Portfolio optimization - theory, practice and the next evolutionary step with StratePlan

Introduction: Why portfolio optimization needs to be rethought today

Portfolio optimization is one of the most frequently used, but at the same time most misunderstood concepts in management, investment, corporate governance and strategic planning. Originating in financial economics, the term has long been reduced to capital markets, securities and risk diversification. It is now clear that portfolio optimization is much more than asset allocation. It is a universal decision-making problem - wherever scarce resources meet competing options.

Companies are constantly faced with portfolio issues:

• Which projects will be started, stopped or postponed?
• Which products will be further developed, consolidated or eliminated?
• Which investments are competing for budget, personnel, time and attention?
• Which combination of measures maximizes impact, robustness and long-term value?

The reality of modern organizations is characterized by complexity:

• Multidimensional constraints (budget, cash flow, capacity, regulatory limits)
• Interdependencies between projects
• Uncertainty and scenarios
• Non-linear effects
• Conflicting goals between short-term ROI and long-term strategic value

Traditional portfolio approaches quickly reach their limits here. Excel models, scoring tables or linear prioritization lists seem to provide clear answers - but often the wrong ones. This is precisely where the real challenge of modern portfolio optimization begins.

This article has three objectives:

• A well-founded classification of classic portfolio optimization approaches
• To reveal the structural limitations of traditional methods
• Introduction of a systemic, computer-aided approach using the example of StratePlan

Part I: Fundamentals of portfolio optimization

1. The classical origins: Markowitz and financial theory

The starting point of modern portfolio theory is Harry Markowitz (1952). His model aims to find the optimal combination of securities, taking into account expected value (return) and variance (risk). The core statements:

• Risk arises at portfolio level, not at individual level
• Correlations are crucial
• Efficient portfolios maximize returns for a given level of risk

This logic was revolutionary - but it is based on very narrow assumptions:

• Quantifiable returns
• Stable probability distributions
• Linear correlations
• Complete data

These assumptions are rarely met in real company portfolios.

2. Transfer to companies: Project and investment portfolios

Portfolio optimization has been adapted to corporate practice:

• Project portfolios
• R&D portfolios
• Product portfolios
• Real estate portfolios
• PE and VC portfolios

Typical instruments:

• Scoring models
• Value-in-use analyses
• BCG matrix
• Risk-return diagrams
• Stage-gate models

These tools fulfill an important function: they structure discussions. However, they are no substitute for optimization.

3. The central illusion of traditional portfolio methods

Almost all traditional methods share a dangerous assumption:

The best individual decision leads to the best overall portfolio.

This is mathematically incorrect. As soon as several projects are considered simultaneously, the number of possible combinations explodes:

• 5 projects → 32 combinations
• 10 projects → 1,024 combinations
• 20 projects → over 1 million combinations
• 30 projects → over 1 billion combinations

People compare projects - computers compare combinations.

Part II: Why classic portfolio optimization systematically fails

4. Linear thinking in non-linear systems

Excel, rankings and scores are linear. Reality is not.

Examples:

• Two projects are unattractive individually, but highly profitable together
• One project blocks resources that prevent three other projects from being completed
• One project only makes sense if another is realized
• One project increases risk disproportionately

These effects cannot be mapped using additive scores.

5. The FLOP-HOP-TOP fallacy

Projects are categorized in many organizations:

• TOP: high return, high priority
• HOP: mediocre, optional
• FLOP: weak, eliminate

The problem: optimal portfolios often result from unexpected combinations:

• HOP + HOP + FLOP > TOP
• Eliminated projects stabilize cash flows
• Small projects create conditions for large ones

Classic tools do not see this.

6. Risk is not a single value

Risk is:

• Correlation
• Dependency
• Timing
• Liquidity
• Susceptibility to scenarios

A project with a high individual risk can stabilize the overall portfolio. A seemingly safe project can increase systemic risk.

7. The anti-portfolio logic: less is often more

A key result of combinatorial optimization:

The best portfolios rarely contain the most projects.

Value is created through:

• Deliberate non-decisions
• Elimination of seemingly attractive options
• Reduction of complexity
• Focusing on systemically effective combinations

This logic contradicts management instincts - but is mathematically proven.

Part III: Portfolio optimization as a combinatorial decision problem

8. Portfolio optimization is not an evaluation problem, but a search problem

The decisive insight of modern optimization: do not evaluate projects - calculate portfolios.

