Agricultural Science Digest

Agricultural Science Digest cover
Volume 46 · 2026/10/08

Forecasting Minimum Support Price-anchored Mandi Prices for Wheat and Maize in India: A Comparative Machine-learning and Time-series Framework

Sai Prabhakar Balantrapu
G. Raghavender Raju
Satya Sai Mudigonda
Lalith Aditya Kalur
172 Views

Submitted

2026/08/14

First Online

2026/10/08

doi

10.18805/ag.D-6567

Cite article - Sai Prabhakar Balantrapu, G. Raghavender Raju, Satya Sai Mudigonda, Lalith Aditya Kalur. (2026). Forecasting Minimum Support Price-anchored Mandi Prices for Wheat and Maize in India: A Comparative Machine-learning and Time-series Framework. Agricultural Science Digest. 46: 129–135. doi: 10.18805/ag.D-6567

Abstract

Background: India’s minimum support price (MSP) regime creates an administered floor for wholesale (mandi) prices, making accurate price forecasting the analytical foundation for farmer revenue-risk assessment, government procurement planning and the design of price- or revenue-based crop insurance instruments. Wheat and maize present a sharp institutional contrast: wheat is procurement-binding through the Food Corporation of India, while maize receives little procurement support and its price is left largely to private trade. This study benchmarks machine-learning and classical time-series forecasters for the monthly modal mandi price of both crops across a structurally distinct, post-pandemic hold-out period and defines, in measurable terms, what it means for such a series to be MSP-anchored.

Methods: This analysis was carried out at Sri Sathya Sai Institute of Higher Learning (SSSIHL), Prasanthi Nilayam, Andhra Pradesh, India, during 2026. Daily AGMARKNET records for wheat and maize (April 2008-March 2025) were aggregated into 204 monthly All-India modal-price observations per crop-the mean of daily modal prices-with the corresponding MSP reported for reference. Eight algorithms-ridge regression, support vector regression, random forest, extremely randomised trees, gradient boosting, XGBoost, ARIMA and SARIMA-together with a naïve persistence benchmark, were trained on 177 months (April 2008-December 2022) and evaluated out-of-sample on 27 held-out months (January 2023-March 2025) using the mean absolute error, root mean squared error and mean absolute percentage error (MAPE).

Result: Ridge regression on lagged prices and a month indicator dominated every fitted ensemble and, with ARIMA and SARIMA evaluated as static multi-step forecasts, every time-series competitor for both crops, achieving a MAPE of 2.75 per cent for wheat (mean absolute error ₹ 68.97/quintal) and 2.59 per cent for maize (₹ 55.97/quintal). Tree-based learners and support vector regression deteriorated sharply on the trending test window; SARIMA was competitive only for maize. Against naïve persistence, however, ridge’s edge was modest for wheat and mixed for maize-persistence matched ridge on MAE and MAPE and narrowly beat it on RMSE-indicating ridge’s advantage lies chiefly over the more flexible learners. This is attributed to the near-unit autocorrelation of MSP-anchored series and the inability of tree ensembles to extrapolate beyond their training range, consistent with, though not proving, the institutional asymmetry between procurement-stabilised wheat and procurement-absent maize. These forecasts form the price engine for companion volume-forecasting and procurement-liability studies within the broader actuarial data-science revenue-protection framework.

