{"id":17771,"date":"2019-02-14T13:44:58","date_gmt":"2019-02-14T18:44:58","guid":{"rendered":"https:\/\/fin-ml.ca\/?page_id=17771"},"modified":"2024-01-31T16:26:55","modified_gmt":"2024-01-31T21:26:55","slug":"stages","status":"publish","type":"page","link":"https:\/\/fin-ml.ca\/fr\/stages\/","title":{"rendered":"Stages"},"content":{"rendered":"<div class=\"vgblk-rw-wrapper limit-wrapper\">\n\n\t<p>Tous les \u00e9tudiants doivent effectuer un stage de recherche appliqu\u00e9e. Le programme Fin-ML FONCER aide les \u00e9tudiants \u00e0 trouver des stages gr\u00e2ce \u00e0 son r\u00e9seau de professionnels et \u00e0 son acc\u00e8s aux institutions financi\u00e8res. Ils sont \u00e9galement encourag\u00e9s \u00e0 trouver leurs stages de leurs propres initiatives ou avec l&rsquo;aide de leur superviseur, sous r\u00e9serve de l&rsquo;approbation du comit\u00e9 des stages de Fin-ML. A la fin du stage, l&rsquo;\u00e9tudiant doit pr\u00e9senter un rapport r\u00e9sumant son travail, et \u00eatre disponible pour pr\u00e9senter devant la communaut\u00e9 FONCER (les donn\u00e9es sensibles et les informations confidentielles seront respect\u00e9es) :<\/p>\n<ul>\n<li>Un rapport r\u00e9sumant les r\u00e9sultats de la recherche<\/li>\n<li>Une conf\u00e9rence bas\u00e9e sur les r\u00e9sultats de la recherche<\/li>\n<li>Un article destin\u00e9 \u00e0 \u00eatre soumis \u00e0 un journal<\/li>\n<li>Un mod\u00e8le de code transf\u00e9r\u00e9 sur papier<\/li>\n<\/ul>\n\t<iframe title=\"Application of GPT to automate financial newsletter production - December 8, 2023\" width=\"1260\" height=\"709\" data-src=\"https:\/\/www.youtube.com\/embed\/YYZrchzQQYY?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4><strong>Utilisation de <\/strong><strong>GPT\u00a0<\/strong><strong>pour automatiser la production de bulletins d&rsquo;information financi\u00e8re<br \/>\n<\/strong><\/h4>\n<p>Date : 08 d\u00e9cembre 2023<\/p>\n<p>Finliti wants to use artificial intelligence to optimize its business operations. One task they want to optimize is the research, writing and editing of a newsletter that is sent out to the company&rsquo;s mailing list every week. Manually writing this letter on average, according to the writer&rsquo;s experience, takes about 5 hours.<\/p>\n<p>Finliti wanted to investigate to what extent using code to automatically generate the newsletter would reduce the time necessary to produce a newsletter of a similar level of quality. The project to determine this consisted of writing the code, testing it out in practice, and revising it to determine the final improvement over manual writing.<\/p>\n<p>Overall, the amount of time needed dropped by about one hour. One suggested further step to reduce the revision time could be to have the code use the complete text from fewer articles if it can be upgraded with internet access.<\/p>\n\t<p>Conf\u00e9rence de <a href=\"http:\/\/linkedin.com\/in\/gavin-orok-19815b151\" target=\"_blank\" rel=\"noopener\">Gavin OROK<\/a>, actuellement \u00e9tudiant \u00e0 la maitrise en Finance quantitative \u00e0 l&rsquo;Universit\u00e9 Waterloo. Il a obtenu son Baccalaur\u00e9at en Math\u00e9matiques financi\u00e8res de cette m\u00eame universit\u00e9. Pour son stage de recherche collaborative, il a travaill\u00e9 au sein de la startup Finliti, bas\u00e9e \u00e0 Toronto, qui a pour mission d&rsquo;enseigner aux jeunes professionnels comment investir au-del\u00e0 d&rsquo;une strat\u00e9gie passive de fonds indiciels. Il \u00e9galement b\u00e9n\u00e9fici\u00e9 d&rsquo;une bourse Fin-ML CREATE sous la supervision de la professeure Christiane Lemieux.<\/p>\n\t<iframe title=\"Classifier Rank - A new model validation method to imbalanced data - November 25, 2022\" width=\"1260\" height=\"709\" data-src=\"https:\/\/www.youtube.com\/embed\/v-DoJ8sRivs?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4><strong>Classification par rang &#8211; Une nouvelle m\u00e9thode de validation des donn\u00e9es d\u00e9s\u00e9quilibr\u00e9es<br \/>\n<\/strong><\/h4>\n<p>Date : 25 novembre 2022<\/p>\n<p>Most of the commonly used confusion matrix-based classification performance metrics, such as\u00a0f1_score, MCC, PRC and lift curve, are sensitive to the class imbalance. To address this problem, we propose a novel classifier evaluation method, called classifier rank, which eliminates the effect of class imbalance by providing the rank of the classifier in the space of all possible classifiers.<\/p>\n<p>To rank a classifier, we find the distribution of performance metrics conditional on arbitrary imbalance rate. However, some metrics like PRC and lift curve are functions of a large sequence of confusion matrices whose joint distribution is difficult to estimate.