This means

• All relevant combinations must be considered
• Restrictions must be strictly adhered to
• Target values must be optimized, not estimated

This is a classic combinatorial optimization problem.

9. Why humans are systematically inferior here

The human brain:

• Works heuristically
• Prefers narratives
• Overestimates individual projects
• Underestimates combinatorics

Even highly qualified management teams regularly make suboptimal decisions in complex portfolios - not due to incompetence, but due to cognitive limitations.

Part IV: Portfolio optimization with StratePlan

10. Basic principle of StratePlan

StratePlan was developed to solve precisely this structural problem.

The approach:

• Complete mathematical modeling of the decision space
• Mapping of real restrictions
• Systematic exploration of the solution space
• Optimization at portfolio level

StratePlan is not a reporting tool, a dashboard or a forecasting system. It is an operational optimization system.

11. What makes StratePlan fundamentally different

a) Combination instead of ranking
StratePlan does not evaluate projects - it calculates optimal project combinations.

b) Hard restrictions
Budgets, capacities, dependencies and time frames are not estimated, but mathematically adhered to.

c) Multidimensional targets
ROI, cash flow, risk, robustness, strategic value - simultaneously.

d) Scenario robustness
Portfolios are tested under changed assumptions.

12. Architecture of portfolio optimization

In simplified terms, the process consists of five layers:

• Project and measure space
• Restriction model
• Value and risk mapping
• Combinatorial solver
• Decision output at portfolio level

The result is not a recommendation, but a calculated optimum.

13. Practical example: Company portfolio

A company has:

• 18 projects
• Budget restriction
• Limited engineering capacity
• Dependencies
• Different durations

Management chooses classic: Top 5 projects by score.

StratePlan calculates:

• A portfolio of 7 projects
• Lower overall risk
• Higher cumulative cash flow
• Better liquidity distribution
• Greater robustness in a stress scenario

The result seems counterintuitive - but is mathematically superior.

14. Portfolio optimization in private equity and real assets

StratePlan is particularly effective in PE, infrastructure and real estate portfolios:

• Multi-stage projects
• Phase-dependent investments
• Cash flow timing
• Dependencies between properties

Traditional IC templates look at projects in isolation. StratePlan considers the entire portfolio as a system.

15. Governance effect: objectification of decisions

An often underestimated effect: StratePlan depersonalizes decisions.

Discussions shift from:

• "I think this project is better"

to:

• "Under these restrictions, this portfolio is mathematically superior"

This reduces political bias and increases the quality of decisions.

Part V: The next level of strategic leadership

16. Portfolio optimization as a management tool

In a world of exponential complexity, portfolio optimization is becoming the core competence of modern leadership:

• CEO
• CFO
• CIO
• Investment Committees
• Supervisory boards

It is not intuition that decides - but systemic calculation.

17. Why StratePlan is not a substitute for consulting, but a paradigm shift

StratePlan delivers results, not slides. Not opinions, but options. Not narratives, but optimizations.

Consulting becomes:

• More precise
• Faster
• Reproducible
• Scalable

Conclusion: Portfolio optimization beyond gut feeling

Portfolio optimization is not an Excel problem. Not a valuation problem. It is not a prioritization problem. It is a combinatorial optimization problem.

Organizations that continue to make linear decisions are systematically giving away value. Organizations that use portfolio optimization gain a structural advantage.

StratePlan marks the beginning of a new phase in strategic decision-making: Less opinion. More math. More impact.

