Pith. sign in

REVIEW 3 major objections 4 minor 41 references

Do Betting Markets Sense a Goal Coming? Evidence from the German Bundesliga

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Betting markets show no sign of sensing an imminent goal.

desk verdict New question, unique dataset, but the bettor-side null is not identified because the latent state can absorb the anticipation effect; the bookmaker side is the credible result. read the letter →

arxiv 2505.21275 v1 pith:IYSJQ4FP submitted 2025-05-27 econ.GN q-fin.EC

David Winkelmann , Christian Deutscher This is my paper
classification econ.GNq-fin.EC
keywords livebettingmarketanticipationstate-spacemodelsregressiontimeseriesanalysisBundesligaefficiency
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper asks whether betting markets can see major news coming, and uses the first goal of a football match as the test event: it is the single most outcome-decisive moment in a low-scoring sport, and its timing becomes public at one well-defined instant. Working with second-by-second odds and stakes from a leading European bookmaker for a full Bundesliga season, the authors test whether implied win probabilities drift toward the team that will score, or whether bettors shift their stake share toward that team, as the goal approaches. Their finding is a uniform absence of anticipation: the inverse-time-to-goal covariate is statistically insignificant for both the bookmaker's odds and the bettors' stakes, after controlling for team strength, time decay, red cards, and expected goals. The practical upshot, if the result is right, is that the in-play market contains no exploitable signal about an imminent goal, and that market participants adjust to news rather than forecast it. The paper also contributes a modelling tool—state-space estimation with a latent market-activity level—that separates any anticipatory signal from the noise and persistence of live betting flows.

What carries the argument

The carrying object is the covariate $\text{mintogoal}_{it}^{-1}$, the reciprocal of the minutes remaining until the first goal in match $i$ at minute $t$. It rises smoothly from near zero early in a scoreless match to exactly 1 in the final minute, so if participants sensed the goal coming, their behaviour should show a level shift proportional to this quantity; the coefficient on it ($\beta_8$ in Equation (4) for the bookmaker and Equation (5) for bettors) is the paper's direct test of anticipation. On the bettor side, the central mechanism is a state-space model in which the relative stake on the scoring team follows a zero-one-inflated $\beta$ distribution whose mean is a logit link of covariates plus a latent activity state $s_t = \phi s_{t-1} + \sigma_s \epsilon_t$; this latent AR(1) state absorbs the strong serial correlation typical of live betting so that the anticipation coefficient is not confounded by market nervousness. Estimation is by maximum likelihood with the state space discretised into $m = 95$ intervals, and the inference on $\beta_8$ is what carries the paper's null conclusion for the bettor side.

What would settle it

Re-estimate Equations (4) and (5) on the same 1 Hz feed aggregated at 10-second resolution instead of one-minute means; a positive, significant coefficient on $\text{mintogoal}_{it}^{-1}$ at that resolution, or a systematic drop in the scoring team's implied win probability in the final 30 seconds before goals relative to other scoreless minutes, would refute the paper's conclusion that neither side anticipates the first goal.

Watch

Extended reading notes

Core claim

Football is a low-scoring game, and in this dataset the team that scores first wins 70.6% of matches, so the first goal is the clearest possible major-news event. The paper's central discovery is a null result, stated positively: on the way to the first goal, neither the bookmaker nor the bettors move in a direction that would indicate they know it is coming. For the bookmaker, adding the inverse minutes-to-go variable $\text{mintogoal}_{it}^{-1}$ to a regression that already includes pre-match strength, elapsed time and its square, red cards, and the expected-goals differential leaves the implied win probability of the scoring team unchanged (coefficient $-0.005$, 95% confidence interval $[-0.012, 0.002]$). For bettors, the same variable entered into a zero-one-inflated $\beta$ state-space model of relative stakes is likewise insignificant (coefficient $0.096$, interval $[-0.031, 0.222]$), in a model whose latent activity state is strongly persistent ($\hat{\phi} = 0.974$). The paper concludes that neither side anticipates the first goal: bookmaker odds are rigid up to expected goals and red cards, and bettor stake share shows no directional drift, so the negative result is symmetric across the two sides of the market.

