Podcast: NYU’s Kolm on transaction costs and machine learning
TCA methodologies that ignore partial fills “might be off by 20% to 30%”
TCA methodologies that ignore partial fills “might be off by 20% to 30%”
Most forms of post-trade transaction cost analysis only consider the price impact of completed orders. But ignoring partially filled orders – which are all too common when trading – produces a distorted measure of execution quality.
“Depending on what methodologies are used, you might be off by 20% to 30%, relative to the true transaction cost,” says Petter Kolm, professor of finance and director of the Mathematics in Finance master’s program at NYU’s Courant Institute of Mathematical Sciences, and our guest for this episode of Quantcast.
Kolm’s latest paper with Nicholas Westray, a visiting researcher in financial machine learning at the Courant Institute, explores the so-called clean-up costs of trades, which they define as the opportunity cost attributed to the part of the order that is unfilled.
Most trading firms use ad hoc techniques to measure the cost of partial fills. The paper proposes a streamlined way to quantify clean-up costs that can be consistently applied to different trading strategies. The setup assumes the market behaves like a propagator model. This allows for the transaction costs of partially filled orders to be modelled as if they were fully executed, capturing the effects on drift of the security as well as the market impact of the trade.
In this podcast, Kolm also discusses his other research interests, including the applications of machine learning and its various branches in finance, one of which is natural language processing (NLP). Kolm and his team have used NLP to gauge investor sentiment on individual stocks by harvesting signals from financial news. Their research has shown that there is indeed a connection between sentiment and the successive behaviour of the stock.
Kolm is also working on various applications of reinforcement learning, which is becoming increasingly popular. He is, however, more cautious than other quants about it application in finance. While the technique is promising, he warns that prior applications such as the Alpha Go system developed by Google’s Deepmind benefited from a large and stable database for training. In finance, quants only have a limited history of prices to work with. “Reinforcement learning has had a bit of hype – all the cool kids are doing it these days, but I think people are starting to understand and separate hype from reality,” he says.
Kolm says his future projects will focus on the application of deep learning and reinforcement learning to optimal execution and the trading of American options, as well as the use of NLP to generate trading signals.
Index
00:00 Intro and transaction cost analysis
02:30 How costly are clean-up costs?
05:10 The problem of quantifying clean-up costs
10:32 Reading financial news with NLP
16:50 Reinforcement learning
22:40 Deep learning and limit order books
26:20 Teaching machine learning techniques
30:20 Future research projects
To hear the full interview, listen in the player above, or download. Future podcasts in our Quantcast series will be uploaded to Risk.net. You can also visit the main page here to access all tracks, or go to the iTunes store or Google Podcasts to listen and subscribe.
Now also available on Spotify.
Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.
To access these options, along with all other subscription benefits, please contact info@risk.net or view our subscription options here: http://subscriptions.risk.net/subscribe
You are currently unable to print this content. Please contact info@risk.net to find out more.
You are currently unable to copy this content. Please contact info@risk.net to find out more.
Copyright Infopro Digital Limited. All rights reserved.
You may share this content using our article tools. Printing this content is for the sole use of the Authorised User (named subscriber), as outlined in our terms and conditions - https://www.infopro-insight.com/terms-conditions/insight-subscriptions/
If you would like to purchase additional rights please email info@risk.net
Copyright Infopro Digital Limited. All rights reserved.
You may share this content using our article tools. Copying this content is for the sole use of the Authorised User (named subscriber), as outlined in our terms and conditions - https://www.infopro-insight.com/terms-conditions/insight-subscriptions/
If you would like to purchase additional rights please email info@risk.net
More on Cutting Edge
Georgios Skoufis on RFRs, convexity adjustments and Sabr
Bloomberg quant discusses his new approach for calculating convexity adjustments for RFR swaps
Infrequent MtM reduces neither value-at-risk nor backtesting exceptions
Frequency of repricing impacts volatility and correlation measures
Dynamic margining long/short equity trading strategies
A repo haircut model extends a previous solution for long-only strategies
SABR convexity adjustment for an arithmetic average RFR swap
A model-independent convexity adjustment for interest rate swaps is introduced
Joint S&P 500/VIX smile calibration in discrete and continuous time
An arbitrage-free model for exotic options that captures smiles and futures is presented
The carbon equivalence principle: minimising the cost to carbon net zero
A method to align incentives with sustainability in financial markets is introduced
The cost of mis-specifying price impact
Expected returns can be significantly affected by the wrong use of impact models
Most read
- Quants are using language models to map what causes what
- Reluctantly, CME moves to clear US Treasuries
- The bank quant who wants to stop gen AI hallucinating