---
title: Rolling-Horizon Optimization (RHO)
url: https://www.emergentmind.com/topics/rolling-horizon-optimization-rho
type: topic
---

# Rolling-Horizon Optimization (RHO)

Rolling-Horizon Optimization (RHO) is a sequential decision procedure in which, at each decision epoch, an optimization problem is solved over a finite future window, only the first-stage action is implemented, new information is observed, and the horizon is shifted forward. In control settings this receding-horizon procedure is equivalent to model predictive control when future states are predicted explicitly from a system model, while in stochastic programming and large-scale mathematical optimization it functions as a tractability device for replacing a single long-horizon problem by a sequence of smaller but coupled subproblems [2504.05883] [2102.04874].

## 1. Canonical formulation and policy interpretation

A general stochastic control formulation used in the RHO literature specifies a state \(S_t\), decision \(x_t\in\mathcal X_t\), exogenous information \(W_{t+1}\), transition map \(S_{t+1}=S^M(S_t,x_t,W_{t+1})\), and per-period cost \(C(S_t,x_t)\). The corresponding base model is

\[
\min_{\pi\in\Pi}\;\mathbb E^\pi\Bigl[\sum_{t=0}^T C\bigl(S_t,X_t^\pi(S_t)\bigr)\;\Big|\;S_0\Bigr],
\]

subject to

\[
S_{t+1}=S^M\bigl(S_t,X_t^\pi(S_t),W_{t+1}\bigr),\quad X_t^\pi(S_t)\in\mathcal X_t.
\]

Within this viewpoint, RHO is not the base problem itself but a policy class for solving it: at time \(t\), one solves a truncated lookahead problem on a smaller horizon, applies the first action, and re-optimizes after the system evolves [1703.04644].

For finite-horizon multistage stochastic programming, the truncated lookahead problem at stage \(t\) with horizon length \(\tau\) is written as \(\mathcal P_t(\tau;x_{t-1},\xi_t)\), where the terminal cost-to-go \(Q_{t+\tau}(\cdot)\) is set to zero. The resulting RH policy implements only the first-stage decision \(x_t^\tau\), then rolls forward one period and repeats [2102.04874]. In scenario-based variants, this first stage is typically coupled by non-anticipativity constraints. In stochastic home energy management, for example, first-hour battery decisions satisfy \(b^+_{s,t}=b^+_{s',t}\), \(b^-_{s,t}=b^-_{s',t}\), and \(s_{s,t}=s_{s',t}\) for all scenarios \(s,s'\), so that the action executed at the current time is scenario-independent [2203.08639].

This formulation makes RHO simultaneously a decision rule, a decomposition strategy, and a way of embedding updated measurements or forecasts into repeated optimization.

## 2. Deterministic, stochastic, robust, and multi-horizon variants

The inner model solved inside an RHO loop varies substantially across the literature. A classical deterministic rolling horizon, often written with point forecasts \(\tilde W_{t,t'}\), solves a short-window deterministic program and implements only the current action. In practice, however, practitioners modify such models with buffer stocks, schedule slack, reserve requirements, or forecast corrections. A central reinterpretation is that these “deterministic” remedies are parametric cost function approximations (CFAs), for example

\[
X_t^{\rm CFA}(S_t\mid\theta) =\argmin_{x_t\in\mathcal X_t}\;\Bigl\{\,C(S_t,x_t) +\sum_{f\in\mathcal F}\theta^c_f\,\phi_f(S_t,x_t)\Bigr\} \;\text{s.t.}\; A_t\,x_t\;\le\;b_t(\theta),
\]

where \(\theta\) adjusts the objective and right-hand sides and is tuned in the stochastic base model, typically by simulation [1703.04644].

A second family uses explicit stochastic lookahead models. In residential energy management, the rolling window is a two-stage stochastic MILP that minimizes expected electricity cost over a 24-hour horizon with 100 demand scenarios drawn from online multivariate probabilistic-load-forecast methods, each with probability \(\pi_s=1/100\) [2203.08639]. In production planning, the rolling subproblem is a two-stage capacitated lot-sizing model with here-and-now lot sizes for \(t\le\tilde T\) and recourse lot sizes, inventories, and backlogs for \(t>\tilde T\), embedded in discrete-event simulation [2402.14506].

