
Nick FiroozyeUniversity College London | UCL · Department of Computer Science
Nick Firoozye
PhD (Maths), Financial Stats/Econometrics/ML current
About
79
Publications
27,309
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
462
Citations
Introduction
Mathematician and former Assistant Prof (UIUC), Financial Econometrician, Statistician, Machine Learning expert and Bond Market Strategist and Sovereign Debt Crisis Specialist (Wolfson Prize Shortlister). Emphasis on time-series methods, credit, law and finance, uncertainty and risk.
Currently, Honorary Sr Lecturer at UCL, working on several papers on Algorithmic Trading and Automatic Trade Idea Recommendation with Adriano Koshiyama, as well as teaching two online classes on Algorithmic Trading Strategies (at Experfy and Quantshub) and one in-person class at UCL (together with Paolo Barucca) this January. I plan some more online courses in due course.
Education
July 1988 - July 1990
January 1982 - June 1986
Publications
Publications (79)
We extend the research into cross-sectional momentum trading strategies. Our main result is our novel ranking algorithm, the naive Bayes asset ranker (nbar), which we use to select subsets of assets to trade from the S&P 500 index. We perform feature representation transfer from radial basis function networks to a curds and whey (caw) multivariate...
We demonstrate a novel application of online transfer learning for a digital assets trading agent. This agent uses a powerful feature space representation in the form of an echo state network, the output of which is made available to a direct, recurrent reinforcement learning agent. The agent learns to trade the XBTUSD (Bitcoin versus US Dollars) p...
Financial time series are both autocorrelated and nonstationary, presenting modelling challenges that violate the independent and identically distributed random variables assumption of most regression and classification models. The prediction with expert advice framework makes no assumptions on the data-generating mechanism yet generates prediction...
We present a novel framework for analyzing the optimal asset and signal combination problem. We reformulate the original problem of portfolio selection from a set of correlated assets and signals into one of selecting from a set of uncorrelated trading strategies through the lens of Canonical Correlation Analysis of Hotelling (1936). The new enviro...
We demonstrate an application of online transfer learning as a digital assets trading agent. This agent makes use of a powerful feature space representation in the form of an echo state network, the output of which is made available to a direct, recurrent reinforcement learning agent. The agent learns to trade the XBTUSD (Bitcoin versus US dollars)...
We explore online inductive transfer learning, with a feature representation transfer from a radial basis function network formed of Gaussian mixture model hidden processing units to a direct, recurrent reinforcement learning agent. This agent is put to work in an experiment, trading the major spot market currency pairs, where we accurately account...
We explore online inductive transfer learning, with a feature representation transfer from a radial basis function network formed of Gaussian mixture model hidden processing units to a direct, recurrent reinforcement learning agent. This agent is put to work in an experiment, trading the major spot market currency pairs, where we accurately account...
Systematic financial trading strategies account for over 80% of trade volume in equities and a large chunk of the foreign exchange market. In spite of the availability of data from multiple markets, current approaches in trading rely mainly on learning trading strategies per individual market. In this paper, we take a step towards developing fully...
We conduct a detailed experiment on major cash fx pairs, accurately accounting for transaction and funding costs. These sources of profit and loss, including the price trends that occur in the currency markets, are made available to our recurrent reinforcement learner via a quadratic utility, which learns to target a position directly. We improve u...
We conduct a detailed experiment on major cash fx pairs, accurately accounting for transaction and funding costs. These sources of profit and loss, including the price trends that occur in the currency markets, are made available to our recurrent reinforcement learner via a quadratic utility, which learns to target a position directly. We improve u...
We conduct a detailed experiment on major cash fx pairs, accurately accounting for transaction and funding costs. These sources of profit and loss, including the price trends that occur in the currency markets, are made available to our recurrent reinforcement learner via a quadratic utility, which learns to target a position directly. We improve u...
We conduct a detailed experiment on major cash fx pairs, accurately accounting for transaction and funding costs. These sources of profit and loss, including the price trends that occur in the currency markets, are made available to our recurrent reinforcement learner via a quadratic utility, which learns to target a position directly. We improve u...
We investigate the benefits of feature selection, nonlinear modelling and online learning when forecasting in financial time series. We consider the sequential and continual learning sub-genres of online learning. The experiments we conduct show that there is a benefit to online transfer learning, in the form of radial basis function networks, beyo...
This paper reviews Artificial Intelligence (AI), Machine Learning (ML) and associated algorithms in future Capital Markets. New AI algorithms are constantly emerging, with each 'strain' mimicking a new form of human learning, reasoning, knowledge, and decisionmaking. The current main disrupting forms of learning include Deep Learning, Adversarial L...
