arXiv subjects

Andrew Green

Publications and source records attributed to Andrew Green.

At least 19 recordsLinked to original sources

arXiv2023

LitSumm: Large language models for literature summarisation of non-coding RNAs

Curation of literature in life sciences is a growing challenge. The continued increase in the rate of publication, coupled with the relatively fixed number of curators worldwide presents a major challenge to developers of biomedical knowledgebases. Very few knowledgebases have resources to scale to the whole relevant literature and all have to prioritise their efforts. In this work, we take a first step to alleviating the lack of curator time in RNA science by generating summaries of literature for non-coding RNAs using large language models (LLMs). We demonstrate that high-quality, factually accurate summaries with accurate references can be automatically generated from the literature using a commercial LLM and a chain of prompts and checks. Manual assessment was carried out for a subset of summaries, with the majority being rated extremely high quality. We apply our tool to a selection of over 4,600 ncRNAs and make the generated summaries available via the RNAcentral resource. We conclude that automated literature summarization is feasible with the current generation of LLMs, provided careful prompting and automated checking are applied.

q-bio.GN↗

arXiv2018

Deeply Learning Derivatives

This paper uses deep learning to value derivatives. The approach is broadly applicable, and we use a call option on a basket of stocks as an example. We show that the deep learning model is accurate and very fast, capable of producing valuations a million times faster than traditional models. We develop a methodology to randomly generate appropriate training data and explore the impact of several parameters including layer width and depth, training data quality and quantity on model speed and accuracy.

q-fin.CP↗

arXiv2017

The Asymmetric Bowshock/PWN of PSR J2124$-$3358

We describe new measurements of the remarkable H$α$/UV/X-ray bowshock and pulsar wind nebula of the isolated millisecond pulsar PSR J2124$-$3358. {\it CXO} imaging shows a one-sided jet structure with a softer equatorial outflow. KOALA IFU spectroscopy shows that non-radiative emission dominates the bow shock and that the H$α$ nebula is asymmetric about the pulsar velocity with an elongation into the plane of the sky. We extend analytic models of the contact discontinuity to accommodate such shapes and compare these to the data. Using {\it HST} UV detections of the pulsar and bow shock, radio timing distance and proper motion measurements and the {\it CXO}-detected projected spin axis we model the 3-D PWN momentum flux distribution. The integrated momentum flux depends on the ionization of the ambient ISM, but for an expected ambient WNM we infer $I = 2.4 \times 10^{45} {\rm g\,cm^2}$. This implies $M_{NS} = 1.6-2.1 M_\odot$, depending on the equation of state, which in turn suggests that the MSP gained significant mass during recycling and then lost its companion. However, this conclusion is at present tentative, since lower ionization allows $\sim 30\%$ lower masses and uncertainty in the parallax allows up to 50% error.

astro-ph.HE↗

arXiv2016

XVA at the Exercise Boundary

XVA is a material component of a trade valuation and hence it must impact the decision to exercise options within a given netting set. This is true for both unsecured trades and secured / cleared trades where KVA and MVA play a material role even if CVA and FVA do not. However, this effect has frequently been ignored in XVA models and indeed in exercise decisions made by option owners. This paper describes how XVA impacts the exercise decision and how this can be readily evaluated using regression techniques (Longstaff and Schwartz 2001). The paper then assesses the materiality of the impact of XVA at the exercise boundary on swaption examples.

q-fin.PR↗

arXiv2016

Option-Based Pricing of Wrong Way Risk for CVA

The two main issues for managing wrong way risk (WWR) for the credit valuation adjustment (CVA, i.e. WW-CVA) are calibration and hedging. Hence we start from a novel model-free worst-case approach based on static hedging of counterparty exposure with liquid options. We say "start from" because we demonstrate that a naive worst-case approach contains hidden unrealistic assumptions on the variance of the hazard rate (i.e. that it is infinite). We correct this by making it an explicit (finite) parameter and present an efficient method for solving the parametrized model optimizing the hedges. We also prove that WW-CVA is theoretically, but not practically, unbounded. The option-based hedges serve to significantly reduce (typically halve) practical WW-CVA. Thus we propose a realistic and practical option-based worst case CVA.