A third family is robust. In microgrid scheduling, uncertain loads, photovoltaic generation, EV demand, and market prices are represented through box and budget uncertainty sets, and a static robust counterpart is repeatedly solved over a moving window [2307.05154]. Related energy-management work combines an online scheduling layer with robust optimization, where the value of launching a new RHO iteration depends on forecast-driven information gain rather than a fixed timetable [2311.11307].

A fourth family is deterministic re-optimization with learned point forecasts rather than explicit scenarios. In network-constrained V2X coordination, the rolling subproblem is an MIQP solved hourly on the shrinking horizon \(\mathcal H_t=\{t,t+1,\dots,24\}\), using realized data up to \(t\) and point forecasts from the GRU-EN-TFD model for later periods [2502.09290].

A fifth family is multistage multi-horizon stochastic programming. In domestic renewable energy system design, strategic investment stages and short-term operational subtrees are linked in a single tactical multi-horizon tree, and a rolling-horizon matheuristic solves submodels containing both a master investment problem and operational recourse, optionally with first- or second-order stochastic dominance constraints that preserve time consistency [2505.15167].

Taken together, these variants show that RHO is compatible with deterministic LPs, MILPs, MIQPs, robust counterparts, and scenario-based stochastic programs; the rolling mechanism is separate from the precise uncertainty model used within each window.

## 3. Horizon length, overlap, and inter-window coupling

The central design question in RHO is how far to look ahead and how much of the computed plan to execute before re-optimization. In infinite-horizon discounted multistage stochastic programming with bounded stage costs \(|f(x,\xi)|\le\kappa\) and discount factor \(\gamma\in(0,1)\), the lookahead policy \(\pi_\tau\) satisfies

\[
V(\pi_\tau)-V(\pi_\infty)\;\le\;\frac{\kappa}{1-\gamma}\;\gamma^\tau,
\]

and an \(\epsilon\)-sufficient forecast horizon is therefore

\[
\tau^*_\epsilon\;=\;\left\lceil \frac{\log\!\big(\tfrac{\epsilon(1-\gamma)}{\kappa}\big)}{\log(\gamma)}\right\rceil.
\]

For finite horizons, Siddig, Song and Khademi propose an ADP-style heuristic that learns a state-dependent horizon map \(T(s)=\sum_{i=0}^P\theta_i\phi_i(s)\), so that the lookahead depth varies by roll rather than remaining fixed [2102.04874].

Overlap is the standard mechanism for preserving continuity across windows. In offline pickup-and-delivery with time windows, the day is decomposed into overlapping windows \(W_i=(i\,t_s,\,i\,t_s+T_w]\); each subproblem is solved on the active requests in the window, but only the first \(t_s\) minutes are frozen before the horizon advances. This overlapping structure provides the boundary stitching that makes temporal decomposition viable for narrow-window PDPTW instances [2303.03475]. In flexible job-shop scheduling, the analogous parameters are planning window \(H\), execution step \(S\le H\), and overlap size \(H-S\); at each iteration the next \(H\) operations are scheduled, but only the first \(S\) are executed [2502.15791].

Other domains formalize inter-window coupling through state carryover or fixed first-stage commitments. In rolling lot-sizing, here-and-now quantities \(Q_{it}\) are fixed across scenarios for \(t\le\tilde T\), while later decisions remain adjustable [2402.14506]. In multi-horizon stochastic MILP for domestic energy system design, once a window has been solved, first-stage investments are fixed and battery state is carried into the next iteration [2505.15167]. In healthcare operating-theatre scheduling, week 1 is implemented and frozen, then weeks 2 through \(W+1\) are replanned under rolling-horizon “no-move-right” constraints [1808.10139]. In stochastic waste collection, inventory levels \(u_i^n\) are passed from one subproblem to the next as a warm start [2405.14499].