<\/p>\n<p>Hence, we propose a discrete hidden Markov model and a continuous directed binary tree model to effectively represent this large-scale joint distribution. As a result, we can estimate the classifier rank using graphical inference algorithms, such as Variable Elimination and Monte-Carlo algorithm.<\/p>\n\t<p>Conf\u00e9rence de <a href=\"http:\/\/linkedin.com\/in\/ningsheng-zhao\" target=\"_blank\" rel=\"noopener\">Ningsheng ZHAO<\/a>, candidat au doctorat en g\u00e9nie de l&rsquo;Information et des syst\u00e8mes \u00e0 l&rsquo;Universit\u00e9 Concordia. Ningsheng a obtenu son dipl\u00f4me de ma\u00eetrise en Statistiques \u00e0 l&rsquo;Universit\u00e9 de Waterloo. Ses int\u00e9r\u00eats de recherche sont l&rsquo;apprentissage automatique et la Science des donn\u00e9es. Ses recherches actuelles portent sur les techniques de diagnostic et d&rsquo;explication des mod\u00e8les d&rsquo;apprentissage automatique et leurs applications dans le domaine des affaires. Il s&rsquo;agit d&rsquo;une recherche conjointe men\u00e9e avec les professeurs Jia Yuan Yu et Dr Krzysztof Dzieciolowski et soutenue par des programmes tel que MITACS, DAESYS, Fin-ML et CUPFA.<\/p>\n\t<iframe title=\"Risk-Aware Bid Optimization for Online Display Advertisement - November 4, 2022\" width=\"1260\" height=\"709\" data-src=\"https:\/\/www.youtube.com\/embed\/ZM4fvGLMbh0?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4><strong>Optimisation des offres en fonction du risque pour les annonces en ligne.<br \/>\n<\/strong><\/h4>\n<p>Date : 04 novembre 2022<\/p>\n<p>This research focuses on the bid optimization problem in the real-time bidding setting for online display advertisements. We proposed risk-neutral and risk-averse models that maximize the expected profit for the advertiser by exploiting historical data to design an upfront bidding policy, mapping the type of advertisement opportunity to a bid price and accounting for the risk of violating the budget constraint during a given period of time. And, we present the numerical results from experiments, which demonstrate that our risk-averse method can effectively control the risk of overspending the budget while achieving a competitive level of profit compared with the risk-neutral model and a state-of-the-art data-driven risk-aware bidding approach.<\/p>\n\t<p>Conf\u00e9rence de <a href=\"http:\/\/linkedin.com\/in\/reneeruifan\" target=\"_blank\" rel=\"noopener\">Rui FAN<\/a>, dipl\u00f4m\u00e9e en maitrise en Sciences de donn\u00e9es et Analyse d&rsquo;affaires de HEC Montr\u00e9al en collaboration avec le GERAD. Rui est actuellement ing\u00e9nieure en donn\u00e9es et intelligence artificielle \u00e0 la Banque Royale du Canada. Ses domaines de recherche incluent l&rsquo;optimisation et l&rsquo;apprentissage automatique. Elle s&rsquo;int\u00e9resse particuli\u00e8rement \u00e0 l&rsquo;optimisation consciente du risque, \u00e0 l&rsquo;apprentissage automatique int\u00e9gr\u00e9 \u00e0 la recherche op\u00e9rationnelle, et \u00e0 ses applications dans les secteurs de la finance et du commerce \u00e9lectronique. Elle a \u00e9galement b\u00e9n\u00e9fici\u00e9 d&rsquo;une bourse Fin-ML CREATE sous la supervision du professeur \u00c9rick Delage.<\/p>\n\t<iframe title=\"Foreseeing the Worst: Forecasting Electricity DART Spikes - September16th, 2022\" width=\"1260\" height=\"709\" data-src=\"https:\/\/www.youtube.com\/embed\/mkoENzuwfCs?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4><strong>Pr\u00e9voir le pire : Pr\u00e9vision des pics DART d&rsquo;\u00e9lectricit\u00e9<br \/>\n<\/strong><\/h4>\n<p>Date : 16 septembre 2022<\/p>\n<p>Statistical learning models are proposed for the prediction of the probability of a spike in the electricity DART (day-ahead minus real-time price) spread. Assessing the likelihood of DART spikes is of paramount importance for virtual bidders, among others. The model&rsquo;s performance is evaluated on historical data for the Long Island zone of the New York Independent System Operator (NYISO). A tailored feature set encompassing novel engineered features is designed. Such a set of features makes it possible to achieve excellent predictive performance and discriminatory power.<\/p>\n<p>Results are shown to be robust to the choice of the predictive algorithm. Lastly, the benefits of forecasting the spikes are illustrated through a trading exercise, confirming that trading strategies employing the model predicted probabilities as a signal generate consistent profits.<\/p>\n\t<p>Conf\u00e9rence de\u00a0<a href=\"http:\/\/linkedin.com\/in\/remi-galarneau-vincent-35726810b\" target=\"_blank\" rel=\"noopener\">R\u00e9mi GALARNEAU-VINCENT<\/a>, candidat au doctorat en Ing\u00e9nierie financi\u00e8re \u00e0 HEC Montr\u00e9al, sous la supervision de la professeure Genevi\u00e8ve Gauthier. Il a d\u00e9couvert le march\u00e9 de l&rsquo;\u00e9lectricit\u00e9 lors d&rsquo;un travail de recherche avec la soci\u00e9t\u00e9 Plant-E corp. C&rsquo;est un passionn\u00e9 d&rsquo;apprentissage automatique.