As there will be a wait­ing room, please join the meet­ing at 8:45 and 13:45.
Open­ing Cer­e­mony
Morn­ing ses­sion 1
chair: An­drás Zem­pléni, ELTE
Clau­dia Klüp­pel­berg (TU Mu­nich):
Mod­el­ling risk in an agent-ob­ject mar­ket by a ran­dom bi­par­tite graph struc­ture
Morn­ing ses­sion 2
chair: Gá­bor Mol­nár-Sáska, Mor­gan Stan­ley
Zsolt Pándi (Mor­gan Stan­ley):
Emerg­ing chal­lenges in risk man­age­ment
Réka Janosik (MSCI):
A new stan­dard for risk man­age­ment
István Barra (Black­Rock):
Out­lier ro­bust volatil­ity fore­cast­ing in prac­tice
Af­ter­noon ses­sion 1
chair: Mik­lós Vörös, MSCI
Carlo Acerbi (Banque Pictet & Cie):
When Quan­ti­ta­tive Fi­nance meets Harsanyi
Árpád Szlávik (Mor­gan Stan­ley):
Model Risk Man­age­ment Evo­lu­tion
Af­ter­noon ses­sion 2
chair: Lás­zló Márkus, ELTE
Lás­zló An­tal (Black Rock):
Risk mea­sures and un­cer­tainty in port­fo­lio con­struc­tion – a case study of Sus­tain­abil­ity
Chen Zhou (De Ned­er­land­sche Bank and Eras­mus Uni­ver­sity of Rot­ter­dam):
Why risk is so hard to mea­sure?

Clau­dia Klüp­pel­berg (TU Mu­nich)

Mod­el­ling risk in an agent-ob­ject mar­ket by a ran­dom bi­par­tite graph struc­ture

We in­tro­duce a ran­dom net­work model for busi­ness re­la­tion­ships as for in­stance an in­sur­ance mar­ket, over­lap­ping port­fo­lios, or Op­er­a­tional Risk. Us­ing Pareto-tailed losses (as are ob­served for large risks) with a de­pen­dence struc­ture in­tro­duced by the graph we study sys­temic risk mea­sures, which are based on the Value-at-Risk and the Ex­pected Short­fall. We show that the de­pen­dence on the net­work struc­ture plays a fun­da­men­tal role for the in­di­vid­ual agen­t’s risk as well as for the mar­ket risk. The fo­cus of our analy­sis lies in the study of the in­flu­ence of the ran­dom graph on risk mea­sures, where we con­sider the Bernoulli graph and a Rasch-type graph as ex­am­ples. In par­tic­u­lar, we ex­plain the in­flu­ence of the net­work struc­ture on di­ver­si­fi­ca­tion in such mod­els. This is joint work with Oliver Kley and Ge­sine Rein­ert.

Chen Zhou (De Ned­er­land­sche Bank and Eras­mus Uni­ver­sity of Rot­ter­dam)

Why risk is so hard to mea­sure?

This pa­per an­a­lyzes the re­li­a­bil­ity of stan­dard ap­proaches for fi­nan­cial risk analy­sis. We fo­cus on the dif­fer­ence be­tween Value-at-Risk and ex­pected short­fall, their small sam­ple prop­er­ties, the scope for un­der­re­port­ing risk and how es­ti­ma­tion can be im­proved. Over­all, we find that risk fore­casts are ex­tremely un­cer­tain at low sam­ple sizes, with Value-at-Risk more ac­cu­rate than ex­pected short­fall. Value-at-Risk is eas­ily de­lib­er­ately un­der­re­ported with­out vi­o­lat­ing reg­u­la­tions and con­trol mech­a­nisms. Fi­nally, we dis­cuss the im­pli­ca­tions for aca­d­e­mic re­search, prac­ti­tion­ers and reg­u­la­tors, along with best prac­tice sug­ges­tions.

Carlo Acerbi (Banque Pictet & Cie)

When Quan­ti­ta­tive Fi­nance meets Harsanyi

In the 100th birth an­niver­sary of the math­e­mati­cian and Eco­nom­ics No­bel win­ner Janos Karoly Harsanyi, we ex­plore pos­si­ble fer­tile con­nec­tions be­tween some of Harsany­i’s key re­sults in game the­ory and open ques­tions in Quan­ti­ta­tive Fi­nance. Harsanyi div­i­dends, un­known to the large risk com­mu­nity pub­lic, rep­re­sent a canon­i­cal way to de­com­pose the re­sult of a co­op­er­a­tive game into co­hort con­tri­bu­tions, show­ing in­ter­fer­ence ef­fects at all or­ders. In­ter­pret­ing risk or re­turn as the game re­sult, and risk fac­tors, port­fo­lio po­si­tions or in­vest­ment de­ci­sions as the play­ers of the game, we im­me­di­ately ob­tain new par­a­digms to risk at­tri­bu­tion, per­for­mance con­tri­bu­tion and per­for­mance at­tri­bu­tion. We com­pare com­mon best prac­tices with these new par­a­digms and we high­light the chal­lenges (com­pu­ta­tional, the­o­ret­i­cal) that these meth­ods open, in the hope to spur fu­ture re­search.