These mechanisms control the basic RHO trade-off: larger horizons reduce myopia but increase subproblem size, whereas smaller horizons improve responsiveness and runtimes but amplify truncation effects.

## 4. Forecasting, learning, and information-driven replanning

A defining feature of contemporary RHO is the tight coupling between forecasting and optimization. In stochastic home energy management, two online probabilistic-load-forecast methods are used—RLS-Free and Copula-Free—and each hourly update draws 100 net-demand scenarios for the 24-hour horizon, with no scenario reduction [2203.08639]. In production planning, forecast evolution is modeled through additive MMFE dynamics, and customer behaviors are parameterized as Type A, Type B, and Type C, corresponding respectively to stable, single late-update, and periodic-update forecasts [2402.14506]. In collaborative caching, demand prediction uses an auto-regressive model \(\tilde\lambda_n^{(t)}=\mu_t[\sum_{\tau=1}^H \beta_\tau \lambda_n^{(t-\tau)}+\epsilon_t]\), and the rolling policy optimizes against cumulative predicted demand over an \(N\)-step lookahead [1907.13516].

Recent work increasingly learns either the forecasts themselves or the structure of the rolling procedure. In V2X value stacking, the GRU-EN-TFD forecaster predicts building load, PV output, and EV arrivals; the RHO then re-solves an MIQP each hour on the remaining part of the day [2502.09290]. In dynamic microgrid management, a knapsack-type model selects the best iteration start times by quantifying the expected value of refined PV forecasts and realized EV arrivals, so that re-optimization is concentrated where information changes most [2307.05154]. In threshold-based online rolling horizon control, an offline longest-path problem produces contribution values \(v_j\), these are collapsed into a threshold \(\tau\), and a new iteration is launched online whenever the estimated contribution exceeds a dynamic threshold \(f(t,\mathrm{used})\tau\); with a median threshold, the resulting online rule is shown to be \(1/4\)-competitive [2311.11307].

Learning also appears inside the optimization loop itself. Zhao et al. use a rolling-horizon MILP decomposition for neural network verification, where overlapping layer subgraphs are processed sequentially to tighten pre-activation bounds [2401.05280]. Li et al. propose L-RHO for long-horizon flexible job-shop scheduling: a neural network predicts which overlapping operations are unlikely to change machine assignment between consecutive windows, and those assignments are fixed in the next subproblem, reducing redundant re-optimization [2502.15791].

This body of work suggests a shift from fixed-calendar rolling schemes toward information-adaptive rolling schemes in which forecasting, scenario construction, and variable fixing are themselves optimized.

## 5. Representative applications and reported performance

Empirical results are strongly domain-specific, but the literature reports recurrent gains when uncertainty is material, horizons are long, or monolithic formulations are intractable.

| Domain | Rolling subproblem | Reported findings |
|---|---|---|
| Energy storage management | Parametric CFA within deterministic LP lookahead | Under perfect forecasts, the unparameterized RHO is optimal; as forecast noise increases, CFA methods outperform the benchmark by 13–26%, and “lookup-table” and “exponential” forms give up to ~26% improvement for large \(\sigma_f\) [1703.04644] |
| Stochastic home energy management | Two-stage stochastic MILP, \(H=24\), 100 scenarios | RLS-SP yields up to 10–20% lower costs than naive in winter and autumn; “Naive+RLS-SP” gives an extra \(\sim 8\%\) cost saving; RLS-SP is within 3–6% of the perfect-information bound, but naive outperforms pure stochastic RHO in spring and summer [2203.08639] |
| Production planning | Scenario-based stochastic lot sizing in simulation | For highly nervous Type C forecasts, stochastic optimization cuts 40–60% vs. MRP and 5–25% vs. deterministic; at \(\alpha=12.5\%\) and 98% utilization, stochastic beats deterministic by ~14%; best number of scenarios is ≈30, with solve times ~15–20 s [2402.14506] |
| Robust microgrid scheduling | Dynamic rolling robust LP | The dynamic rolling horizon model outperforms the classical version by up to 57% in costs and increases the local use of PV by up to 11% [2307.05154] |
| Distributed UAV coverage | Rolling-horizon distributed MIP | With \(T=6\), mission completion is roughly 15% faster than myopic coverage and still 6× more computationally efficient than centralized RHO; per-step solve times are ∼0.2–1 s [2504.05883] |
| Neural network verification | Rolling-horizon OBBT MILPs on layer subgraphs | OBBT-RH verifies 80 of 90 instances and is 1.25× faster overall than full OBBT+MIP while preserving near-OBBT bound quality [2401.05280] |
| Online energy management | Threshold-scheduled robust RHO | Total electricity costs are reduced by more than 85% compared to a classical rolling horizon framework and by more than 50% compared to a dynamic offline rolling horizon framework [2311.11307] |
| Long-horizon flexible job-shop scheduling | Learning-guided restricted CP-SAT subproblems | L-RHO accelerates RHO by up to 54% and improves objective value by up to 21% relative to Default RHO [2502.15791] |