<\/p>\n\t<iframe title=\"Generate &amp; predict multivariate time series - 11 March  2022\" width=\"1260\" height=\"709\" data-src=\"https:\/\/www.youtube.com\/embed\/s0gMqWn-nXo?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4><strong>G\u00e9n\u00e9rer et pr\u00e9dire des s\u00e9ries temporelles multivariables<\/strong><\/h4>\n<p>Date : 11 mars 2022<\/p>\n<p>Time series forecasting is to try to predict future realizations based on past observations. Those predictions can be precise or probabilistic depending on the model. Recent development in machine learning has accelerated the quantity of interesting models. The model selection is very much dependent of the use case and needs a thorough investigation. It is however harder to compare different models to measure their prediction quality, robustness, computational performance and data requirements. We will present a package written in Python to measure the performance of multiple models. For that purpose, we will focus on the simulation of time series to use as benchmarks.<\/p>\n\t<p>Conf\u00e9rence de <a href=\"https:\/\/www.researchgate.net\/profile\/Francis-Huot-Chantal\" class=\"broken_link\">Francis HUOT-CHANTAL<\/a>, candidat au doctorat en Math\u00e9matiques appliqu\u00e9es \u00e0 l&rsquo;Universit\u00e9 de Montr\u00e9al, sous la supervision du professeur Fabian Bastin. Le sujet de sa recherche porte sur la tenue de march\u00e9 \u00e0 haute fr\u00e9quence. Cependant, il aimerait approfondir la preuve th\u00e9orique de la convergence. Comme la partie la plus importante de sa th\u00e8se est consacr\u00e9e \u00e0 la mise en application, il travaille principalement avec les langages de programmation Julia et Python.<\/p>\n\t<iframe title=\"Valuation of syndicated loan adjusted for presence of financial covenant - Feb 11, 2022\" width=\"1260\" height=\"709\" data-src=\"https:\/\/www.youtube.com\/embed\/2ghlxv4PhO8?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4>\u00c9valuation de cr\u00e9dits syndiqu\u00e9s en pr\u00e9sence d&rsquo;un avenant financier<\/h4>\n<p>Date : 11 F\u00e9vrier 2022<\/p>\n<p>Nous \u00e9valuons l&rsquo;impact de l&rsquo;inclusion d&rsquo;une clause de s\u00e9curit\u00e9 dans les contrats de pr\u00eats syndiqu\u00e9s. Nous proposons un mod\u00e8le de jeu dynamique stochastique des ajustements des contrats de pr\u00eats syndiqu\u00e9s en pr\u00e9sence de la clause. Le mod\u00e8le tient compte du droit du pr\u00eateur de punir ou de tol\u00e9rer toute violation de la clause, et de la flexibilit\u00e9 de l&#8217;emprunteur dans l&rsquo;ajustement de sa strat\u00e9gie d&rsquo;investissement et de prise de risque. Nos exp\u00e9riences num\u00e9riques montrent que, si une clause de s\u00e9curit\u00e9 am\u00e9liore la valeur du pr\u00eat dans la plupart des \u00c9tats, elle peut avoir un effet n\u00e9gatif lorsque le risque de faillite devient important. Une \u00e9tude suppl\u00e9mentaire montre que le pr\u00eateur peut tol\u00e9rer de mani\u00e8re optimale un certain d\u00e9faut technique pour \u00e9viter cet effet n\u00e9gatif. Nous constatons \u00e9galement que la valeur de la clause diminue lorsque la solvabilit\u00e9 de l&#8217;emprunteur s&rsquo;am\u00e9liore.<\/p>\n\t<p>Conf\u00e9rence de <a href=\"http:\/\/linkedin.com\/in\/tiguenenabassaga\" target=\"_blank\" rel=\"noopener\">Tigu\u00e9n\u00e9 NABASSAGA<\/a>, candidat au doctorat en Ing\u00e9nierie financi\u00e8re \u00e0 HEC Montr\u00e9al, sous la supervision de la professeure Mich\u00e8le Breton. Tigu\u00e9n\u00e9 est actuellement gestionnaire au sein d&rsquo;Ernst &amp; Young Global Consulting Services.<\/p>\n\t<iframe title=\"Deep unsupervised Anomaly Detection in the derivatives market - December 3, 2021\" width=\"1260\" height=\"709\" data-src=\"https:\/\/www.youtube.com\/embed\/ybByZyLbEic?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4>D\u00e9tection automatique d&rsquo;anomalies dans le march\u00e9 des produits d\u00e9riv\u00e9s \u00e0 haute-fr\u00e9quence par apprentissage profond non-supervis\u00e9<\/h4>\n<p>Date : 3 D\u00e9cembre 2021<\/p>\nThe recent rise in the interest of market participants towards financial derivatives directly lead to an augmentation in the quantity of high-frequency data processed continually by all exchanges. Regulators, who constantly monitor markets to unveil potential infractions, still perform their investigations manually with the help of deterministic rule-based algorithms. Thus, this notable growth in data generation poses a heightened risk of fraudulent orders going unnoticed.