Load-bearing premise

The test assumes that anticipation, if it exists, would appear as a smooth climb in odds or betting share that intensifies as the goal approaches, and that this climb would show up in minute-by-minute averages; any signal concentrated in the final seconds of a minute, or shaped differently, would escape detection.

Editorial extensions

If this is right

  • No betting strategy built on reading an imminent goal—backing the team whose odds or stake share drifts upward in the minutes before a goal—can earn positive returns, because the paper finds no such drift to exploit.
  • Bookmakers are not at a competitive disadvantage from failing to anticipate goals: their odds already incorporate expected goals and match dynamics, and the residual rigidity costs them nothing if bettors cannot forecast goals either.
  • The result separates anticipation from insider information: since goal news is public at a single instant and no pre-event adjustment appears, the market can be characterised as reacting to information rather than forecasting it, a distinction the paper notes is hard to draw in financial markets.
  • The strongly persistent latent activity state ($\hat{\phi} \approx 0.97$) shows that volatility clustering, not anticipation, dominates the dynamics of live betting flows, so any future test of pre-event behaviour must control for the market activity level.
  • Against this baseline, a significant rise in stakes on a team shortly before a goal becomes interpretable as a red flag for wrongdoing such as match-fixing, an application the paper explicitly points to.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The authors leave the resolution question open: their 1 Hz feed is averaged into one-minute bins, so anticipation concentrated in the final seconds of a minute would be invisible to the test exactly as specified; re-aggregating at 10-second resolution is a direct extension of their own data.
  • The same inverse-time-to-event design transfers to other step-change news with publicly known timing—red cards and penalty awards in football, wickets in cricket, or scheduled corporate announcements—where pre-event drift can be measured against the same null.
  • If the null survives finer resolution, the binding constraint is the forecasting difficulty of goals, not the market's processing speed; the efficiency property would then belong to the information problem itself, and thinner-information settings such as lower-division football remain the natural place to hunt for anticipation.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper uses 1 Hz in-play betting data from a leading European bookmaker covering the 2018/19 German Bundesliga season to test whether bookmakers and bettors anticipate the first goal of a match. The authors aggregate the data to one-minute intervals and model bookmaker behavior with linear regressions of implied win probability on pre-match probability, elapsed time, red cards, expected-goal differences, and the inverse minutes-to-goal (Equation 4). Bettor behavior is modeled with zero-one-inflated beta state-space models in which relative stakes on the team that eventually scores depend on the same covariates plus a latent AR(1) 'market activity' state (Equation 5). In both cases the coefficient on 1/mintogoal is statistically insignificant, and the paper concludes that neither bookmakers nor bettors anticipate the first goal. An appendix regression for bettors without the latent state, however, reports a significant negative coefficient that the main text does not discuss.

Significance. If the negative result is correct, it would be a useful contribution to the sports-betting and market-efficiency literatures, using a rare high-frequency dataset to show that an imminent, high-impact event (the first goal) is not priced in ahead of time by either market side. The bookmaker-side analysis is transparent and sensible, and the state-space machinery is state of the art for this type of betting data. However, the bettor-side test suffers from an identification problem: the highly persistent latent state can absorb exactly the gradual anticipatory movements the test is designed to detect, and the appendix evidence shows material instability in the coefficient of interest. The current manuscript therefore does not establish the universal negative that is its headline claim. The paper is, nonetheless, worth revising because the data are unique and the bookmaker result is a solid baseline; the bettor question can in principle be addressed with event-study designs or alternative specifications.