Zsolt Pándi (Mor­gan Stan­ley)

Emerg­ing chal­lenges in risk man­age­ment

The en­vi­ron­ment in which fi­nan­cial ser­vice providers op­er­ate has changed sig­nif­i­cantly since the fi­nan­cial cri­sis. Risk man­age­ment had to ad­dress var­i­ous is­sues over that pe­riod mo­ti­vated in part by stricter reg­u­la­tions. How­ever, changes in the global econ­omy as well as changes in the fo­cus of global reg­u­la­tory agenda pre­sent new ques­tions for risk man­age­ment to an­swer. This talk in­tends to high­light some of these from a prac­ti­tion­er’s point of view.

Árpád Szlávik (Mor­gan Stan­ley)

Model Risk Man­age­ment Evo­lu­tion

Model Risk Man­age­ment (MRM) within banks has been evolv­ing from model val­i­da­tion to­wards be­com­ing an ef­fec­tive and value-cen­tric func­tion. What are the dri­vers, stages, and re­sults of this evo­lu­tion? Is it pos­si­ble to mea­sure model risk? How can we use Ar­ti­fi­cial In­tel­li­gence (AI) / Ma­chine Learn­ing (ML) mod­els? These are the ques­tions we will ad­dress in the talk.

Réka Janosik (MSCI)

A new stan­dard for risk man­age­ment

The de­mand for mar­ket risk mea­sure­ment over long hori­zons in­creases as mu­tual funds and pen­sion plan man­agers want to con­trol their long-term risk/​re­turn pro­file. The naïve ex­ten­sion of the clas­sic risk man­age­ment frame­work fails to cap­ture key com­po­nents of long term risk. Mean re­ver­sion and regime shift­ing are im­por­tant fac­tors shap­ing long hori­zon sce­nar­ios, while port­fo­lio re­bal­anc­ing with state de­pen­dent trans­ac­tion cost, re­demp­tions and forced trad­ing are ex­am­ples of wrong way liq­uid­ity risk.

Lás­zló An­tal (Black­Rock)

Risk mea­sures and un­cer­tainty in port­fo­lio con­struc­tion – a case study of Sus­tain­abil­ity

In­vestors are be­com­ing in­creas­ingly aware of the fun­da­men­tal idea of in­evitable un­cer­tainty in fu­ture es­ti­mates, and would like to re­flect this in their as­set al­lo­ca­tion de­ci­sions. The tec­tonic shift of fi­nan­cial mar­kets to­wards sus­tain­able so­lu­tions un­der­lines the im­por­tance of this mind­set, given the vari­a­tion in dif­fer­ent views and be­liefs, and the mag­ni­tude of its im­pact. We de­velop a sys­tem­atic way to con­struct port­fo­lios, re­sid­ing on ro­bust and sto­chas­tic op­ti­mi­sa­tion tech­niques, min­imis­ing port­fo­lio VaR/​CVaR. We demon­strate how this method can be used to ac­count for un­cer­tainty in re­turn es­ti­mates, and cre­ate less con­cen­trated port­fo­lios.

István Barra (Black­Rock)

Out­lier ro­bust volatil­ity fore­cast­ing in prac­tice

We in­tro­duce the gen­er­al­ized au­tore­gres­sive score (GAS) mod­els by Creal et al 2014 which pro­vide a flex­i­ble yet com­pu­ta­tion­ally tractable frame­work to model la­tent vari­ables such as volatil­ity or de­fault in­ten­sity. By ty­ing the dy­nam­ics of the la­tent vari­ables to the dis­tri­b­u­tional as­sump­tion on the ob­ser­va­tions GAS mod­els co­her­ently ac­com­mo­date, and ap­pro­pri­ately mod­er­ate, the im­pact of out­liers on pa­ra­me­ter es­ti­mates. We show­case novel em­pir­i­cal adap­ta­tions of GAS mod­els to eq­uity risk fore­cast­ing, high­light­ing their in­tu­itive ap­peal, com­pu­ta­tional sim­plic­ity and em­pir­i­cal ben­e­fits rel­a­tive to cur­rent in­dus­try stan­dard volatil­ity fore­cast­ing mod­els.

Pro­gram com­mit­tee:

  • A. Zem­pléni (chair),
  • L. Márkus,
  • N.M. Arató,
  • J. Gáll,
  • Gy. Michalet­zky,
  • G. Mol­nár-Sáska,
  • V. Prokaj,
  • M. Rá­sonyi

Lo­cal or­gan­is­ers:

  • A. Zem­pléni (chair),
  • Á. Back­hausz,
  • V. Csiszár


The main spon­sor is This work­shop is also sup­ported by and
Registration form