<br \/>\nI will present an ongoing research work that aims to solve the problem of algorithmic fraud detection in the derivatives market. This work, in collaboration with TMX, relies on the vast literature of anomaly detection to locate potential fraudulent patterns in the limit order books&rsquo; time series. Because of the complex nature of that data and the very low amount of real fraud cases, a new framework based on deep unsupervised learning algorithms specifically designed for high-frequency time series is proposed.\n\t<p>Conf\u00e9rence de <a href=\"http:\/\/linkedin.com\/in\/cedricpoutre\" target=\"_blank\" rel=\"noopener\">C\u00e9dric POUTR\u00c9<\/a>, candidat au doctorat en Math\u00e9matiques financi\u00e8res \u00e0 l&rsquo;Universit\u00e9 de Montr\u00e9al, sous la supervision du Professeur Manuel Morales. Il s&rsquo;int\u00e9resse aux diverses applications de l&rsquo;apprentissage machine dans le trading \u00e0 haute fr\u00e9quence et la microstructure des march\u00e9s.<\/p>\n\t<iframe title=\"An Input-Output HMM to describe states of energy price in NYISO Market - November 26, 2021\" width=\"1260\" height=\"945\" data-src=\"https:\/\/www.youtube.com\/embed\/-5YKOe0w0e8?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4>Description des \u00e9tats du prix de l&rsquo;\u00e9nergie sur le march\u00e9 NYISO \u00e0 l&rsquo;aide du mod\u00e8le \u00e0 cha\u00eene de Markov cach\u00e9e d&rsquo;entr\u00e9e-sortie<\/h4>\n<p>Date : 26 Novembre 2021<\/p>\nThe energy market in New York is centralized and managed by an independent system operator called NYISO. The price of energy on this market is determined by an auction mechanism. The Analysis of this price reveals that it is very volatile for several reasons (uncertain and inelastic demand, dynamic changes in the behavior of participants, etc.) that can be represented by models indexed by Markov chains.<br \/>\nIn this work, an Input-Output HMM models is used to predict the price of energy and describe its hidden states. This model is well known for its ability to predict conditional price density. In order to be able to better describe the states (especially if they are numerous), a parsimonious parameterization is used.\n\t<p>Conf\u00e9rence d&rsquo;<a href=\"http:\/\/linkedin.com\/in\/ismael-assani\" target=\"_blank\" rel=\"noopener\">Isma\u00ebl ASSANI<\/a>, candidat au doctorat en statistique \u00e0 l&rsquo;Universit\u00e9 de Montr\u00e9al, sous la supervision du professeur Maciej Augustyniak. Isma\u00ebl s&rsquo;int\u00e9resse \u00e0 la mod\u00e9lisation des s\u00e9ries financi\u00e8res, en particulier aux mod\u00e8les index\u00e9s par des cha\u00eenes de Markov cach\u00e9es (discr\u00e8tes et continues), \u00e0 la couverture et \u00e0 l&rsquo;\u00e9valuation des produits d\u00e9riv\u00e9s ainsi qu&rsquo;\u00e0 l&rsquo;application des m\u00e9thodes d&rsquo;apprentissage automatique en finance.<\/p>\n\t<iframe title=\"Using RL to maximize Customer Profitability &amp; CLV at Financial Institutions - November 19, 2021\" width=\"1260\" height=\"709\" data-src=\"https:\/\/www.youtube.com\/embed\/7jNEsXduNKU?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4>Utilisation de l&rsquo;apprentissage par renforcement pour maximiser la rentabilit\u00e9 des clients et la valeur de la client\u00e8le dans les institutions financi\u00e8res<\/h4>\n<p>Date : 19 Novembre 2021<\/p>\nCustomer Lifetime Value, CLV, is a popular measure to understand the future profitability of customers to allocate resources in more efficient ways to keep the company alive during difficult economic situations.<br \/>\nWe use Machine Learning tools to predict the expected revenue from each customer during one year of his\/her relationship with the institution as the CLV of the customer. The approach is implemented on two datasets from two international financial institutions. Different feature engineering techniques were applied to improve the prediction power of the model. We used two stage or three stage prediction models.<br \/>\nIn the second phase, we train a reinforcement learning algorithm based on the history of marketing activities and the CLV as the state of customers to determine the optimum marketing action for customers in each state to maximize their profitability.\n\t<p>Conf\u00e9rence de <a href=\"http:\/\/linkedin.com\/in\/meisam-soltani-05777bb\" target=\"_blank\" rel=\"noopener\">Meisam SOLTANI<\/a>. Apr\u00e8s une formation en G\u00e9nie \u00e9lectrique et dans l&rsquo;industrie automobile, Meisam est pass\u00e9 \u00e0 l&rsquo;univers de la Science des donn\u00e9es et de l&rsquo;analytique. Il a obtenu un dipl\u00f4me en Sciences de la gestion \u00e0 l&rsquo;Universit\u00e9 de Waterloo en 2014. Tout en poursuivant son doctorat en Analyse de gestion \u00e0 l&rsquo;Universit\u00e9 Queens, sous la supervision du professeur Mikhail Nediak, Meisam a particip\u00e9 \u00e0 de nombreux projets en lien avec la Science des donn\u00e9es et d&rsquo;analytique \u00e0 la Banque Scotia.