Across these studies, the strongest gains typically occur when forecast revisions are frequent, constraints are tight, or the baseline is either myopic or static. Reported exceptions are equally informative: perfect forecasts can eliminate the need for corrective parameterization, and finite-window myopia can make simple fallback policies preferable in low-volatility regimes.

## 6. Limitations, misconceptions, and current research directions

A persistent misconception is that the main methodological divide is between “deterministic rolling horizon” and “stochastic programming.” The parametric-CFA perspective argues instead that both are policies for a stochastic base model, differing mainly in where approximation is placed: scenario trees and recourse structure on one side, tunable objective and constraint modifications on the other [1703.04644].

Myopia remains the principal technical limitation. In waste collection, Theorem 3.2 shows that when \(R=0\) there exist feasible instances for which \(z^{RH,1}=-\infty\) although the full multistage model is feasible; empirically, \(W=1\) produced infinite or very large profit loss in several small instances, whereas \(W=2,3\) traded ∼29.8% average profit gap for 90–98% CPU-time savings [2405.14499]. Similar finite-window effects appear in residential energy management, where pure stochastic RHO can misjudge the long-term value of PV self-consumption in spring and summer [2203.08639].

Safeguards against such pathologies are an active research direction. In situationally aware multi-tier load restoration, safeguarded constraints are added so that each segment remains above a \((1-\varepsilon)\)-fraction of its optimal restoration potential, explicitly mitigating the myopia of plain rolling horizon control [2510.02502]. In large multi-horizon stochastic MILPs, lower-bound schemes such as Strategic Wait-and-See, Scenario Multistage Grouping, and Scenario Multistage Clustering are used to assess the optimality gap of rolling-horizon heuristics; with \(\hat h=2\), \(\hat h^R=1\), and \(\phi_e=1/3\), the SFR3 matheuristic achieved solutions within 0.05–0.2% of optimum on small and medium instances and within 0.7% of the best lower bound on large cases [2505.15167].

A second limitation is computational scaling. Longer horizons tighten bounds and improve non-myopic behavior, but they enlarge the MILP or MIQP rapidly. In rolling-horizon verification, larger layer horizons yield diminishing returns while subproblem costs grow super-linearly [2401.05280]. In distributed coverage control, longer horizons add \(\mathcal O(T\times M\times|\mathcal P|)\) binary variables [2504.05883]. This has motivated decomposition, parallelism, online scheduling of iteration times, and learning-guided fixing of stable decisions.

Current work therefore concentrates on three complementary directions: stronger theoretical guidance for selecting horizon length, more adaptive policies for deciding when to re-optimize, and hybrid schemes that combine rolling windows with simulation, learning, or lower-bound certification. The cumulative evidence does not treat RHO as a single algorithmic template; rather, it identifies a broad family of sequential optimization policies whose effectiveness depends on how horizon truncation, uncertainty representation, and inter-window coupling are engineered for the underlying problem class.

Source: https://www.emergentmind.com/topics/rolling-horizon-optimization-rho