q-fin.PR↗

arXiv2016

Time Evolution and Deterministic Optimisation of Correlator Product States

We study a restricted class of correlator product states (CPS) for a spin-half chain in which each spin is contained in just two overlapping plaquettes. This class is also a restriction upon matrix product states (MPS) with local dimension $2^n$ ($n$ being the size of the overlapping regions of plaquettes) equal to the bond dimension. We investigate the trade-off between gains in efficiency due to this restriction against losses in fidelity. The time-dependent variational principle formulated for these states is numerically very stable. Moreover, it shows significant gains in efficiency compared to the naively related matrix product states - the evolution or optimisation scales as $2^{3n}$ for the correlator product states versus $2^{4n}$ for the unrestricted matrix product state. However, much of this advantage is offset by a significant reduction in fidelity. Correlator product states break the local Hilbert space symmetry by the explicit selection of a local basis. We investigate this dependence in detail and formulate the broad principles under which correlator product states may be a useful tool. In particular, we find that scaling with overlap/bond order may be more stable with correlator product states allowing a more efficient extraction of critical exponents - we present an example in which the use of correlator product states is several orders of magnitude quicker than matrix product states.

quant-ph↗

arXiv2015

Which measure for PFE? The Risk Appetite Measure, A

Potential Future Exposure (PFE) is a standard risk metric for managing business unit counterparty credit risk but there is debate on how it should be calculated. The debate has been whether to use one of many historical ("physical") measures (one per calibration setup), or one of many risk-neutral measures (one per numeraire). However, we argue that limits should be based on the bank's own risk appetite provided that this is consistent with regulatory backtesting and that whichever measure is used it should behave (in a sense made precise) like a historical measure. Backtesting is only required by regulators for banks with IMM approval but we expect that similar methods are part of limit maintenance generally. We provide three methods for computing the bank price of risk from readily available business unit data, i.e. business unit budgets (rate of return) and limits (e.g. exposure percentiles). Hence we define and propose a Risk Appetite Measure, A, for PFE and suggest that this is uniquely consistent with the bank's Risk Appetite Framework as required by sound governance.

q-fin.RM↗

arXiv2015

The SAMI Galaxy Survey: Can we trust aperture corrections to predict star formation?

In the low redshift Universe (z<0.3), our view of galaxy evolution is primarily based on fibre optic spectroscopy surveys. Elaborate methods have been developed to address aperture effects when fixed aperture sizes only probe the inner regions for galaxies of ever decreasing redshift or increasing physical size. These aperture corrections rely on assumptions about the physical properties of galaxies. The adequacy of these aperture corrections can be tested with integral-field spectroscopic data. We use integral-field spectra drawn from 1212 galaxies observed as part of the SAMI Galaxy Survey to investigate the validity of two aperture correction methods that attempt to estimate a galaxy's total instantaneous star formation rate. We show that biases arise when assuming that instantaneous star formation is traced by broadband imaging, and when the aperture correction is built only from spectra of the nuclear region of galaxies. These biases may be significant depending on the selection criteria of a survey sample. Understanding the sensitivities of these aperture corrections is essential for correct handling of systematic errors in galaxy evolution studies.

astro-ph.GA↗

arXiv2015

First Light Results from the Hermes Spectrograph at the AAT

The High Efficiency and Resolution Multi Element Spectrograph, HERMES, is a facility-class optical spectrograph for the Anglo-Australian Telescope (AAT). It is designed primarily for Galactic Archaeology, the first major attempt to create a detailed understanding of galaxy formation and evolution by studying the history of our own galaxy, the Milky Way. The goal of the GALAH survey is to reconstruct the mass assembly history of the Milky Way through a detailed chemical abundance study of one million stars. The spectrograph is based at the AAT and is fed by the existing 2dF robotic fiber positioning system. The spectrograph uses volume phase holographic gratings to achieve a spectral resolving power of 28,000 in standard mode and also provides a high-resolution mode ranging between 40,000 and 50,000 using a slit mask. The GALAH survey requires an SNR greater than 100 for a star brightness of V ?= 14 in an exposure time of one hour. The total spectral coverage of the four channels is about 100 nm between 370 and 1000 nm for up to 392 simultaneous targets within the 2-degree field of view. HERMES has been commissioned over three runs, during bright time in October, November, and December 2013, in parallel with the beginning of the GALAH pilot survey, which started in November 2013. We present the first-light results from the commissioning run and the beginning of the GALAH survey, including performance results such as throughput and resolution, as well as instrument reliability.

