Paper: arXiv 2309.12082
Authors: Tobias Wand, Timo Wiedemann, Jan Harren, Oliver Kamps
Abstract
The Geometric Brownian Motion (GBM) is a standard model in quantitative finance, but the potential function of its stochastic differential equation (SDE) cannot include stable nonzero prices. This article generalises the GBM to an SDE with polynomial drift of order q and shows via model selection that q=2 is most frequently the optimal model to describe the data. Moreover, Markov chain Monte Carlo ensembles of the accompanying potential functions show a clear and pronounced potential well, indicating the existence of a stable price.
Complexity vs Empirical Score
- Math Complexity: 8.5/10
- Empirical Rigor: 6.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper features advanced mathematics including stochastic differential equations, likelihood functions, and polynomial drift terms, while also applying a method to real stock market data (TAQ and CRSP) with specific data processing steps.
Research Flowchart
flowchart TD
A["Research Goal: Estimating Stable Fixed Points<br>and Langevin Potentials"] --> B["Methodology: SDEs with<br>Polynomial Drift of Order q"]
B --> C["Data: Financial Time Series<br>e.g., Stock Indices"]
C --> D["Computational Process:<br>Model Selection via MLE/BIC"]
D --> E{"Model Selection Outcome"}
E -- Optimal Model --> F["Key Finding: q=2 SDE<br>Most Frequently Optimal"]
F --> G["Key Finding: Markov Chain Monte Carlo<br>Reveals Potential Well/Stable Price"]
E -- Other q Values --> D
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 a generalisation of the Geometric Brownian Motion to an SDE with polynomial drift, finding that a second‑order polynomial (q=2) most frequently provides the best description of stock price data.
Source sentence
This article generalises the GBM to an SDE with polynomial drift of order q and shows via model selection that q=2 is most frequently the optimal model to describe the data.
Assumes 6
- The analysis is performed directly on price series rather than returns, assuming non‑stationary price dynamics are of interest. data
Source sentence
Note that we are directly analysing the price time series Pt instead of the returns r = log(Pt+1/Pt).
- The drift term of the SDE can be represented by a polynomial of order q. model
Source sentence
the drift term −dV/dP (P) is a polynomial of order q = 3.
- Microstructure effects are mitigated by using quoted prices, removing crossed quotes, and limiting the bid‑ask spread to below $5. market
Source sentence
require the bid-ask-spread to be below 5$, and finally use the last valid available quote within every 30-minute interval.
- Forward‑filled quotes are acceptable because they occur almost never for the very liquid stocks studied. data
Source sentence
We forward fill quotes if there is no valid entry for a given time interval. However, this does almost never happen for very liquid stocks such as those chosen in this paper.
Show 2 more
- Calm and turbulent market periods do not materially affect the model order selection results. market
Source sentence
the choice of calm or turbulent periods does not seem to influence our results.
- Maximum likelihood estimation and AIC model selection are appropriate for choosing the polynomial order. statistical
Source sentence
we find that a second order polynomial for the drift term often offers a very suitable description of the data.
Data
- TAQ intraday quoted price data market-prices commercial
Source sentence
We use intraday data from the TAQ (Trade and Quote) database. To account for microstructure related issues, such as the bid-ask-bounce or infrequent trading, we rely upon quoted prices that we re-sample to a 30-minute frequency.
- CRSP daily stock data market-prices commercial
Source sentence
CRSP: We also consider lower-frequency, daily, data from the Center of Research in Security Prices (CRSP), which is one of the most used database in economics and finance.
Costs
Unclear - sources disagree
Source sentence (the authors' words)
require the bid-ask-spread to be below 5$, and finally use the last valid available quote within every 30-minute interval.
Finds 5
- Model selection via AIC most frequently chooses a second‑order polynomial drift (q=2) as the best description of the price data. statistical
Source sentence
we find that a second order polynomial for the drift term often offers a very suitable description of the data.
- MCMC‑sampled potentials for q=2 exhibit pronounced wells, indicating stable fixed points at non‑zero prices. theoretical
Source sentence
Sampling the posterior density of the parameters via MCMC reveals that the second order polynomials’ potentials show very pronounced potential wells (i.e. stable minima).
- Higher‑order polynomials (q=3, q=4) are rarely selected, and q=4 is never supported by the 68% credible interval. statistical
Source sentence
... the maximum likelihood estimation lies within the 68% credible interval for all orders q except for q = 4.
- The frequency of stable fixed points (q=2) does not differ between calm and turbulent market periods. statistical
Source sentence
our results do not seem to show a difference between the calm and turbulent market period (cf. figure 3), perhaps indicating that the market can quickly adjust to such turbulent behaviour.
- Despite the superiority of q=2, the standard GBM (q=1) is still selected frequently, showing it remains a reasonable model. statistical
Source sentence
Finally, it is a reassuring result that the AIC selection still frequently suggests q = 1 (the standard GBM model) as the best polynomial order.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 22 related papers). Coverage is partial, so this is neither replication nor refutation.
3 papers in the archive
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.