Spot Regressions with Candlesticks
Pith reviewed 2026-05-18 07:08 UTC · model grok-4.3
The pith
Candlestick high and low prices yield lower-risk estimators for spot betas than conventional returns.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
By minimizing a quadratic risk criterion under a fixed-k asymptotic framework, the paper constructs candlestick-based estimators for the parameters of spot regressions, including spot beta; these estimators exploit the additional information contained in intra-period high and low prices and are shown to reduce finite-sample estimation risk relative to estimators that use only open-to-close returns while supporting valid hypothesis tests.
What carries the argument
Quadratic-risk-minimizing candlestick estimator under fixed-k asymptotics, which folds high and low prices into the regression parameter estimates.
If this is right
- Spot beta estimates become more accurate in short samples typical of high-frequency windows.
- Tests for market neutrality or other linear restrictions on spot parameters gain power without sacrificing asymptotic validity.
- Risk-management and asset-pricing applications can exploit intra-period price ranges rather than discarding them.
- Deviations from neutrality in assets such as Bitcoin are detectable and vary with volatility regime.
Where Pith is reading between the lines
- The same candlestick construction could be applied to other regression-based quantities such as spot factor loadings or time-varying covariances.
- Portfolio managers might obtain more responsive beta hedges by updating estimates at the frequency of the underlying candlesticks.
- Extensions to multivariate systems would allow joint inference on several spot betas using the same high-low information.
Load-bearing premise
The fixed-k asymptotic framework combined with quadratic risk minimization continues to deliver correct size and lower risk when applied to real candlestick series that contain high and low prices.
What would settle it
A Monte Carlo experiment or real-data exercise in which the candlestick estimator exhibits higher mean-squared error than the return-based estimator or in which the proposed test rejects at rates far from the nominal level.
read the original abstract
Betas from spot regressions are central to asset pricing and risk management, as measures of systematic risk. This paper develops a new estimation and inference framework for spot regressions by leveraging high-frequency candlesticks, extending conventional (open-to-close) returns with intra-period high/low prices. Specifically, I construct candlestick-based estimators of regression parameters, including spot beta, by minimizing a quadratic risk under a fixed-k asymptotic framework. I then develop a feasible hypothesis testing procedure for spot betas with correct asymptotic size. Simulation results show that the proposed estimator reduces estimation risk relative to return-based estimators, especially in small samples, and the test achieves notably higher power. I apply the framework to assess the market neutrality of Bitcoin using 1-minute data on IBIT and SPY, finding deviations from neutrality, particularly in high-volatility periods.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper develops a framework for estimating spot regression parameters such as beta by constructing candlestick-based estimators (leveraging open, high, low, and close prices) that minimize quadratic risk under a fixed-k asymptotic regime. It derives a feasible test for spot betas that achieves correct asymptotic size, reports simulation evidence of lower estimation risk and higher power relative to conventional return-based estimators (especially in small samples), and applies the method to 1-minute IBIT-SPY data to document deviations from market neutrality in Bitcoin, concentrated in high-volatility periods.
Significance. If the results hold, the approach could meaningfully improve finite-sample precision for systematic-risk measurement in asset pricing and risk management by extracting additional information from intra-period extremes. The fixed-k quadratic-risk construction and the reported simulation gains in estimation risk and test power would represent a practical advance for high-frequency applications, particularly in volatile assets, provided the underlying continuous-path assumptions are robust.
major comments (2)
- The quadratic-risk minimization that defines the candlestick estimator is derived from the joint law of the (open, high, low, close) vector under a continuous semimartingale specification inside each interval. When the true price path contains jumps, realized highs and lows deviate from this law, so the optimized weights no longer minimize actual risk; the simulation improvements and the Bitcoin application (high-volatility regimes) are therefore conditional on the no-jump DGP used in both derivation and Monte Carlo design.
- The fixed-k asymptotic framework is presented as delivering correct size and improved finite-sample performance, yet the manuscript provides limited derivation details, data-exclusion rules, and error-bar reporting. This leaves unexamined potential bias-variance trade-offs that could affect the central claim of superior estimation risk and test power.
minor comments (2)
- The abstract states that the test 'achieves notably higher power' without reporting specific power values, sample sizes, or direct comparisons to the return-based benchmark; adding these numbers would strengthen the simulation section.
- Clarify the exact mapping from the candlestick vector to the quadratic-risk objective and state whether any tuning parameters beyond k are involved.
Simulated Author's Rebuttal
We thank the referee for the detailed and constructive report. We address each major comment below, providing the strongest honest responses consistent with the manuscript's assumptions and results, and indicate revisions where the manuscript will be updated.
read point-by-point responses
-
Referee: The quadratic-risk minimization that defines the candlestick estimator is derived from the joint law of the (open, high, low, close) vector under a continuous semimartingale specification inside each interval. When the true price path contains jumps, realized highs and lows deviate from this law, so the optimized weights no longer minimize actual risk; the simulation improvements and the Bitcoin application (high-volatility regimes) are therefore conditional on the no-jump DGP used in both derivation and Monte Carlo design.
Authors: We agree that the quadratic-risk minimization and the resulting optimal weights are derived under the continuous semimartingale assumption without jumps, which is the maintained framework throughout the paper. In the presence of jumps the realized high and low will generally differ from the continuous-path law, so the estimator is no longer guaranteed to be quadratic-risk optimal. The Monte Carlo design and the 1-minute IBIT-SPY application are therefore conducted under this maintained assumption. We have added an explicit discussion of this limitation in the revised manuscript, including a caveat on the Bitcoin results in high-volatility intervals and a brief outline of how jump-detection filters could be combined with the candlestick approach in future work. The continuous-path case remains the relevant benchmark for the fixed-k theory and for many high-frequency applications where jumps are infrequent at the 1-minute horizon. revision: partial
-
Referee: The fixed-k asymptotic framework is presented as delivering correct size and improved finite-sample performance, yet the manuscript provides limited derivation details, data-exclusion rules, and error-bar reporting. This leaves unexamined potential bias-variance trade-offs that could affect the central claim of superior estimation risk and test power.
Authors: We appreciate the call for greater transparency on the fixed-k asymptotics. In the revised version we have expanded the main-text derivation of the quadratic-risk objective and the feasible test statistic, moved additional technical steps to a new appendix subsection, and stated the precise data-exclusion rules applied in both the simulations and the empirical exercise (intervals with zero volume or price changes exceeding five times the local volatility are dropped). We have also added error bars to all simulation tables and figures and included a new set of Monte Carlo results that vary k explicitly, documenting the associated bias-variance trade-offs. These changes directly support the claims of lower estimation risk and higher test power while clarifying the scope of the fixed-k results. revision: yes
Circularity Check
No significant circularity; estimator defined via explicit quadratic-risk minimization on candlestick observables
full rationale
The paper constructs its candlestick-based spot-beta estimator by minimizing an asymptotic quadratic risk criterion under a fixed-k framework. This is a direct definition of the estimator from the observed open-high-low-close vector and the maintained continuous-semimartingale law inside each interval; it does not reduce by construction to a fitted parameter taken from prior literature, nor does any load-bearing step rely on a self-citation whose content is itself unverified. The derivation therefore remains self-contained against external benchmarks and receives the default non-finding.
Axiom & Free-Parameter Ledger
free parameters (1)
- k
axioms (1)
- domain assumption Candlestick high and low prices provide additional information beyond open-to-close returns that improves spot beta estimation under high-frequency sampling.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.