astro-ph.IM↗

arXiv2015

Dirac Processes and Default Risk

We introduce Dirac processes, using Dirac delta functions, for short-rate-type pricing of financial derivatives. Dirac processes add spikes to the existing building blocks of diffusions and jumps. Dirac processes are Generalized Processes, which have not been used directly before because the dollar value of non-Real numbers is meaningless. However, short-rate pricing is based on integrals so Dirac processes are natural. This integration directly implies that jumps are redundant whilst Dirac processes expand expressivity of short-rate approaches. Practically, we demonstrate that Dirac processes enable high implied volatility for CDS swaptions that has been otherwise problematic in hazard rate setups.

q-fin.PR↗

arXiv2015

MVA: Initial Margin Valuation Adjustment by Replication and Regression

Initial margin requirements are becoming an increasingly common feature of derivative markets. However, while the valuation of derivatives under collateralisation (Piterbarg 2010, Piterbarg2012), under counterparty risk with unsecured funding costs (FVA) (Burgard2011, Burgard2011, Burgard2013) and in the presence of regulatory capital (KVA) (Green2014) are established through valuation adjustments, hitherto initial margin has not been considered. This paper further extends the semi-replication framework of (Burgard2013a), itself later extended by (Green2014), to cover the cost of initial margin, leading to Margin Valuation Adjustment (MVA). Initial margin requirements are typically generated through the use of VAR or CVAR models. Given the form of MVA as an integral over the expected initial margin profile this would lead to excessive computational costs if a brute force calculation were to be used. Hence we also propose a computationally efficient approach to the calculation of MVA through the use of regression techniques, Longstaff-Schwartz Augmented Compression (LSAC).

q-fin.PR↗

arXiv2015

Self-Financing Trading and the Ito-Doeblin Lemma

The objective of the note is to remind readers on how self-financing works in Quantitative Finance. The authors have observed continuing uncertainty on this issue which may be because it lies exactly at the intersection of stochastic calculus and finance. The concept of a self-financing trading strategy was originally, and carefully, introduced in (Harrison and Kreps 1979) and expanded very generally in (Harrison and Pliska 1981).

q-fin.PR↗

arXiv2015

Warehousing Credit (CVA) Risk, Capital (KVA) and Tax (TVA) Consequences

Credit risk may be warehoused by choice, or because of limited hedging possibilities. Credit risk warehousing increases capital requirements and leaves open risk. Open risk must be priced in the physical measure, rather than the risk neutral measure, and implies profits and losses. Furthermore the rate of return on capital that shareholders require must be paid from profits. Profits are taxable and losses provide tax credits. Here we extend the semi-replication approach of Burgard and Kjaer (2013) and the capital formalism (KVA) of Green, Kenyon, and Dennis (2014) to cover credit risk warehousing and tax, formalized as double-semi-replication and TVA (Tax Valuation Adjustment) to enable quantification.

q-fin.PR↗

arXiv2014

Efficient XVA Management: Pricing, Hedging, and Attribution using Trade-Level Regression and Global Conditioning

Banks must manage their trading books, not just value them. Pricing includes valuation adjustments collectively known as XVA (at least credit, funding, capital and tax), so management must also include XVA. In trading book management we focus on pricing, hedging, and allocation of prices or hedging costs to desks on an individual trade basis. We show how to combine three technical elements to radically simplify XVA management, both in terms of the calculations, and the implementation of the calculations. The three technical elements are: trade-level regression; analytic computation of sensitivities; and global conditioning. All three are required to obtain the radical efficiency gains and implementation simplification. Moreover, many of the calculations are inherently parallel and suitable for GPU implementation. The resulting methodology for XVA management is sufficiently general that we can cover pricing, first- and second-order sensitivities, and exact trade-level allocation of pricing and sensitivities within the same framework. Managing incremental changes to portfolios exactly is also radically simplified.

