Hi, I'm
Marcel-Philippe Metzger
|
My background spans quantitative finance, machine learning, and statistical modeling. Passionate about bridging the gap between data science and financial decision-making.
import torch
import torch.nn as nn
class MarketSignal(nn.Module):
def __init__(self, dim=64, heads=4):
super().__init__()
self.lstm = nn.LSTM(dim, dim, batch_first=True)
self.attn = nn.MultiheadAttention(dim, heads, batch_first=True)
self.head = nn.Linear(dim, 1)
def forward(self, x):
h, _ = self.lstm(x)
context, _ = self.attn(h, h, h)
return torch.sigmoid(self.head(context[:, -1]))
# >>> model.predict(asset='MPM')
# {'confidence': 1.00, 'signal': 'STRONG_BUY'}01. About Me
Welcome! I'm Marcel-Philippe — currently a trainee in Finance & Risk, fascinated by the intersection of financial markets, data analysis, and algorithmic decision-making.
Continuously developing my skills in quantitative methods, Python programming, and statistical modeling. My goal: building robust, data-driven trading strategies and solving complex financial problems with analytical precision.
Outside of work, I dive deep into machine learning applications for financial markets, time series analysis, and backtesting frameworks.
02. Experience
Finance & Risk Trainee
@ SantanderAsset Based Finance
- Contribution to ongoing ABS transactions, including automation of cashflow and waterfall reports
Risk Modelling
- Analysis and documentation of credit risk models for internal and external auditors (SAS)
- Statistical validation of scorecard applicability (HCBE)
- Development of automated scraping tool with GUI and analysis dashboard for dealer verification (Risk Execution)
Regulatory Information
- Analysis and preparation of regulatory data for regulatory requirements (SQL, Python)
Data Science Intern
@ Deutsche Bundesbank- Development of end-to-end scraping and data pipeline for automated processing of annual financial statements
- Built LLM-based classification model for industry classification of companies (local LLM, API-based)
- Fine-tuning of deep learning model for extraction of location and geodata from company reports using self-annotated data
Teaching Assistant
@ University of Mannheim- Tutor for Linear Algebra 1 and Linear Algebra 2A at the Chair of Algebraic Geometry
- Held weekly tutorial sessions for groups of 20+ students
- Grading of exams and weekly assignments
M.Sc. Mathematics for Business and Economics
@ University of MannheimExploring ideal lattice structures for secure post-quantum cryptographic systems
- Relevant Coursework: Mathematical Finance, Derivatives, Investments, Game Theory, International Asset Management
03. Skills & Technologies
Quantitative Methods
Programming & Data
Tools & Platforms
04. Projects
Financial Data Pipeline
End-to-end system for automated collection, processing, and storage of financial market data designed for ML models and derivatives valuation. Features GARCH(1,1) volatility forecasting, cross-validation between data sources, and intelligent missing value handling using Kalman filtering. Implements multi-layer outlier detection combining statistical methods with Isolation Forest.
Trading News Aggregator
Multi-source news aggregation system for financial markets with automated ticker extraction and semantic story clustering. Integrates seven data sources including Finnhub, SEC EDGAR, GDELT, and German financial press (FAZ, Handelsblatt). Features NLP-based ticker recognition from headlines and vector similarity analysis using sentence transformers for story deduplication.
05. Contact
Let's connect!
I'm always open to discussions about exciting projects, career opportunities or quantitative finance. Feel free to reach out!