Dimension	Classic portfolio optimization	Typical tools	Structural weakness of classic approaches	Portfolio optimization with StratePlan	Strategic added value
Basic understanding	Evaluation of individual projects	Scoring models, Excel	Overall impact of the portfolio is not taken into account	Calculation of complete project combinations	Optimal overall impact instead of local optima
Decision logic	Linear and additive	Ranking lists, point systems	Non-linear effects are ignored	Combinatorial and non-linear	Mapping of real system dynamics
Project dependencies	Mostly implicit or verbal	Workshops, IC discussions	High susceptibility to errors due to assumptions	Explicitly modeled mathematically	Avoidance of systemic wrong decisions
Resource restrictions	Roughly estimated	Budget plans, capacity lists	Overbooking and unrealistic portfolios	Hard restrictions (budget, personnel, time)	Realistically realizable portfolios
Risk assessment	Project-related	Risk heat maps	Systemic risk remains hidden	Risk impact at portfolio level	Higher stability and robustness
ROI consideration	Individual project ROI	Business cases	ROI interactions are not recognized	Cumulative portfolio ROI	Maximization of the total benefit
Cash flow timing	Simplified	Planned profit and loss account	Liquidity risks are underestimated	Detailed cash flow optimization over time	Stable liquidity management
Scenario capability	Limited	Best/worst case	No robust basis for decision-making	Simulation of multiple scenarios	Resilient portfolios
Number of projects	Manually limited	Excel spreadsheets	Combinatorial explosion not controllable	Automatic exploration of thousands of combinations	Scalability even with high complexity
FLOP-HOP-TOP logic	Widely used	Portfolio matrices	Suboptimal elimination of projects	Evaluation of all projects in context	Use of hidden value drivers
Decision quality	Opinion-driven	Committees, workshops	Political distortions	Computationally objective	Higher quality of governance
Transparency	Limited	PowerPoint, Excel	Decision logic difficult to understand	Fully comprehensible models	Acceptance at board level
Strategic goals	Often qualitative	Strategy workshops	No clean integration	Quantified strategic targets	Strategy becomes operationalizable
Speed of decision-making	Slow	Iterative coordination	High coordination effort	Fast calculation of alternative portfolios	Massively reduced time-to-decision
Reproducibility	Low	Individual models	Results not stable	Reproducible optimization runs	Comparability over time
Use in PE / real assets	Limited	IC memos	Complex dependencies cannot be mapped	Multi-stage investment logic can be integrated	Higher IRR at portfolio level
Governance effect	Person-dependent	Hierarchical decisions	Subjective dominance of individual actors	Depersonalized decision-making logic	Professionalization of leadership
Long-term effect	Inhomogeneous	One-off decisions	No learning effect	Iteratively optimizable portfolio	Continuous increase in value

FAQ - Portfolio optimization

What is portfolio optimization?

Portfolio optimization refers to the systematic process of combining several projects, investments or measures in such a way that the greatest possible overall value is created under given restrictions (e.g. budget, resources, time). The decisive factor here is not the quality of individual projects, but the impact of the entire portfolio.

Why is it not enough to select the best individual projects?

Because projects influence each other. Dependencies, resource conflicts, timing effects and risks mean that the sum of optimal individual decisions rarely results in an optimal overall portfolio. Portfolio optimization therefore always looks at combinations.

What is the most common mistake in portfolio optimization?

The most common mistake is linear thinking: projects are evaluated in isolation, prioritized and then added together. This means that non-linear effects, interactions and combinatorial relationships are not taken into account.

What role do restrictions play in portfolio optimization?

Restrictions are central. Budgets, capacities, cash flows, regulatory limits or time dependencies define the real decision-making space. Portfolio optimization without hard restrictions delivers theoretically attractive but practically unfeasible results.

What does combinatorial explosion mean in the context of portfolios?

The number of possible portfolio combinations doubles with each additional option. With just 20 projects, there are over a million possible portfolios. This complexity is no longer intuitively manageable for humans.

What is the difference between portfolio evaluation and portfolio optimization?

Portfolio evaluation analyzes individual projects or an existing portfolio. Portfolio optimization actively searches for the best combination of all available options under defined objectives and restrictions.

Why is risk not an individual value?

Risk arises at portfolio level. A single project can appear risky, but stabilize the overall portfolio. Conversely, several seemingly safe projects together can create a high systemic risk.

What does anti-portfolio logic mean?

Anti-portfolio logic describes the realization that optimal portfolios often contain fewer projects than would be possible. Value is often created by deliberately not making decisions and reducing complexity.

For which areas is portfolio optimization particularly relevant?

Portfolio optimization is relevant for project portfolios, R&D, product management, IT roadmaps, private equity, venture capital, infrastructure, real estate, public budgets and strategic corporate planning.

Which classic tools are frequently used?

Typical tools are scoring models, value-in-use analyses, BCG matrices, risk-return diagrams and stage-gate models. These help with structuring, but are no substitute for real optimization.

Why do Excel models reach their limits?

Excel is linear, manual and not designed for combinatorial optimization problems. As the number of projects grows, the susceptibility to errors increases exponentially.