<\/p>\n\t<iframe title=\"Enhanced Rating Prediction with Text Data - 22 Oct. 20211, by Ernest Tafolong\" width=\"1260\" height=\"709\" data-src=\"https:\/\/www.youtube.com\/embed\/TbC7j3GBm5k?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4>Am\u00e9lioration de la pr\u00e9diction des notations \u00e0 l&rsquo;aide de donn\u00e9es textuelles<\/h4>\n<p>Date : 22 Octobre 2021<\/p>\n<p>Over recent years, Natural Language processing (NLP) is widely used in many areas including finance and economics. Over the last decades, some influential papers have investigated whether negative words in the press can affect significantly affect individual stocks and aggregate market (Tetlock, 2007). They provide strong empirical evidence between their dictionary and 10K\/10Q filing and return, trading volume, unexpected earning, and fraud.<\/p>\n<p>The aim of this research is to fill gap and provide a framework that assess the credit rating of the firm more broadly using natural language processing from firm fillings. The proposed model can be used to assess the discriminating power of the identified variables that have been shown to be related to the rating.<\/p>\n\t<p>Conf\u00e9rence d&rsquo;<a href=\"http:\/\/linkedin.com\/in\/ernest-tafolong-frm-b510922a\" target=\"_blank\" rel=\"noopener\">Ernest Tafolong<\/a>, dipl\u00f4m\u00e9 d&rsquo;une maitrise en Sciences de donn\u00e9es \u00e0 HEC Montr\u00e9al, sous la supervision du Professeur Erik Delage. Ernest est actuellement directeur du d\u00e9partement d&rsquo;ing\u00e9nierie financi\u00e8re d&rsquo;une soci\u00e9t\u00e9 de cr\u00e9dit alternatif \u00e9tablie au Canada, o\u00f9 il dirige le d\u00e9veloppement de tous les mod\u00e8les d&rsquo;apprentissage automatique pour les pr\u00eats automobiles et personnels. Son exp\u00e9rience ant\u00e9rieure comprend la quantification des risques de march\u00e9 \u00e0 la Banque Nationale du Canada, il a \u00e9galement \u00e9t\u00e9 conseiller en ing\u00e9nierie financi\u00e8re chez Desjardins o\u00f9 il a con\u00e7u et mis en \u0153uvre des mod\u00e8les quantitatifs pour les strat\u00e9gies d&rsquo;investissement et la gestion des risques. En outre, il a fourni des conseils quantitatifs pour la gestion du risque de mod\u00e8le dans de grandes banques fran\u00e7aises. Il a \u00e9galement obtenu une ma\u00eetrise en ing\u00e9nierie financi\u00e8re \u00e0 l&rsquo;Universit\u00e9 Laval en 2011.<\/p>\n\t<iframe title=\"Natural Language Processing for fraud detection in derivatives market - 17 June 2021,  Pan LIU\" width=\"1260\" height=\"709\" data-src=\"https:\/\/www.youtube.com\/embed\/CK6L8Y5zzxM?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4>D\u00e9tection de la fraude sur le march\u00e9 des produits d\u00e9riv\u00e9s gr\u00e2ce au Traitement du langage naturel (NLP)<\/h4>\n<p>Date : 17 juin 2021<\/p>\nWith increasing activities in the financial derivatives market, exchange regulators are seeking to build smarter market surveillance systems to detect potential frauds. Current systems are often based on rules capturing known suspicious patterns reflected in the structured market data, while not having the ability to process information conveyed by unstructured textual data such as business news, which, however, can have crucial real-time impact on the market. Thus, it is of great interest to leverage NLP to extract analytics from textual data, so that the surveillance system could assess trading behaviors more accurately due to the added awareness of the business context.<br \/>\nIn this talk, I will introduce a work in progress that develops a financial events analysis NLP pipeline that extracts important events from financial news then analyzes their characteristics, and a framework to leverage those analytics to help detect financial frauds, especially illegal insider trading, in the derivatives market.\n\t<p>Conf\u00e9rence de <a href=\"http:\/\/linkedin.com\/in\/pan-liu\" target=\"_blank\" rel=\"noopener\">Pan LIU<\/a>, candidat au doctorat en Sciences de donn\u00e9es \u00e0 HEC Montr\u00e9al, sous la supervision du professeur Gilles Caporossi. Son int\u00e9r\u00eat de recherche est de d\u00e9velopper des mod\u00e8les de traitement du langage naturel et d&rsquo;apprentissage automatique pour r\u00e9soudre des probl\u00e8mes en finance.<\/p>\n\t<iframe title=\"Options Pricing via Neural SDEs and Martingale Pricing Theory - 28 May 2021, Timothy DeLise\" width=\"1260\" height=\"709\" data-src=\"https:\/\/www.youtube.com\/embed\/G5UMXNgy81E?