Paper: arXiv 2406.19403
Authors: Wojciech Wisniewski, Yuri Kalnishkan, David Lindsay, Siân Lindsay
Abstract
Financial organisations such as brokers face a significant challenge in servicing the investment needs of thousands of their traders worldwide. This task is further compounded since individual traders will have their own risk appetite and investment goals. Traders may look to capture short-term trends in the market which last only seconds to minutes, or they may have longer-term views which last several days to months. To reduce the complexity of this task, client trades can be clustered. By examining such clusters, we would likely observe many traders following common patterns of investment, but how do these patterns vary through time? Knowledge regarding the temporal distributions of such clusters may help financial institutions manage the overall portfolio of risk that accumulates from underlying trader positions. This study contributes to the field by demonstrating that the distribution of clusters derived from the real-world trades of 20k Foreign Exchange (FX) traders (from 2015 to 2017) is described in accordance with Ewens’ Sampling Distribution. Further, we show that the Aggregating Algorithm (AA), an on-line prediction with expert advice algorithm, can be applied to the aforementioned real-world data in order to improve the returns of portfolios of trader risk. However we found that the AA ‘struggles’ when presented with too many trader ``experts’’, especially when there are many trades with similar overall patterns. To help overcome this challenge, we have applied and compared the use of Statistically Validated Networks (SVN) with a hierarchical clustering approach on a subset of the data, demonstrating that both approaches can be used to significantly improve results of the AA in terms of profitability and smoothness of returns.
Complexity vs Empirical Score
- Math Complexity: 4.5/10
- Empirical Rigor: 6.0/10 (capped from 7.0: transaction costs not modelled; see penalty flags)
- Quadrant: Street Traders — empirical and applied, lower math complexity (effective empirical rigor capped from 7.0)
Why this score: The paper uses advanced statistical concepts like Ewens’ Sampling Distribution and hypergeometric tests, but the math is presented at a high level without deep derivations, keeping complexity moderate. It demonstrates strong empirical rigor by analyzing a large proprietary dataset (20k traders, 3 years of FX data), applying clustering algorithms (SVN, hierarchical), and testing prediction algorithms (Aggregating Algorithm) with clear profitability metrics.
Research Flowchart
flowchart TD
A["Research Goal<br>Cluster 20k FX Traders & Predict Risk Portfolios<br>using Expert Advice"] --> B{"Key Data & Inputs<br>20k FX Traders (2015-2017)<br>Transaction Trades"}
B --> C["Computation: Clustering<br>Derive Temporal Clusters &<br>Validate Distribution Ewens' Sampling"]
C --> D{"Computation: Prediction<br>Aggregating Algorithm AA<br>Applied to Real-World Data"}
D --> E{"Key Findings / Outcomes"}
E --> F["Outcome 1<br>Clusters follow Ewens' Distribution"]
E --> G["Outcome 2<br>AA 'Struggles' with too many experts<br>(Overfitting/Similarity issues)"]
E --> H["Solution Applied<br>Statistically Validated Networks (SVN)<br>& Hierarchical Clustering"]
H --> I["Final Outcome<br>SVN/Clustering significantly improves<br>AA Profitability & Return Smoothness"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes applying the Aggregating Algorithm with trader clustering, showing that cluster distributions follow Ewens' Sampling Distribution and that clustering improves AA performance.
Source sentence
This study contributes to the field by demonstrating that the distribution of clusters derived from the real‑world trades of 20k Foreign Exchange (FX) traders (from 2015 to 2017) is described in accordance with Ewens' Sampling Distribution.
Assumes 3
- The set of all possible outcomes (outcome space) is known in advance. model
Source sentence
We assume that the set of all possible outcomes (outcome space) Ωis known to us in advance
- The validity of the Aggregating Algorithm holds under mild regularity assumptions on the game. statistical
Source sentence
The validity of the AA holds under some mild regularity assumptions on the game
- Orders must be closed in their entirety once opened. execution
Source sentence
Online trading platforms often operate under the stipulation that once an order is placed (opened), it must be closed in its entirety.
Data
2015 – 2017 · Foreign Exchange (FX) traders · fx
- Proprietary FX client trades dataset alternative proprietary
Source sentence
The proprietary dataset comprises the trades made by over 20k clients during 2015-2017.
Costs
Not modeled
Finds 4
Sharpe 2.00 (benchmark 0.60) · max drawdown 1.1%
- Cluster sizes follow Ewens' Sampling Distribution. statistical
Source sentence
This study contributes to the field by demonstrating that the distribution of clusters
- The Aggregating Algorithm improves portfolio returns but struggles when many similar experts are present. performance
Source sentence
can be applied to the aforementioned real‑world data in order to improve the returns of portfolios of trader risk. However we found that the AA 'struggles' when presented with too many trader ``experts'', especially when there are many trades with similar overall patterns.
- Statistically Validated Networks and hierarchical clustering significantly improve AA profitability and return smoothness. performance
Source sentence
both approaches can be used to significantly improve results of the AA in terms of profitability and smoothness of returns.
- Hierarchical clustering yields better risk measures than standard AA, though return is lower. performance
Source sentence
however for hierarchical clustering all other risk measures are significantly better than the standard AA besides the return.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 26 related papers). Coverage is partial, so this is neither replication nor refutation.
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.