major comments (3)
  1. [Section 4.2, Table 4 and Appendix D, Table 7] The paper's conclusion that bettors do not anticipate goals rests on the insignificant coefficient on mintogoal^-1 in the state-space models (0.096, [-0.031, 0.222] in the final SSM), but the beta regression without the latent state (Table 7) reports -0.109 with 95% CI [-0.216, -0.003], which is significant at the 5% level. This sign reversal is not discussed anywhere in the manuscript. Because the state-space model is the basis for the bettor-side conclusion, the authors must explain this instability or present additional evidence that the latent state is not absorbing the anticipation effect.
  2. [Section 4.1, Equation (5) and Table 4] The latent AR(1) state has estimated persistence phi = 0.974, giving it nearly one parameter per minute per match. A gradual ramp-up in relative stakes before the goal—the precise pattern the test targets—can be absorbed by the state sequence rather than by the 1/mintogoal regressor. The insignificant coefficient therefore only shows that no incremental effect of the specific functional form remains after the latent process has been fit; it does not establish the absence of anticipation. The authors should provide an identification argument or a simulation study showing that the SSM can recover a known anticipatory effect.
  3. [Section 2 and Equations (4)-(5)] The anticipation test is built on the covariate 1/mintogoal, with data aggregated to one-minute means. Any anticipation effect concentrated in the final seconds of a minute, or following a non-monotonic time pattern, would not be captured by this model. Given that the central claim is a universal negative ('neither odds nor stakes show significant adjustments right before the first goal'), the authors should either use the underlying 1 Hz data for an event-study around the goal, or test alternative functional forms, to rule out anticipation at higher frequency.
minor comments (4)
  1. [Appendix B, Table 6] The first row/column label appears as 'nrinterval' though the variable is t; please check the alignment of the correlation matrix.
  2. [Section 3.2] The claim that 'this result remains robust even when the expected goals covariate is excluded' is not shown in any table; please provide the corresponding estimates.
  3. [Figure 2] The gray hatched area corresponds to halftime, but it is not clear how the halftime minutes are treated in the regressions (e.g., excluded or included with t=45-60); please clarify.
  4. [Section 5] The statement 'This suggests that bookmakers do not face negative consequences from failing to anticipate goals' is a strong inference from the insignificant coefficient; consider softening or providing an economic interpretation.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's central claim is an empirical null result based on observational data, not a derivation that reduces to its own inputs.

full rationale

The paper asks whether bookmakers and bettors anticipate the first goal, and answers by testing the coefficient on 1/mintogoal in regression and state-space models (Eqs. 4 and 5). The regressor is constructed from observed goal times and is not a fitted parameter, and the response variables (implied probability, relative stakes) are measured data. No parameter is fitted to a subset and then renamed a prediction, no uniqueness theorem is imported from prior work by the same authors, and no ansatz is smuggled in via citation: the SSM framework and zero-one-inflated beta distribution are standard tools cited from the literature, including some co-authored work, but their use does not force the conclusion. The Appendix D sensitivity analysis (the 1/mintogoal coefficient changes from -0.109 without the latent state to +0.096 with it) raises a legitimate identification concern: the highly persistent latent state (phi = 0.974) could absorb gradual anticipatory stake movements, making the bettor-side null not fully identified. However, this is a statistical modeling limitation, not circularity: the latent state is not defined in terms of the goal-time regressor, and the test remains a falsifiable empirical probe. The acknowledged drawback about high information availability in top leagues is an external-validity concern, not a circular step. Thus the analysis is self-contained against external benchmarks and scores 0.

Assumptions & free parameters 3 free parameters · 5 assumptions · 1 invented entities

The paper rests on standard statistical model choices rather than dedicated derivations. The main test uses a single covariate to represent goal proximity, and the latent state is a flexible construct that can absorb systematic patterns. No physical or economic entities are invented; the only new unobservable is the AR(1) market activity state.