References

  1. Acharya, S.S. (2004). Agricultural Marketing in India (4 th ed.). Oxford and IBH Publishing. Akaike, H. (1974). A new look at the statistical model identification. IEEE Transactions on Automatic Control. 19(6): 716- 723. Balkrishna, A., Arya, V. and Singh, S.K. (2023). Minimum support price under the aegis of universal basic income: Understanding the implications and way forward: A review. Bhartiya Krishi Anusandhan Patrika . 38(3): 227-237. doi: 10.18805/BKAP640 . Borah, S., Talukdar, U., Barman, U., Deka, N., Baruah, S. and Ahmed, A.A. (2025). Market arrivals and price behaviour of some selected crops: A study from major markets of Assam. Bhartiya Krishi Anusandhan Patrika. 40(2): 180-186. doi: 10.18805/BKAP796 . Box, G.E.P. and Jenkins, G.M. (1976). Time Series Analysis: Forecasting and Control (Rev. ed.). Holden-Day. Breiman, L. (2001). Random forests. Machine Learning. 45(1): 5- 32. Chand, R. (2017). Doubling farmers’ income: Rationale, strategy, prospects and action plan (NITI Aayog Policy Paper No. 1/2017). NITI Aayog. Chen, T. and Guestrin, C. (2016). XGBoost: A Scalable Tree Boosting System. In Proceedings of the 22 nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (pp. 785-794). Cortes, C. and Vapnik, V. (1995). Support-vector networks. Machine Learning. 20(3): 273-297. Dickey, D.A. and Fuller, W.A. (1979). Distribution of the estimators for autoregressive time series with a unit root. Journal of the American Statistical Association. 74(366): 427- 431. Directorate of Marketing and Inspection. (n.d.). AGMARKNET: Agricultural Marketing Information Network [Data set]. Ministry of Agriculture and Farmers Welfare, Government of India. https://agmarknet.gov.in. Drucker, H., Burges, C.J.C., Kaufman, L., Smola, A. and Vapnik, V. (1997). Support vector regression machines. Advances in Neural Information Processing Systems. 9: 155-161. Friedman, J.H. (2001). Greedy function approximation: A gradient boosting machine. Annals of Statistics. 29(5): 1189-1232. Geurts, P., Ernst, D. and Wehenkel, L. (2006). Extremely randomized trees. Machine Learning. 63(1): 3-42. Gulati, A., Saini, S. and Jain, S. (2013). Monsoon 2013: Estimating the Impact on Agriculture (ICRIER Working Paper No. 269). Indian Council for Research on International Economic Relations. Hoerl, A.E. and Kennard, R.W. (1970). Ridge regression: Biased estimation for nonorthogonal problems. Technometrics. 12(1): 55-67. https://www.rma.usda.gov/revenue-protection. Mahajan, S., Sharma, M. and Gupta, A. (2020). ARIMA modelling for forecasting of rice production: A case study of India. Agricultural Science Digest. 40(4): 404-407. doi: 10.18805/ag.D-5029 . Makridakis, S., Spiliotis, E. and Assimakopoulos, V. (2018). Statistical and machine learning forecasting methods: Concerns and ways forward. Plos One . 13(3): e0194889. Makridakis, S., Spiliotis, E. and Assimakopoulos, V. (2020). The M4 competition: 100,000 time series and 61 forecasting methods. International Journal of Forecasting. 36(1): 54-74. Meena, G.L., Sharma, L., Singh, H., Bairwa, K.C. and Kumar, K. (2023). Cost and return analysis of kharif sorghum ( Sorghum bicolor ) in Bhilwara district, Rajasthan. Agricultural Science Digest. 43(2): 196-200. doi: 10.18805/ag.D-5566 . Severini, S., Zinnanti, C., Borsellino, V. and Schimmenti, E. (2021). EU income stabilization tool: Potential impacts, financial sustainability and farmer’s risk aversion. Agricultural and Food Economics . 9: 34. https://doi.org/10.1186/ s40100-021-00205-4.
EDITORIAL BOARD

Guided by Experienced Academic Leadership

Distinguished scholars and subject experts committed to editorial excellence, scientific integrity and rigorous peer review.

View Full Board
Arvind kumar
Chief Editor

Arvind kumar

Former Vice- ChancellorRani Lakshmi Bai Central Agricultural Uni., Jhansi, U.P., INDIA

R.G. Upadhyay
Associate chief editor

R.G. Upadhyay

ProfessorEx -Director Extension Education, VCSG, Uttarakhand University of Horticulture and Forestry, Pauri Garhwal (Uttarakhand) India

Bouderoua Kaddour
Editor

Bouderoua Kaddour

DirectorUniversity Abdelhamid ibn Badis of Mostaganem, Algeria.

G. Rama Rao
Editor

G. Rama Rao

DeanAcharya NG Ranga Agricultural University, India