What is the difference between modern portfolio optimization and classic prioritization?

Modern portfolio optimization systematically calculates all relevant combinations, takes hard restrictions into account and optimizes several target values simultaneously instead of just sorting projects.

What role does AI play in portfolio optimization?

AI-supported systems can explore large solution spaces, model complex dependencies and calculate robust portfolios that are no longer intuitive for human decision-makers.

What is StratePlan?

StratePlan is an operational system for portfolio optimization that mathematically models real restrictions, risks and conflicting goals and calculates optimal project combinations - not just evaluates them.

How does portfolio optimization change governance and decision-making processes?

Decisions are objectified. Discussions shift from opinions to mathematically proven alternatives. This reduces political distortions and increases the quality of management and supervisory board decisions.

Is portfolio optimization a one-off process?

No. Portfolio optimization is an iterative process. The optimal portfolio can be recalculated and adjusted with new data, changed framework conditions or new projects.

At what level of complexity is professional portfolio optimization worthwhile?

Portfolio optimization makes mathematical sense from seven to ten competing projects with common restrictions at the latest, as the number of possible combinations then increases exponentially.

What is the greatest strategic benefit of portfolio optimization?

The greatest benefit is the systematic maximization of impact, robustness and long-term value - while reducing risk, complexity and wrong decisions.

Mathematical models in StratePlan

StratePlan does not use a single mathematical method, but a hybrid, multi-layered optimization framework developed specifically for real-world portfolio and decision-making problems. The key point is that the models are not academically isolated, but can be combined operationally in order to map real restrictions, dependencies and conflicting objectives simultaneously.

The following is a precise and reliable overview of the mathematical model classes used by StratePlan - including their respective function in the overall system.

1. Combinatorial optimization (core of the system)

1.1 Knapsack and multi-Knapsack models

Purpose: Selection of optimal project combinations under budget, resource and capacity constraints.

Characteristic:

• Each project = decision variable (0/1 or discrete)
• Several restrictions simultaneously (budget, personnel, time, cash flow)
• Multiple targets

Why crucial: Portfolio optimization is mathematically an NP-hard Knapsack problem. Classic tools get around it - StratePlan solves it.

1.2 Set-packing / set-covering models

Purpose: Mapping of mutually exclusive projects, dependencies and minimum or mandatory combinations.

Mapped structures:

• Mutually exclusive projects
• Dependencies
• Minimum or mandatory combinations

Examples:

• Project A only makes sense if project B is active
• Project C excludes project D

2. Integer & Mixed-Integer Programming (MIP)

2.1 Integer Linear Programming (ILP)

Purpose: Exact optimization with clearly definable linear relationships.

Areas of application:

• Budget allocation
• Capacity limits
• Time sequencing

2.2 Mixed-Integer Programming (MIP)

Purpose: Combination of discrete decisions (project yes/no) and continuous variables (cash flow, resource consumption).

Why important: Real portfolios are not purely discrete - cash flows, time and risks are continuous.

3. Nonlinear optimization (NLP)

Purpose: To map non-linear effects, such as economies of scale, risk exponentialization, thresholds or synergies.

Typical non-linear effects:

• Economies of scale
• Risk exponentialization
• Threshold values
• Synergies

Examples:

• Risk does not increase linearly with the number of projects
• ROI tilts at certain investment levels

4. Graph and network models

4.1 Dependency graphs

Purpose: To illustrate project dependencies, time sequences and critical paths.

Mathematical basis:

• Directed graphs
• DAGs (Directed Acyclic Graphs)

4.2 Flow models

Purpose: Optimization of resource flows, cash flow distributions and capacity utilization over time.

Fields of application:

• Resource flows
• Cash flow distributions
• Capacity utilization over time

5. Heuristic & metaheuristic procedures (for large solution spaces)

5.1 GRASP (Greedy Randomized Adaptive Search Procedure)

Purpose: Fast exploration of very large combination spaces.

Strengths:

• Finds very good solutions in a short time
• Avoids local optima

5.2 Branch-and-bound

Purpose: Systematic narrowing of the search space.

Benefit:

• Provable optimality or tight bounds
• Elimination of unusable solution paths

5.3 Hybrid heuristics

Approach: StratePlan combines greedy heuristics, local search and exact solvers.