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4>Fixation du prix des options par le biais des \u00e9quations diff\u00e9rentielles stochastiques neurales et de la th\u00e9orie de l&rsquo;\u00e9valuation de Martingale<\/h4>\n<p>Date : 28 Mai 2021<\/p>\nThis research investigates pricing financial options based on the traditional martingale theory of arbitrage pricing combined with neural SDEs. We treat neural SDEs as universal It\u00f4 process approximators. In this way we can lift all assumptions on the form of the underlying price process, and compute theoretical option prices numerically based on empirical data.<br \/>\nAlong the way we propose a new variation to train neural SDEs by implementing the Wassterstein distance metric as a loss function. While similar to the GAN formulation, this technique does not need a discriminator and works will in the experiments, albeit limited in scope. Furthermore, it is conjectured that the error of the option price implied by the learnt model can be bounded by the very Wasserstein distance metric that was used to fit the empirical data.\n\t<p>Conf\u00e9rence de <a href=\"https:\/\/www.researchgate.net\/scientific-contributions\/Timothy-DeLise-2148723686\" target=\"_blank\" rel=\"noopener\" class=\"broken_link\">Timothy DeLISE<\/a>, candidat au doctorat en Math\u00e9matiques \u00e0 l&rsquo;Universit\u00e9 de Montr\u00e9al, sous la supervision du professeur Manuel Morales. Ses int\u00e9r\u00eats de recherche sont les math\u00e9matiques, l&rsquo;apprentissage automatique et le d\u00e9veloppement de logiciels.<\/p>\n\t<iframe title=\"Deep hedging methods for pricing &amp; hedging financial derivatives - 14 May 2021, Alexandre CARBONNEAU\" width=\"1260\" height=\"709\" data-src=\"https:\/\/www.youtube.com\/embed\/rWCoiKIkfpM?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4>M\u00e9thodes complexes pour la fixation des prix et la protection des produits financiers d\u00e9riv\u00e9s<\/h4>\n<p>Date : 14 Mai 2021<\/p>\nI will present a reinforcement learning approach for the problem of optimally hedging financial derivatives. I will start by giving an overview of the hedging problem with classical approaches considered in the literature.<br \/>\nAfterwards, I will present the approach called deep hedging which I extensively studied in my PhD research thesis. This class of algorithms represents trading policies by neural networks. In other words, asset positions used for hedging are obtained as the output vectors of neural networks. Various numerical experimentations of this approach will be presented for pricing and hedging different financial derivatives.\n\t<p>Conf\u00e9rence d&rsquo;<a href=\"http:\/\/linkedin.com\/in\/alexandre-carbonneau-26945698\" target=\"_blank\" rel=\"noopener\">Alexandre CARBONNEAU<\/a> qui a obtenu un doctorat en Finance math\u00e9matique \u00e0 l&rsquo;Universit\u00e9 Concordia sous la direction du Professeur Fr\u00e9d\u00e9ric Godin. Il est titulaire d&rsquo;un Baccalaur\u00e9at en Math\u00e9matique avec une sp\u00e9cialisation en actuariat de l&rsquo;Universit\u00e9 de Montr\u00e9al, ainsi qu&rsquo;une Maitrise en finance math\u00e9matique et computationnelle de la m\u00eame institution.<\/p>\n\t<iframe title=\"Analyse comparative librairies de pr\u00e9diction de s\u00e9ries chronologiques - 7 May 2021 - Samuel TREMBLAY\" width=\"1260\" height=\"709\" data-src=\"https:\/\/www.youtube.com\/embed\/1aK0pm8WlEE?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4>Analyse comparative des librairies de pr\u00e9diction de s\u00e9ries chronologiques<\/h4>\n<p>Date : 7 Mai 2021<\/p>\nLes diff\u00e9rents mod\u00e8les permettant l&rsquo;analyse et la pr\u00e9diction de s\u00e9ries chronologiques sous le langage de programmation Python sont list\u00e9s et cat\u00e9goris\u00e9s. Ensuite, des donn\u00e9es sur sept (7) contrats \u00e0 terme diff\u00e9rents sont obtenues et nettoy\u00e9es afin d&rsquo;\u00eatre utilis\u00e9es lors de l&rsquo;analyse comparative. Une division en 20 \u00e9chantillons d&rsquo;entra\u00eenement et de test est r\u00e9alis\u00e9e dans le but d&rsquo;\u00e9valuer l&rsquo;erreur de pr\u00e9diction. Une s\u00e9lection restreinte de librairies adapt\u00e9es au pr\u00e9sent cas d&rsquo;utilisation est effectu\u00e9e. Le mod\u00e8le ARIMA \u00e9merge comme \u00e9tant le meilleur mod\u00e8le en termes du MAE vis-\u00e0-vis les autres types de mod\u00e8les pour la s\u00e9rie S2.<br \/>\nDans le but d&rsquo;explorer plus en d\u00e9tails les mod\u00e8les d&rsquo;apprentissage profond, ceux-ci sont revisit\u00e9s. L&rsquo;architecture du r\u00e9seau de neurones r\u00e9currents \u00e0 m\u00e9moire court et long terme (en anglais LSTM) est retenue comme \u00e9tant la plus performante.