Paper: arXiv 2511.06177
Authors: Dmitrii Vlasiuk, Mikhail Smirnov
Abstract
We test the hypothesis that consecutive intraday price changes in the most liquid U.S. equity ETF (SPY) are conditionally nonrandom. Using NBBO event-time data for about 1,500 regular trading days, we form for every lag L ordered pairs of a backward price increment (“push”) and a forward price increment (“response”), standardize them, and estimate the expected responses on a fine grid of push magnitudes. The resulting lag-by-magnitude maps reveal a persistent structural shift: for short lags (1-5,000 ticks), expected responses cluster near zero across most push magnitudes, suggesting high short-term efficiency; beyond that range, pronounced tails emerge, indicating that larger historical pushes increasingly correlate with nonzero conditional responses. We also find that large negative pushes are followed by stronger positive responses than equally large positive pushes, consistent with asymmetric liquidity replenishment after sell-side shocks. Decomposition into symmetric and antisymmetric components and the associated dominance curves confirm that short-horizon efficiency is restored only partially. The evidence points to an intraday, lag-resolved anomaly that is invisible in unconditional returns and that can be used to define tradable pockets and risk controls.
Complexity vs Empirical Score
- Math Complexity: 2.5/10
- Empirical Rigor: 9.0/10
- Quadrant: Street Traders — empirical and applied, lower math complexity
Why this score: The paper relies on straightforward statistical conditioning (standardized push-response analysis, dominance curves) without complex stochastic calculus or advanced theory, but implements a rigorous, data-heavy methodology using event-time NBBO data across 1,500+ days, with detailed preprocessing and multi-lag estimation suitable for direct backtesting.
Research Flowchart
flowchart TD A["Research Goal<br>Test conditional non-randomness<br>of consecutive SPY intraday price changes"] --> B["Data Source & Prep<br>1,500 days NBBO event-time data<br>Regular trading hours only"] B --> C["Methodology<br>Form ordered pairs of push (backward increment)<br>and response (forward increment) for lag L"] C --> D["Computation<br>Standardize increments<br>Estimate expected responses on grid of<br>push magnitudes for each lag L"] D --> E["Key Findings<br>Short Lag (1-5k ticks):<br>Responses cluster near zero<br>High efficiency"] D --> F["Key Findings<br>Long Lag (>5k ticks):<br>Pronounced tails appear<br>Larger pushes correlate with non-zero responses"] D --> G["Key Findings<br>Asymmetry: Large negative pushes<br>followed by stronger positive responses<br>Consistent with asymmetric liquidity replenishment"]
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 push‑response decomposition that treats the initial price change as a push and the subsequent change as a response, analyzed over a lag‑resolved surface.
Source sentence
In the paper, rather than observing and analyzing aggregate autocorrelation, we propose the push-response decomposition, where pushes represent the initial price change at a fixed lag and responses are the corresponding price change that follows.
Assumes 6
- Only quotes with the R condition flag are used, assuming they are regular NBBO‑eligible quotes. data
Source sentence
the dataset was restricted to records carrying the R condition flag, which indicates regular quotes eligible for NBBO formation under Regulation National Market System.
- Mid‑price jumps larger than $1.50 are treated as data errors and removed. data
Source sentence
was treated as a data error, and both affected events were excluded from the sample
- Observations are indexed in event time, assuming all three points of a push‑response triplet belong to the same regular session. model
Source sentence
we form triplets (t −L, t, t + L) out of event time only when all three belong to the same regular session.
- No‑dynamic‑arbitrage restrictions are not imposed a priori; any detected conditional response is examined without enforcing those constraints. other
Source sentence
it is important to note that we do not impose no-arbitrage conditions initially and aim to look at the output produced by the data.
Show 2 more
- Symmetric and antisymmetric components separate magnitude‑driven adjustment from sign‑driven drift. model
Source sentence
S captures magnitude-conditioned adjustment that does not depend on the sign of the push. A isolates sign-conditioned drift.
- The dominance index ρ(L,|j|) reflects which mechanism (symmetry vs antisymmetry) prevails at a given lag and push magnitude. statistical
Source sentence
Values near +1 indicate local antisymmetry dominance. Values near −1 indicate local symmetry dominance.
Data
- NBBO quote data for SPY market-prices commercial
Source sentence
The analysis is based on the National Best Bid and Offer (NBBO) quote data provided by TAQ for the SPDR S&P 500 ETF Trust (SPY) covering the period from January 2018 to December 2023.
Cboe NYSE TAQ
Costs
Not applicable
Finds 5
- For short lags (1‑5,000 ticks) expected responses cluster near zero, while for longer lags pronounced tails appear, indicating non‑zero conditional responses for larger pushes. statistical
Source sentence
for short lags (1-5,000 ticks), expected responses cluster near zero across most push magnitudes, suggesting high short-term efficiency; beyond that range, pronounced tails emerge, indicating that larger historical pushes increasingly correlate with nonzero conditional responses.