free parameters (3)
  • phi (autoregressive persistence parameter) = 0.974 (final SSM)
    Estimated from data; controls the persistence of the latent market activity state. If anticipation created a slow drift, phi could capture it, but this is part of the fitted model.
  • sigma_s (state error standard deviation) = 0.183 (final SSM)
    Estimated state noise; together with phi it defines the latent activity process.
  • beta_8 (coefficient on 1/mintogoal) = 0.096 (final SSM, CI [-0.031, 0.222]); -0.109 (Appendix D, CI [-0.216, -0.003])
    The key coefficient for the anticipation hypothesis; it is fitted to data and its insignificance drives the main conclusion. The differing signs across models are a source of concern.
assumptions (5)
  • domain assumption Relative stakes follow a zero-one-inflated beta distribution.
    Section 4.1. This distributional choice is standard for proportional outcomes with boundary values, but it constrains how stake shares can vary.
  • domain assumption The latent market activity state follows a Gaussian AR(1) process.
    Section 4.1 and Appendix C. The AR(1) structure is standard in stochastic volatility models, but it imposes a specific dynamic shape on unobserved activity.
  • ad hoc to paper Anticipation, if present, is captured by the linear predictor term beta * (1/mintogoal) in Equations (4) and (5).
    This is the paper's chosen functional form for an impending goal. It is not derived from theory; it is a modeling choice that determines the test's power.
  • domain assumption Expected goals (xG) data accurately summarize all observable pre-goal in-match information available to market participants.
    The bookmaker model controls for xgdiff/t as the measure of in-match dynamics. If xG mismeasures information, the residual anticipation test could be biased.
  • standard math Implied probabilities derived from odds via the margin-correction formula reflect the bookmaker's true probabilities.
    Section 2. This is a common transformation in sports economics, but it assumes the margin is distributed proportionally.
invented entities (1)
  • Latent market activity state s_t
    purpose: Captures unobserved serial correlation in the level of betting activity and mean relative stakes.
    Introduced in Section 4.1 as a continuous AR(1) latent variable. It has no direct observable counterpart outside the model; it is inferred from the stake series and could absorb gradual pre-goal drifts, which makes its identification with 'activity' rather than 'anticipation' a modeling assumption.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Do Betting Markets Sense a Goal Coming? Evidence from the German Bundesliga." pith.science (2026). https://pith.science/paper/IYSJQ4FP

@misc{pith2026250521275,
  author       = {Pith},
  title        = {Pith review of: Do Betting Markets Sense a Goal Coming? Evidence from the German Bundesliga},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IYSJQ4FP}},
  note         = {Machine review of arXiv:2505.21275}
}
read the original abstract

We use the fertile ground of betting markets to study the anticipation of major news in financial markets. While there is a considerable body of literature on the accuracy and efficiency of betting markets after important in-match events, there are no studies dealing with the anticipation of such events. This paper tracks bookmaker odds and betting stakes to provide insights into the movement of both prior to goals. Utilising high-resolution (1 Hz) data from a leading European bookmaker for a full season of the top German football league, we analyse whether market participants anticipate major news. In particular, we consider the case of the first goal scored within a match, with its strong impact on the match outcome. Using regression models and state-space models (SSMs) accounting for an underlying market activity level, we investigate whether the bookmaker adjusts odds and bettors tend to place higher stakes on the scoring team right before the first goal is scored. Our results indicate that neither side of the market anticipates goals by significantly adjusting their behaviour.

Figures

Figures reproduced from arXiv: 2505.21275 by the authors.

Figure 1
Figure 1. Boxplot of the minute Ti in which the first goal was scored in the 289 non-scoreless matches of the 2018/19 German Bundesliga season [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Implied probabilities (solid lines) and relative stakes (dotted lines) in the 2018/19 [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Dependence structure of the SSM with latent state [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

41 extracted references · 41 canonical work pages

  1. [1]

    and Wilfling, B

    Al-Anaswah, N. and Wilfling, B. (2011). Identification of speculative bubbles using state-space models with M arkov-switching. Journal of Banking & Finance , 35(5):1073--1086

    work page 2011

  2. [2]

    J., and Lucas, A

    Barra, I., Hoogerheide, L., Koopman, S. J., and Lucas, A. (2017). Joint B ayesian analysis of parameters and states in nonlinear non- G aussian state space models. Journal of Applied Econometrics , 32(5):1003--1026

    work page 2017

  3. [3]

    Ben-Rephael, A., Da, Z., and Israelsen, R. D. (2017). It depends on where you search: i nstitutional investor attention and underreaction to news. Review of Financial Studies , 30(9):3009--3047

    work page 2017

  4. [4]

    Buhagiar, R., Cortis, D., and Newall, P. W. (2018). Why do some soccer bettors lose more money than others? Journal of Behavioral and Experimental Finance , 18:85--93

    work page 2018

  5. [5]

