Paper: arXiv 2310.02322
Authors: Christa Cuchiero, Janka Möller
Abstract
In the context of stochastic portfolio theory we introduce a novel class of portfolios which we call linear path-functional portfolios. These are portfolios which are determined by certain transformations of linear functions of a collections of feature maps that are non-anticipative path functionals of an underlying semimartingale. As main example for such feature maps we consider the signature of the (ranked) market weights. We prove that these portfolios are universal in the sense that every continuous, possibly path-dependent, portfolio function of the market weights can be uniformly approximated by signature portfolios. We also show that signature portfolios can approximate the growth-optimal portfolio in several classes of non-Markovian market models arbitrarily well and illustrate numerically that the trained signature portfolios are remarkably close to the theoretical growth-optimal portfolios. Besides these universality features, the main numerical advantage lies in the fact that several optimization tasks like maximizing (expected) logarithmic wealth or mean-variance optimization within the class of linear path-functional portfolios reduce to a convex quadratic optimization problem, thus making it computationally highly tractable. We apply our method also to real market data based on several indices. Our results point towards out-performance on the considered out-of-sample data, also in the presence of transaction costs.
Complexity vs Empirical Score
- Math Complexity: 9.5/10
- Empirical Rigor: 8.5/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper uses advanced mathematics including stochastic calculus, signature methods, and rough path theory, with rigorous theoretical proofs (e.g., universal approximation theorems), while also presenting substantial empirical work with simulated data, real market data, and transaction costs, leading to a highly tractable convex quadratic optimization.
Research Flowchart
flowchart TD
A["Research Goal: <br>Develop universal, <br>computationally tractable <br>portfolio construction method"] --> B["Key Methodology: <br>Linear Path-Functional Portfolios <br>using Signature of Market Weights"]
B --> C{"Data Inputs"}
C --> D["Simulated Non-Markovian <br>Market Models"]
C --> E["Real Market Data: <br>Equity Indices"]
D & E --> F["Computational Process: <br>Convex Quadratic Optimization <br>Maximize Expected Logarithmic Wealth"]
F --> G["Key Findings & Outcomes"]
G --> H["Theoretical: <br>Universality - Approximates <br>any continuous portfolio & Growth-Optimal Portfolio"]
G --> I["Empirical: <br>Out-performance on Out-of-Sample Data <br>Robust even with Transaction Costs"]
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 introduces linear path-functional portfolios based on signature feature maps for stochastic portfolio theory.
Source sentence
In the context of stochastic portfolio theory we introduce a novel class of portfolios which we call linear path-functional portfolios.
Assumes 6
- The price process is a continuous semimartingale satisfying No Unbounded Profit with Bounded Risk (NUPBR). model
Source sentence
the price process is only assumed to be a (continuous) semimartingale satisfying No Unbounded Profit with Bounded Risk
- Ranked market weights are defined by ordering the components of the semimartingale and breaking ties by label order. other
Source sentence
Xi_t, where we break ties by allocating a lower rank to smaller labels.
- Finite generated wealth holds if and only if NUPBR is satisfied. model
Source sentence
Assuming that their generated wealth is almost surely finite this holds if and only if NUPBR is satisfied
- Ergodicity/stationarity conditions justify time‑average approximations of returns. statistical
Source sentence
Recall that sufficient ergodicity/stationarity conditions guaranteeing that these approximations are justified have been discussed in Remark 5.10.
Show 2 more
- Portfolios are re‑balanced daily. execution
Source sentence
we want to emphasize, that even with transaction costs our portfolios are re-balanced daily.
- Transaction‑cost regularization uses a lower bound β ≥10⁻⁸. execution
Source sentence
When we include the regularization cost, we choose β0 = 0.5 and enforce a lower bound β ≥10⁻⁸.
Data
- Swiss Market Index (SMI) universe market-prices commercial
Source sentence
For the SMI, we consider the universe of stocks which survived between 2000-2022
- S&P500 universe market-prices commercial
Source sentence
to … 378 stocks in the S&P500 universe. For both markets, we obtained the data from Reuters Datastream.
Costs
Modeled
Transaction costs are included via a regularization parameter β (e.g., β ≥10⁻⁸) with scenarios of 0%, 1% and 5% costs; portfolios are re‑balanced daily.
Source sentence
we want to emphasize, that even with transaction costs our portfolios are re-balanced daily.
Finds 5
- Signature portfolios achieve mean log‑relative wealth virtually identical to the theoretical growth‑optimal portfolio on 100,000 test samples. performance
Source sentence
wealth of the theoretical growth-optimal and signature portfolios, evaluated on 100’000 test-samples,
- JL‑signature portfolios consistently outperform randomized‑signature and true‑signature portfolios out‑of‑sample in both log‑relative wealth and Sharpe‑ratios. performance
Source sentence
JL‑signature portfolios did consistently achieve a better out-of-sample performance both in terms of log-relative wealth and Sharpe-ratios than the randomized signature portfolios.
- All but one of the optimized portfolios out‑performed the universe portfolio over the out‑of‑sample period. performance
Source sentence
All but one of the optimized portfolios out-performed the universe portfolio over the out-of-sample period.
- Including transaction‑cost regularization improves portfolio performance under transaction costs. performance
Source sentence
the regularization for transaction cost proved effective, since all portfolios performed better with the regularization than without, under the respective level of transaction costs.
- Computing the true signature portfolio for the S&P500 is infeasible; JL‑signature is the most favorable in terms of feasibility and performance. theoretical
Source sentence
computing the true signature portfolio of the S&P500 universe would not have been feasible, since the number of stocks would have been too high. However, the JL‑signature portfolios did consistently achieve a better out-of-sample performance.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 18 related papers). Coverage is partial, so this is neither replication nor refutation.
9 papers in the archive : Universal randomised signatures for generative time series … ; Kernel Learning for Mean-Variance Trading Strategies
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.