- Large negative pushes are followed by stronger positive responses than equally large positive pushes, showing asymmetric liquidity replenishment. statistical
Source sentence
by stronger positive responses than equally large positive pushes, which is consistent with asymmetric liquidity replenishment after sell-side shocks.
- Symmetry dominates for moderate push magnitudes (|zp| ≲ 0.7), as shown by the dominance index. statistical
Source sentence
Figure 2 shows that near |zp| ≲0.7 symmetry tends to dominate throughout, which matches the flat bowl in Figure 1.
- The classic “market‑mill” reversal pattern is reproduced: positive pushes map to negative expected responses and negative pushes map to positive responses. statistical
Source sentence
the reversal curves are evident, with positive pushes mapping to negative expected responses and negative pushes mapping to positive responses, strongest in the wings and attenuating near zero.
- The lag‑resolved push‑response surface rises with lag up to a plateau for strong positive pushes, then flattens, while negative pushes show no plateau. statistical
Source sentence
surface rises with L up to a plateau for strong positive pushes: beyond that point the structure flattens, which indicates that predictability does not keep compounding at the far end of the grid once the information in the initiating push has diffused. However, for negative pushes, we observe an apparent lack of plate
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 20 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.
Cited References (48)
- Continuous auctions and insider trading (1985)
- Trades, Quotes and Prices: Financial Markets Under the Microscope (2018)
- Hedging Demand and Market Intraday Momentum (2021)
- The cross-section of intraday and overnight returns (2021)
- The short-term predictability of returns in order book markets: a deep learning perspective (2024)
- Market Mill Dependence Pattern in the Stock Market: Asymmetry Structure, Nonlinear Correlations and Predictability (2006)
- Market Mill Dependence Pattern in the Stock Market: Distribution Geometry, Moments and Gaussization (2006)
- Market Mill Dependence Pattern in the Stock Market: Individual Portraits (2006)
and 40 more
- Market Mill Dependence Pattern in the Stock Market: Modeling of Predictability and Asymmetry via Multi-Component Conditional Distribution (2007)
- Market mill dependence pattern in the stock market: Multiscale conditional dynamics (2009)
- Why Do Security Prices Change? A Transaction-Level Analysis of NYSE Stocks (1997)
- The components of the bid–ask spread: A general approach (1997)
- Autoregressive conditional duration: A new model for irregularly spaced transaction data (1998)
- Modeling and forecasting realized volatility (2003)
- Econometric analysis of realized volatility and its use in estimating stochastic volatility models (2002)
- Trade execution costs and market quality after decimalization (2003)
- Limit order book as a market for liquidity (2005)
- Does algorithmic trading improve liquidity? (2011)
- High frequency trading and the new market makers (2013)
- High-frequency trading and price discovery (2014)
- Low-latency trading (2013)
- A Hausman test for the presence of market microstructure noise in high-frequency data (2019)
- When are contrarian profits due to stock market overreaction? (1990)
- Ultra high frequency volatility estimation with dependent microstructure noise (2011)
- Time and the process of security price adjustment (1992)
- Algorithmic trading and the market for liquidity (2013)
- No-dynamic-arbitrage and market impact (2010)
- Evidence on the speed of convergence to market efficiency (2005)
- The limits of arbitrage (1997)
- On the Autocorrelation of the Stock Market (2021)
- The short of it: Investor sentiment and anomalies (2012)
- Returns to buying winners and selling losers: Implications for stock market efficiency (1993)
- The other side of value: The gross profitability premium (2013)
- Mean reversion in stock prices: Evidence and implications (1988)
- Intraday Patterns in the Cross-Section of Stock Returns (2010)
- A Simple Implicit Measure of the Effective Bid-Ask Spread in an Efficient Market (1984)
- Estimating the Components of the Bid/Ask Spread (1988)
- Measuring the Information Content of Stock Trades (1991)
- An Empirical Analysis of the Limit Order Book and the Order Flow in the Paris Bourse (1995)
- Fluctuations and Response in Financial Markets: The Subtle Nature of ``Random’’ Price Changes (2004)
- Theory for Long-Memory in Supply and Demand (2005)
- Optimal Execution of Portfolio Transactions (2001)
- Direct Estimation of Equity Market Impact (2005)
- Inferring Trade Direction from Intraday Data (1991)
- Market Intraday Momentum (2018)
- Proof that Properly Anticipated Prices Fluctuate Randomly (1965)
- Random Walks in Stock Market Prices (1965)
- Efficient Capital Markets: A Review of Theory and Empirical Work (1970)
0 of 48 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.