    Che, X., Feddersen, A., and Humphreys, B. R. (2017). Price setting and competition in fixed odds betting markets. In The Economics of Sports Betting , pages 38--51. Edward Elgar Publishing

    work page 2017

  6. [6]

    and Hui, S

    Choi, D. and Hui, S. K. (2014). The role of surprise: understanding overreaction and underreaction to unanticipated events using in-play soccer betting market. Journal of Economic Behavior & Organization , 107:614--629

    work page 2014

  7. [7]

    Chua, C. L. and Tsiaplias, S. (2019). Information flows and stock market volatility. Journal of Applied Econometrics , 34(1):129--148

    work page 2019

  8. [8]

    Clarke, J., Chen, H., Du, D., and Hu, Y. J. (2019). Fake news, investor attention, and market reaction. PSN: Political Communication (Topic)

    work page 2019

Show all 41 references
  1. [9]

    and Reade, J

    Croxson, K. and Reade, J. J. (2014). Information and efficiency: goal arrival in soccer betting. The Economic Journal , 124(575):62--91

    2014

  2. [10]

    R., Gaire, S., Mahatara, S., Joshi, R

    Dahal, K., Pokhrel, N. R., Gaire, S., Mahatara, S., Joshi, R. P., Gupta, A., Banjade, H., and Joshi, J. (2023). A comparative study on effect of news sentiment on stock price prediction with deep learning architecture. PLOS ONE , 18(4):e0284695

    2023

  3. [11]

    Deutscher, C., Frick, B., and \"O tting, M. (2018). Betting market inefficiencies are short-lived in G erman professional football. Applied Economics , 50(30):3240--3246

    2018

  4. [12]

    J., and Singleton, C

    Elaad, G., Reade, J. J., and Singleton, C. (2020). Information, prices and efficiency in an online betting market. Finance Research Letters , 35:101291

    2020

  5. [13]

    R., and Soebbing, B

    Feddersen, A., Humphreys, B. R., and Soebbing, B. P. (2017). Sentiment bias and asset prices: evidence from sports betting markets and social media. Economic Inquiry , 55(2):1119--1129

    2017

  6. [14]

    and McHale, I

    Forrest, D. and McHale, I. G. (2019). Using statistics to detect match fixing in sport. IMA Journal of Management Mathematics , 30(4):431--449

    2019

  7. [15]

    Gil, R. G. R. and Levitt, S. D. (2007). Testing the efficiency of markets in the 2002 W orld C up. The Journal of Prediction Markets , 1(3):255--270

    2007

  8. [16]

    and Rizvi, S

    Haroon, O. and Rizvi, S. A. R. (2020). COVID -19: Media coverage and financial markets behavior -- a sectoral inquiry. Journal of Behavioral and Experimental Finance , 27:100343

    2020

  9. [17]

    Heston, S. L. and Sinha, N. R. (2017). News vs. sentiment: predicting stock returns from news stories. Financial Analysts Journal , 73(3):67--83

    2017

  10. [18]

    G., and Rossi, P

    Jacquier, E., Polson, N. G., and Rossi, P. E. (2002). Bayesian analysis of stochastic volatility models. Journal of Business & Economic Statistics , 20(1):69--87

    2002

  11. [19]

    and Sunderman, M

    Jain, P. and Sunderman, M. (2014). Stock price movement around the merger announcements: insider trading or market anticipation? Managerial Finance , 40(8):821--843

    2014

  12. [20]

    Jarrell, G. A. and Poulsen, A. B. (1989). Stock trading before the announcement of tender offers: insider trading or market anticipation? The Journal of Law, Economics, and Organization , 5(2):225--248

    1989

  13. [21]

    Kitagawa, G. (1987). Non- G aussian state-space modeling of nonstationary time series. Journal of the American Statistical Association , 82(400):1032--1041

    1987

  14. [22]

    Lago‐Peñas, C., Gómez-Ruano, M., Megías-Navarro, D., and Pollard, R. (2016). Home advantage in football: examining the effect of scoring first on match outcome in the five major E uropean leagues. International Journal of Performance Analysis in Sport , 16(2):411--421