Systematic trading strategies are algorithmic procedures that allocate assets aiming to optimize a certain performance criterion. To obtain an edge in a highly competitive environment, an analyst needs to appropriately fine-tune their strategy, or discover how to combine weak signals in novel alpha creating manners. Both aspects, namely fine-tuning...
In this work we introduce QuantNet: an architecture that is capable of transferring knowledge over systematic trading strategies in several financial markets. By having a system that is able to leverage and share knowledge across them, our aim is two-fold: to circumvent the so-called Backtest Overfitting problem; and to generate higher risk-adjuste...
Dynamic trading strategies, in the spirit of trend-following or mean-reversion, represent an only partly understood but lucrative and pervasive area of modern finance. Assuming Gaussian returns and Gaussian dynamic weights or signals, (e.g., linear filters of past returns, such as simple moving averages, exponential weighted moving averages, foreca...
Systematic trading strategies are rule-based procedures which choose portfolios and allocate assets. In order to attain certain desired return proles, quantitative strategists must determine a large array of trading parameters. Backtesting, the attempt to identify the appropriate parameters using historical data available, has been highly criticize...
Dynamic trading strategies, in the spirit of trend-following or mean-reversion, represent an only partly understood but lucrative and pervasive area of modern nance. Assuming Gaussian returns and Gaussian dynamic weights or signals, (e.g., linear lters of past returns, such as simple moving averages, exponential weighted moving averages, forecasts...
Systematic trading strategies are rule-based procedures which choose portfolios and allocate assets. In order to attain certain desired return profiles, quantitative strategists must determine a large array of trading parameters. Backtesting, the attempt to identify the appropriate parameters using historical data available, has been highly critici...
Derivative traders are usually required to scan through hundreds, even thousands of possible trades on a daily basis. Up to now, not a single solution is available to aid in their job. Hence, this work is aimed to develop a trading recommendation system, and to apply this system to the so‐called Mid‐Curve Calendar Spread (MCCS) trade. To suggest th...
Dynamic trading strategies, in the spirit of trend-following or mean reversion, rep-resent an only partly understood but lucrative and pervasive area of modern finance. By assuming Gaussian returns and Gaussian dynamic weights or “signals” (eg, linear filters of past returns, such as simple moving averages, exponential weighted moving averages and...
Systematic trading strategies are algorithmic procedures that allocate assets aiming to optimize a certain performance criterion. To obtain an edge in a highly competitive environment, the analyst needs to proper fine-tune its strategy, or discover how to combine weak signals in novel alpha creating manners. Both aspects, namely fine-tuning and com...
Dynamic trading strategies, in the spirit of trend-following or mean-reversion, represent an only partly understood but lucrative and pervasive area of modern finance. Assuming Gaussian returns, we are able to derive closed-form expressions for the first four moments of the strategy’s returns, in terms of correlations between the random signals and...
Derivative traders are usually required to scan through hundreds, even thousands of possible trades on a daily basis. Up to now, not a single solution is available to aid in their job. Hence, this work aims to develop a trading recommendation system, and apply this system to the so-called Mid-Curve Calendar Spread (MCCS), an exotic swaption-based d...
Derivative traders are usually required to scan through hundreds, even thousands of possible trades on a daily-basis; a concrete case is the so-called Mid-Curve Calendar Spread (MCCS). The actual procedure in place is full of pitfalls and a more systematic approach where more information at hand is crossed and aggregated to find good trading picks...
Creating a Long vega index as a cheaper way of hedging tail risk. Comparing to tail-risk (most long vol) alternatives, it outperforms in almost all metrcis.
• Many so-called “liquidity premia” are just beta in disguise
• A robust, genuinely different risk premium for providing liquidity does exist
• Providing liquidity earns this risk premium, much like market makers do
• It is equivalent to mean reversion positioning as is well documented in academic work
• It is different from beta and other risk pre...
Managing Uncertainty, Mitigating Risk proposes that financial risk management broaden its approach, maintaining quantification where possible, but incorporating uncertainty. The author shows that by using broad quantification techniques, and using reason as the guiding principle, practitioners can see a more holistic and complete picture. © Nick B....
We have discussed both the types of unknown and the sources of knowledge which can be used to characterise and mitigate the effects of the unknown. The last chapter of Part II on the essentials covers the mathematical underpinnings of uncertainty, and its quantification, leading the way to its ultimate management. As we saw in Chapter 4, probabilit...
As we discussed in Chapter 9, we will be using knowledge sources which are broader than mere data, combined with new methods which are appropriate to dealing with wider senses of the unknown — structural models which allow us to examine deeply each risk scenario. In effect we model their drivers and their consequences. Together, this should allow u...
We begin by setting out the historical landscape for the management of risk and uncertainty in banking. Uncertainty has fallen to the wayside despite being acknowledged by many of the leading mathematicians over the years from Bernoulli to Kolmogorov and the founders of all the major schools of economics including Keynes, Hayek and Friedman. In tod...
