Skip to content

Tony Kenck insight

How to Choose a Portfolio Method

Start with the decision, the inventory, the urgency, and what you already have, not with the most advanced technique you can name.

Start with appetite, inventory, urgency, and existing model

Method choice is a decision in its own right and is part of the analytic frame. Before reaching for a technique, ask what management is prepared to use, how large and varied the inventory is, how quickly the decision is needed, and what analysis already exists. Those four answers rule most methods in or out.

Standalone analysis

When you have a single opportunity and need a defensible read fast, disciplined standalone economics with shared assumptions may be enough to make the decision visible.

Get the insights fast, then work the answers in detail.

Tony Kenck, Strategic Business Portfolio Management

Comparison and inventory

As the count of opportunities grows, the leverage shifts to a consistent inventory and honest comparison: same standards, same clock.

Rollups

Rollups that aggregate default choices into enterprise views across time are alogical next step after comparison. The emphasis shift from comparing individual opportunities to enterprise performance. They answer how choices combine, which is the question standalone cases ignore.

Simulation

When uncertainty and correlation drive the decision, simulation shows ranges and the drivers behind them, provided the inputs deserve that much precision.

Optimization and combined methods

Optimization helps search large, constrained spaces for high-performance coordinated sets of decisions. Combining optimization with simulation tests whether an efficient portfolio is also robust. Power here is real but easy to misuse.

Sparse data and advanced methods

Sparse-data techniques help when data is thin.

Stochastic optimization is for decisions whose stakes and structure genuinely require it.

A method-selection checklist

Name the decision and its deadline. Size the inventory. Identify what dominates: comparison, time, uncertainty, or interactions. Choose the simplest method that reaches that. Get the insights fast, then work the answers in detail where it may change the decision.

Model Selection FrameworkA five stage decision tree. Stage one, clarify the need: three gates joined by Or, is there an inventory of assets and opportunities, is there management appetite for portfolio, and urgency. Any No, or Urgent with no model, exits to HIPPO and standalone analysis. All Yes joined by And, or a normal or existing model, flows down. Stage two, assess the data: opportunity data availability. Rich and granular data goes to requirement complexity or dependencies; sparse or coarse data feeds inventory analysis, simple roll-up and layering, and deterministic optimization. Loose future requirements with no entity dependencies lead to stage three, time span of interest: under one year reaches comparison rank and cut, inventory analysis, and simple roll-up and layering including simulation; over one year with analysis or few decisions reaches scenarios and manual what-if rollups. Multiple organizational constraints lead to stage four, is performance uncertainty important: No, with many coordinated decisions, reaches deterministic optimization. Yes leads to stage five, number of decisions and model complexity: under five gives probabilistic simulation, over three hundred gives deterministic optimization then simulation, and five to three hundred gives stochastic programming.1CLARIFYTHE NEED2ASSESSTHE DATA3DEFINE TIME& SCOPE4EVALUATEUNCERTAINTY& DECISIONS5DETERMINECOMPLEXITYMODEL SELECTION FRAMEWORKNoOrNoOrUrgent,No modelIs There anInventory ofAssetsand Opps?Mgmt.Appetitefor Pf?Urgency?YesAndYesAndNormal orExisting modelOpportunityDataAvailabilityRich and granularSparse or CoarseRequirementComplexityor DependenciesNo, or very loose,future requirements,no entity dependenciesMultipleorganizationalconstraints,requirements,or entity dependenciesTime Spanof Interest<1 year<1 year<1 year>1 year>1 yearAnalysis orfew decisionsIsPerformanceUncertaintyImportant?NoManycoordinateddecisionsYesNumber ofDecisions(n)/ ModelComplexity< 5> 3005 < n < 300HIPPO andStandaloneAnalysisComparison(rank & cut)InventoryAnalysisSimple roll-upand layering,incl simulationScenarios,Manual What-IfRollupsDeterministicOptimizationProbabilisticSimulationDeterministicOptimizationthen SimulationStochasticProgramming

Scroll the diagram sideways to follow the whole tree.

Tony's method selection framework. Start with the need and available information, then choose the analysis that fits.

Put the idea to work

Take the next useful step

Go deeper in the field manual, or start a low-pressure conversation about the decision in front of your team.

Buy bookConnect with Tony