    2016

  15. [23]

    and King, R

    Langrock, R. and King, R. (2013). Maximum likelihood estimation of mark-recapture-recovery models in the presence of continuous covariates. The Annals of Applied Statistics , 7(3):1709--1732

    2013

  16. [24]

    L., and Zucchini, W

    Langrock, R., MacDonald, I. L., and Zucchini, W. (2012). Some nonstandard stochastic volatility models and their estimation using structured hidden M arkov models. Journal of Empirical Finance , 19(1):147--161

    2012

  17. [25]

    Mews, S., Langrock, R., \"O tting, M., Yaqine, H., and Reinecke, J. (2024). Maximum approximate likelihood estimation of general continuous-time state-space models. Statistical Modelling , 24(1):9--28

    2024

  18. [26]

    Michels, R., \"O tting, M., and Langrock, R. (2023). Bettors’ reaction to match dynamics: evidence from in-game betting. European Journal of Operational Research , 310(3):1118--1127

    2023

  19. [27]

    Miller Jr, T. W. and Rapach, D. E. (2013). An intra-week efficiency analysis of bookie-quoted NFL betting lines in NYC . Journal of Empirical Finance , 24:10--23

    2013

  20. [28]

    Na, S., Su, Y., and Kunkel, T. (2019). Do not bet on your favourite football team: the influence of fan identity-based biases and sport context knowledge on game prediction accuracy. European Sport Management Quarterly , 19(3):396--418

    2019

  21. [29]

    and Ferrari, S

    Ospina, R. and Ferrari, S. L. (2012). A general class of zero-or-one inflated beta regression models. Computational Statistics & Data Analysis , 56(6):1609--1623

    2012

  22. [30]

    \"O tting, M., Langrock, R., and Deutscher, C. (2018). Integrating multiple data sources in match-fixing warning systems. Statistical Modelling , 18(5-6):483--504

    2018

  23. [31]

    \"O tting, M., Michels, R., Langrock, R., and Deutscher, C. (2024). Demand for live betting: an analysis using state-space models. Applied Stochastic Models in Business and Industry , 40(2):527--541

    2024

  24. [32]

    A., Stasinopoulos, M

    Rigby, R. A., Stasinopoulos, M. D., Heller, G. Z., and De Bastiani, F. (2019). Distributions for Modeling Location, Scale, and Shape: Using GAMLSS in R . Chapman and Hall/CRC

    2019

  25. [33]

    and Grund, S

    Robitzsch, A. and Grund, S. (2022). miceadds: some additional multiple imputation functions, especially for 'mice' . R package version 3.13-12

    2022

  26. [34]

    Sauer, R. D. (1998). The economics of wagering markets. Journal of Economic Literature , 36(4):2021--2064

    1998

  27. [35]

    Stan e k, R. (2017). Home bias in sport betting: evidence from C zech betting market. Judgment and Decision Making , 12(2):168--172

    2017

  28. [36]

    Thaler, R. H. and Ziemba, W. T. (1988). Anomalies: parimutuel betting markets: racetracks and lotteries. Journal of Economic Perspectives , 2(2):161--174

    1988

  29. [37]

    Tunyi, A. A. (2021). Revisiting acquirer returns: evidence from unanticipated deals. Journal of Corporate Finance , 66:101789

    2021

  30. [38]

    Winkelmann, D., Deutscher, C., and \"O tting, M. (2021). Bookmakers’ mispricing of the disappeared home advantage in the G erman B undesliga after the COVID-19 break. Applied Economics , 53(26):3054--3064

    2021

  31. [39]

    Winkelmann, D., \"O tting, M., Deutscher, C., and Makarewicz, T. (2024). Are betting markets inefficient? E vidence from simulations and real data. Journal of Sports Economics , 25(1):54--97

    2024

  32. [40]

    Zeileis, A. (2004). Econometric computing with HC and HAC covariance matrix estimators. Journal of Statistical Software , 11(10):1--17

    2004

  33. [41]

    L., and Langrock, R

    Zucchini, W., MacDonald, I. L., and Langrock, R. (2017). Hidden M arkov M odels for T ime S eries: A n I ntroduction U sing R . Chapman and Hall/CRC, New York

    2017

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.