q-fin.CP↗

arXiv2014

KVA: Capital Valuation Adjustment

Credit (CVA), Debit (DVA) and Funding Valuation Adjustments (FVA) are now familiar valuation adjustments made to the value of a portfolio of derivatives to account for credit risks and funding costs. However, recent changes in the regulatory regime and the increases in regulatory capital requirements has led many banks to include the cost of capital in derivative pricing. This paper formalises the addition of cost of capital by extending the Burgard-Kjaer (2013) semi-replication approach to CVA and FVA to include an addition capital term, Capital Valuation Adjustment (KVA, i.e. Kapital Valuation Adjustment to distinguish from CVA.) The utilization of the capital for funding purposes is also considered. The use of the semi-replication approach means that the flexibility around the treatment of self-default is carried over into this analysis. The paper further considers the practical calculation of KVA with reference to the Basel II (BCBS-128) and Basel III (BCBS-189) capital regimes and their implementation via CRD IV. The paper also assesses how KVA may be hedged, given that any hedging transactions themselves lead to regulatory capital requirements and hence capital costs. Finally a number of numerical examples are presented to gauge the cost impact of KVA on vanilla derivative products.

q-fin.PR↗

arXiv2014

Regulatory-Optimal Funding

Funding is a cost to trading desks that they see as an input. Current FVA-related literature reflects this by also taking funding costs as an input, usually constant, and always risk-neutral. However, this funding curve is the output from a Treasury point of view. Treasury must consider Regulatory-required liquidity buffers, and both risk-neutral (Q) and physical measures (P). We describe the Treasury funding problem and optimize against both measures, using the Regulatory requirement as a constraint. We develop theoretically optimal strategies for Q and P, then demonstrate a combined approach in four markets (USD, JPY, EUR, GBP). Since we deal with physical measures we develop appropriate statistical tests, and demonstrate highly significant (p<0.00001), out-of-sample, improvements on hedged funding with a combined approach achieving 44% to 71% of a perfect information criterion. Thus regulatory liquidity requirements change both the funding problem and funding costs.

q-fin.PR↗

arXiv2014

Regulatory-Compliant Derivatives Pricing is Not Risk-Neutral

Regulations impose idiosyncratic capital and funding costs for holding derivatives. Capital requirements are costly because derivatives desks are risky businesses; funding is costly in part because regulations increase the minimum funding tenor. Idiosyncratic costs mean no single measure makes derivatives martingales for all market participants. Hence Regulatory-compliant pricing is not risk-neutral. This has implications for exit prices and mark-to-market.

q-fin.PR↗

arXiv2014

VAR and ES/CVAR Dependence on data cleaning and Data Models: Analysis and Resolution

Historical (Stressed-) Value-at-Risk ((S)VAR), and Expected Shortfall (ES), are widely used risk measures in regulatory capital and Initial Margin, i.e. funding, computations. However, whilst the definitions of VAR and ES are unambiguous, they depend on input distributions that are data-cleaning- and Data-Model-dependent. We quantify the scale of these effects from USD CDS (2004--2014), and from USD interest rates (1989--2014, single-curve setup before 2004, multi-curve setup after 2004), and make two standardisation proposals: for data; and for Data-Models. VAR and ES are required for lifetime portfolio calculations, i.e. collateral calls, which cover a wide range of market states. Hence we need standard, i.e. clean, complete, and common (i.e. identical for all banks), market data also covering this wide range of market states. This data is historically incomplete and not clean hence data standardization is required. Stressed VAR and ES require moving market movements during a past (usually not recent) window to current, and future, market states. All choices (e.g. absolute difference, relative, relative scaled by some function of market states) implicitly define a Data Model for transformation of extreme market moves (recall that 99th percentiles are typical, and the behaviour of the rest is irrelevant). Hence we propose standard Data Models. These are necessary because different banks have different stress windows. Where there is no data, or a requirement for simplicity, we propose standard lookup tables (one per window, etc.). Without this standardization of data and Data Models we demonstrate that VAR and ES are complex derivatives of subjective choices.

q-fin.RM↗