While we have introduced risk and its mathematics, probability, its history and impact on modern finance and financial risk management, all areas which are generally well known, the need to broaden the discussion to include a wider range of the unknown requires a significant departure from previously used methods. In particular, we need to define a...
We present a short case study on the eurozone sovereign debt crisis concentrating on the Greek exit threat in November 2011. The method was outlined in some detail in Chapter 10. We do not wish to rehash the method again here, but rather to demonstrate how the method can be applied to that specific episode in the ongoing Greek crisis. We will illus...
Much of the risk dealt with in banking is Knightian risk, which is by definition the preserve of probability theory. Probability, with its roots in games of chance,1 and statistics, with its roots in the analysis of government census data,2 have both been applied in modern finance with great success. Meanwhile, they have developed independently as...
Risk management has evolved considerably from its introduction and will continue to evolve as the industry changes and the demands that are placed on risk managers grow. Risk management has tackled a great many areas of risk, but systemic risk and one-offs are a new chapter and require new methods for risk managers to assess and mitigate. While unc...
As we have seen in Chapter 2, while academics is including Keynes, Hayek and Friedman saw risk and uncertainty from a more holistic point of view, modern finance has been dominated by quantitative methods which focus predominantly on measuring risk. Risk is easily quantifiable and easily transferrable (in whole or in part), and the industry’s succe...
Uncertainty is often neglected as a topic of study, especially among those with quantitative backgrounds. It is not easily quantified, and thus it is not easy to put into modelling frameworks. A significant number of financial outcomes are dictated by uncertainty, rather than probability. Elections are often modelled using probability, while board...
So far we have discussed the various definitions of risk together with classifications of the unknown, the historical evolution of risk modelling and how risks are managed in banking and finance. We showed that qualitative inputs such as expertise and insight are consistently disregarded, thus making it difficult to acknowledge and assess the possi...
Probability, statistics and stochastic differential equations have had considerable success in finance and economics. We have discussed the evolution of the concept of risk in modern finance in Chapter 2. Here we will touch on many of the major modelling-based developments in modern risk and finance and their statistical underpinnings. We present a...
Eurozone break-up risk has risen notably over the past few months, as European policy makers have failed to put in place a credible backstop for the larger Eurozone bond markets. Given this increased risk, investors should pay close attention to the ‘redenomination risk’ of various assets. There are important legal dimensions to this risk, includin...
The CDO at the heart of the Eurozone 2 9 J U N E 2 0 1 1
The announcement of the expansion of the loan capacity of the European Financial Stability Facility was accompanied by an update to the entity structure. The new structure includes a greater level of guarantees allowing for the removal of the cash collateral requirement, a previous requ...
The EFSF achieved an AAA rating on 20 September by making a Faustian deal of agreeing to use only the AAA guarantees and a substantial cash-buffer, effectively limiting the size of any loan facilities dramatically. While this appears to us to be highly suboptimal and overlooks the political achievements of the non-AAA countries in passage of the pa...
We present an extension of Wiener's theorem, a nonlinear filter called the measure filter which, when applied to a measure, filters out all but the lowest dimensional part of that measure. We also extend the work of Strichartz in the case of codimension one, for functions of bounded variation, by allowing for a wide range of kernels which can be us...
It is well known that in two space dimensions, if a solution to Poisson’s equation has right-hand side in H loc 1 \mathcal {H}_{{\text {loc}}}^1 , then this solution is actually continuous. The corresponding result for n -Laplacians is shown to be false for n ≥ 3 n \geq 3 ; we construct two examples with right-hand sides in H loc 1 ( ℜ n ) \mathcal...
It is well known that in two space dimensions, if a solution to Poisson's equation has right-hand side in $\mathscr{H}^1_\operatorname{loc}$, then this solution is actually continuous. The corresponding result for n-Laplacians is shown to be false for n ≥ 3; we construct two examples with right-hand sides in $\mathscr{H}^1_\operatorname{loc}(\mathf...
We consider the following question: given a set of matrices with no rank-one connections, does it support a nontrivial Young measure limit of gradients? Our main results are these: (a) a Young measure can be supported on four incompatible matrices; (b) in two space dimensions, a Young measure cannot be supported on finitely many incompatible elasti...
This paper discusses the relaxation of a multiwell energy of the special form W = mini{|▽u − ai|2}. We explain how the relaxation QW can be expressed in terms of certain “tensors of geometric parameters.” The exact set ℱθ of attainable geometric parameters is not known, but we show that it must lie inside an explicitly given convex set ℱ
θ
U
. This...
The translation method has been used with great success in bounding the effective moduli of composite materials. We consider here the analogous method for bounding the relaxations of variational problems. We optimize the bound over the set of all available translations. Our method is to cast this in the form of a minmax problem. Using techniques of...