<br \/>\nEnfin, ces r\u00e9sultats sont discut\u00e9s puis des pistes d&rsquo;am\u00e9lioration sont pr\u00e9sent\u00e9es.\n\t<p>Conf\u00e9rence de <a href=\"http:\/\/linkedin.com\/in\/tremblaysamuel\" target=\"_blank\" rel=\"noopener\">Samuel TREMBLAY<\/a>, conseiller en ing\u00e9nierie des donn\u00e9es au Mouvement Desjardins, o\u00f9 il accompagne de nombreux secteurs d&rsquo;affaires dans la valorisation de leurs m\u00e9gadonn\u00e9es. Samuel est titulaire d&rsquo;un Baccalaur\u00e9at en G\u00e9nie a\u00e9rospatiale de l&rsquo;\u00c9cole Polytechnique de l&rsquo;Universit\u00e9 de Montr\u00e9al et d&rsquo;une Ma\u00eetrise en Intelligence d&rsquo;affaires \u00e0 HEC Montr\u00e9al, sous la co-supervision des professeurs Jean-Francois Plante et Gregory Vial.<\/p>\n\t<iframe title=\"Arbitrage-free Bond price prediction - April 23, 2021 - Xiang GAO\" width=\"1260\" height=\"709\" data-src=\"https:\/\/www.youtube.com\/embed\/enNaoDRxhcY?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4>Pr\u00e9vision du prix des obligations bas\u00e9e sur un filtre RNN et une r\u00e9gularisation sans arbitrage<\/h4>\n<p>Date : 23 Avril 2021<\/p>\nWe review the class of HJM forward rate model and the dynamic term structure model in forecasting the yield curves and the prices of the coupon bonds by filter-based recurrent neural networks.<br \/>\nWe propose model in the Kalman filter and the particle filter with the arbitrage-free restriction as regularization using the deep neural networks. We mainly build on LSTM to extract the useful information from historical time series of daily closed bond data and other type of networks to process the sequential errors for yield curve modeling and calibrating.<br \/>\nThe purpose of the research is mainly to study the effect of arbitrage penalty on different forecasting horizons. We use data from U.S. Treasuries and coupon bonds of twelve corporate issuers. We provide analysis of the prediction error, error distribution, the average excess return, and results of out-of-sample test to show the efficiency and performance of our models.\n\t<p>Conf\u00e9rence de <a href=\"http:\/\/linkedin.com\/in\/xiang-gao-795b7142\" target=\"_blank\" rel=\"noopener\">Xiang GAO<\/a>, titulaire d&rsquo;un doctorat en Math\u00e9matiques appliqu\u00e9es \u00e0 l&rsquo;Universit\u00e9 Concordia sous la supervision du professeur Cody Hyndman. Ses int\u00e9r\u00eats de recherche sont: les math\u00e9matiques financi\u00e8res et l&rsquo;apprentissage automatique.<\/p>\n\t<iframe title=\"A WaveNet based model for Portfolio Management by using Deep RL - April 16, 2021 - Saeed MARZBAN\" width=\"1260\" height=\"709\" data-src=\"https:\/\/www.youtube.com\/embed\/q6uc2Zg3lk0?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4>Recours \u00e0 l&rsquo;apprentissage profond par renforcement pour la gestion de portefeuilles bas\u00e9e sur un mod\u00e8le WaveNet<\/h4>\n<p>Date : 16 Avril 2021<\/p>\nPortfolio management is a challenging task in its nature because of the abundance of factors that should be simultaneously considered: the investors&rsquo; risk preferences and constraints, the investment environments and their limitations, and complicated features that affect the future price movements.<br \/>\nIn this study, we propose a model based on actor-only reinforcement learning that takes all of these factors into account and provides a significant improvement in the portfolio performance compared to the most recent deep reinforcement learning models. In particular, we build on WaveNet, a model that was originally proposed for generating audio waveforms, to extract the useful information from historical time series of assets for portfolio management. Contrary to the similar previous models, we show that the performance of our approach is not affected by how the assets are permuted when fed into the model, and this could make the output of the model more reliable by practitioners.\n\t<p>Conf\u00e9rence de <a href=\"http:\/\/linkedin.com\/in\/saeed-marzban-07891a56\" target=\"_blank\" rel=\"noopener\">Saeed MARZBAN<\/a>. Sous la supervision du professeur \u00c9rick Delage, Saeed termine un doctorat en Ing\u00e9nierie financi\u00e8re \u00e0 HEC Montr\u00e9al avec une concentration sur les mati\u00e8res quantitatives tels que les probabilit\u00e9s, les statistiques, le calcul stochastique, l&rsquo;analyse num\u00e9rique et l&rsquo;apprentissage automatique. En particulier, sa th\u00e8se porte sur l&rsquo;application de diff\u00e9rents types d&rsquo;apprentissage par renforcement de gradient de politique \u00e0 l&rsquo;analyse de s\u00e9ries chronologiques pour l&rsquo;\u00e9valuation des options, la couverture et la gestion de portefeuilles (en collaboration avec une soci\u00e9t\u00e9 d&rsquo;investissement). Il poss\u00e8de aussi des comp\u00e9tences et une exp\u00e9rience approfondie en Programmation dans divers langages et cadres orient\u00e9s objet, ainsi que des exp\u00e9riences industrielles et de gestion.