sequence of events from the initial placement through all the re-pegs to the final fill or cancel as a single passive
order chain and measure its price performance relative to mid at arrival, i.e. the time of original order placement.
Since not all such passive chains result in a full fill, as a cleanup cost, we use the far-touch price at time of cancel
for any quantity that was not filled. In order to compare performance across pairs, we normalize slippage by the
average reference bid-ask spread. In particular, the performance metric is defined as follows:
𝑆𝑙𝑖𝑝𝑝𝑎𝑔𝑒 = (1 𝑓𝑜𝑟 𝑏𝑢𝑦, −1 𝑓𝑜𝑟 𝑠𝑒𝑙𝑙) × 𝐶ℎ𝑎𝑖𝑛𝐹𝑖𝑙𝑙𝑃𝑟𝑖𝑐𝑒 – 𝐴𝑟𝑟𝑖𝑣𝑎𝑙𝑀𝑖𝑑𝑝𝑜𝑖𝑛𝑡𝑃𝑟𝑖𝑐𝑒
𝐴𝑣𝑔𝐵𝑖𝑑𝐴𝑠𝑘𝑆𝑝𝑟𝑒𝑎𝑑
Note, smaller values of slippage indicate better performance.
Order quantity and venue are chosen randomly by the strategy prior to order placement, which enables a fair,
apples-to-apples comparison of chains across different venues and order sizes. When comparing the performance
of primary ECNs to secondary ECNs, only orders for 1M, which are eligible to trade on both types of venues, are
considered.
* C.A. #215
Introduction
Venue selection is an important part of trading in FX markets where a variety of ECNs and liquidity providers
compete for client flow. An FX algo strategy that attempts to capture spread by placing passive orders must
consider the quality of liquidity available at a particular venue and how much quantity to commit to that venue in
order to maximize the probability of getting a fill at a favorable price. The strategy must also balance trading at the
“primary” ECNs, EBS and Reuters, that impose 1M order minimums and secondary ECNs that allow trading in
smaller sizes.
In this study we analyze over 500,000 passive order chains placed by clients via FX TWAP and VWAP strategies for a
variety of directly traded currency pairs over a period of two years from February 2020 to January 2022. By
measuring differences in arrival slippage, we observe that trading in smaller sizes results in price improvement on
the order of 7% of spread over 1M orders. For the clients that prefer trading in larger sizes, we also note that for
1M orders, trading on secondary ECNs outperforms primary ECNs for both EBS and Reuters currency pairs.
Methodology
An FX strategy that follows a pre-determined trajectory such as TWAP or VWAP may attempt to execute passively
by placing a passive order at the reference near-touch price on an ECN. If the market moves away, the passive
order will re-peg by modifying the limit price or canceling and sending an updated order until it either gets filled or
the strategy has to cancel it and cross the spread in order to keep up with the trajectory. We consider the full
sequence of events from the initial placement through all the re-pegs to the final fill or cancel as a single passive
order chain and measure its price performance relative to mid at arrival, i.e. the time of original order placement.
Since not all such passive chains result in a full fill, as a cleanup cost, we use the far-touch price at time of cancel
for any quantity that was not filled. In order to compare performance across pairs, we normalize slippage by the
average reference bid-ask spread. In particular, the performance metric is defined as follows:
𝑆𝑙𝑖𝑝𝑝𝑎𝑔𝑒 = (1 𝑓𝑜𝑟 𝑏𝑢𝑦, −1 𝑓𝑜𝑟 𝑠𝑒𝑙𝑙) × 𝐶ℎ𝑎𝑖𝑛𝐹𝑖𝑙𝑙𝑃𝑟𝑖𝑐𝑒 – 𝐴𝑟𝑟𝑖𝑣𝑎𝑙𝑀𝑖𝑑𝑝𝑜𝑖𝑛𝑡𝑃𝑟𝑖𝑐𝑒
𝐴𝑣𝑔𝐵𝑖𝑑𝐴𝑠𝑘𝑆𝑝𝑟𝑒𝑎𝑑
Note, smaller values of slippage indicate better performance.
Order quantity and venue are chosen randomly by the strategy prior to order placement, which enables a fair,
apples-to-apples comparison of chains across different venues and order sizes. When comparing the performance
of primary ECNs to secondary ECNs, only orders for 1M, which are eligible to trade on both types of venues, are
considered.