<\/p>\n\t<iframe title=\"Building a Recommender System using Graph Neural Networks  - Feb 12, 2021 - J\u00e9r\u00e9mi DEBLOIS-BEAUCAGE\" width=\"1260\" height=\"945\" data-src=\"https:\/\/www.youtube.com\/embed\/hvTawbQnK_w?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4>D\u00e9veloppement d&rsquo;un syst\u00e8me de recommandations \u00e0 l&rsquo;aide de r\u00e9seaux de neurones artificiels<\/h4>\n<p>Date : 12 F\u00e9vrier 2021<\/p>\nRecommender systems are now a well-known technology used by many e-commerce platforms. Those platforms often have access to diverse sources of data, and traditional recommender systems might not accurately leverage all the available information.<br \/>\nThis conference introduces the usage of Graph Neural Networks (GNNs) for personalized recommendation in a large-scale and heterogeneous data setting. We will observe how popular GNN models can be used for the task and what enhancements worked best for Decathlon, notably concerning how to handle evolving and seasonal customers&rsquo; preferences.\n\t<p>Conf\u00e9rence de <a href=\"http:\/\/linkedin.com\/in\/jdebloisbeaucage\" target=\"_blank\" rel=\"noopener\">Jeremi DEBLOIS-BEAUCAGE<\/a>, actuellement ing\u00e9nieur des donn\u00e9es chez QuantumBlack, une entreprise de McKinsey. Il a \u00e9t\u00e9 stagiaire de recherche en Intelligence artificielle chez D\u00e9cathlon Canada. Jeremi a obtenu une ma\u00eetrise en Intelligence d&rsquo;affaires \u00e0 HEC Montr\u00e9al, sous la supervision du professeur Laurent Charlin. Il a \u00e9galement \u00e9t\u00e9 r\u00e9cipiendaire d&rsquo;une bourse MITACS Accelerate.<\/p>\n\t<iframe title=\"Hedging Basket Options with Deep Learning - Jan 29, 2021 - by Pierre ROSIN\" width=\"1260\" height=\"709\" data-src=\"https:\/\/www.youtube.com\/embed\/gcg84IkFL3w?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen src=\"data:image\/gif;base64,R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw==\" class=\"lazyload\" data-load-mode=\"1\"><\/iframe>\n\t<h4>Couverture des options sur panier avec l&rsquo;apprentissage profond<\/h4>\n<p>Date : 29 Janvier 2021<\/p>\n<p>This conference introduces an effective model-free hedging strategy for a Basket Option using its component assets. By using Principal Components Analysis, the cost of trading can be reduced significantly and mitigate the effect of trading with non-liquid assets. Through numerical analysis, we will observe that an LSTM architecture for solving such a problem is very robust to compute an optimal solution. The model will also express the capacity of the neural network to fit an hedging strategy based on a custom risk function of an agent.<\/p>\n\t<p>Conf\u00e9rence de <a href=\"http:\/\/linkedin.com\/in\/pierre-rosin\" target=\"_blank\" rel=\"noopener\">Pierre ROSIN<\/a>, actuellement analyste de donn\u00e9es \u00e0 l&rsquo;Autorit\u00e9 des march\u00e9s financiers. Il est titulaire d&rsquo;une maitrise en Finance math\u00e9matique et computationnelle de l&rsquo;Universit\u00e9 de Montr\u00e9al, sous la supervision du Professeur Fabian Bastin.<\/p>\n\n<\/div><!-- .vgblk-rw-wrapper -->","protected":false},"excerpt":{"rendered":"<p>Tous les \u00e9tudiants doivent effectuer un stage de recherche appliqu\u00e9e. Le programme Fin-ML FONCER aide les \u00e9tudiants \u00e0 trouver des stages gr\u00e2ce \u00e0 son r\u00e9seau de professionnels et \u00e0 son acc\u00e8s aux institutions financi\u00e8res. Ils sont \u00e9galement encourag\u00e9s \u00e0 trouver leurs stages de leurs propres initiatives ou avec l&rsquo;aide de leur superviseur, sous r\u00e9serve de&#8230;<\/p>\n","protected":false},"author":1,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"inline_featured_image":false,"footnotes":""},"class_list":["post-17771","page","type-page","status-publish","hentry"],"jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/fin-ml.ca\/fr\/wp-json\/wp\/v2\/pages\/17771","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/fin-ml.ca\/fr\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/fin-ml.ca\/fr\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/fin-ml.ca\/fr\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/fin-ml.ca\/fr\/wp-json\/wp\/v2\/comments?post=17771"}],"version-history":[{"count":0,"href":"https:\/\/fin-ml.ca\/fr\/wp-json\/wp\/v2\/pages\/17771\/revisions"}],"wp:attachment":[{"href":"https:\/\/fin-ml.ca\/fr\/wp-json\/wp\/v2\/media?parent=17771"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}