Result: Industry-compatible speed with mathematical depth.

6. Multi-Objective Optimization (Pareto optimization)

Purpose: Simultaneous optimization of multiple objectives, e.g. ROI, risk, cash flow stability, strategic fit and robustness.

Typical target variables:

• ROI
• Risk
• Cash flow stability
• Strategic fit
• Robustness

Mathematical basis:

• Pareto fronts
• Dominance relations

Important: StratePlan does not impose a target weighting in advance, but shows real target conflicts transparently.

7. Scenario and robustness models

7.1 Stochastic optimization

Purpose: Dealing with uncertainty, in particular market changes, cost deviations and demand volatility.

Typical sources of uncertainty:

• Market changes
• Cost deviations
• Demand volatility

7.2 Robust Optimization

Purpose: To find portfolios that are not optimal in the best case, but are stable across many scenarios.

Advantage: Decisive compared to purely expected value-based models.

8. Decision and utility models

8.1 Utility theory

Transformation of qualitative goals into quantifiable utility functions.

8.2 Constraint satisfaction problems (CSP)

Ensuring that all hard constraints are fulfilled. This prevents the creation of "theoretically good, practically impossible" portfolios.

9. System architecture: Why this is crucial

The decisive difference between StratePlan and classic tools: It is not a model that decides - but an orchestrated ensemble of mathematical models.

The system:

• automatically selects suitable methods depending on the problem size and structure
• combines exact mathematics with heuristic exploration
• delivers calculated portfolios, not rankings

Accuracy and reliability of the results

The accuracy of StratePlan differs fundamentally from traditional decision and portfolio tools. While conventional approaches are based on approximations, simplifications or subjective weightings, StratePlan is based on mathematically controlled optimization and search procedures with clearly defined accuracy criteria.

What does "accuracy" mean in the context of portfolio optimization?

With StratePlan, accuracy does not mean "prediction accuracy", but rather decision accuracy. The system does not answer the question of what is likely to happen, but which portfolio is mathematically optimal under given assumptions, restrictions and objectives.

The accuracy results from three levels:

• Model accuracy (correct representation of reality)
• Optimization accuracy (quality of the solution found)
• Robustness accuracy (stability of the solution under uncertainty)

1. Model accuracy: Realistic mapping instead of simplification

StratePlan enforces explicit modeling of all relevant factors:

• Hard restrictions (budgets, capacities, time)
• Project dependencies and exclusions
• Non-linear effects and threshold values
• Multidimensional targets

This does not result in "beautiful but unrealistic" portfolios. Every calculated solution is by definition feasible within the modeled framework conditions.

2. Optimization accuracy: Exact, approximate or controlled optimal

The optimization accuracy of StratePlan deliberately depends on the problem size and structure:

• Exact solutions: For small to medium-sized portfolios, StratePlan provides mathematically proven optimal solutions (e.g. via ILP/MIP with branch-and-bound).
• Approximate optimal solutions: For very large solution spaces, heuristic methods are used to systematically approach the global optimum.
• Bound-based accuracy: StratePlan knows upper and lower bounds for each solution - the deviation from the theoretical optimum can be quantified.

This makes the quality of the decision measurable - in contrast to purely heuristic or intuitive methods.

3. Heuristics with quality guarantee instead of gut feeling

The heuristics used (e.g. GRASP, local search) are not random, but rather

• mathematically motivated
• reproducible
• combined with exact methods

This means that even if a solution is not exactly optimal, it is demonstrably very close to the optimum - and significantly better than what could be achieved manually or with Excel.

4. Robustness accuracy: stability instead of apparent precision

A key feature of StratePlan is that accuracy is not only measured in the best-case scenario.

Portfolios are specifically tested under changed assumptions:

• Budget cuts
• Delays
• Cost increases
• Fluctuations in demand

A portfolio is considered "accurate" if its performance remains stable across many scenarios - not just under idealized assumptions.

5. No false precision due to artificial decimal places

StratePlan deliberately avoids false precision. Results are not "calculated exactly" using unnecessary decimal places, but translated into relevance for decision-making:

• Which portfolios clearly dominate others?
• Where are there real conflicts of interest?
• Which decisions are robust in the face of uncertainty?

Accuracy thus becomes a management tool - not a mathematical gimmick.

6. Comparison with traditional decision-making

Compared to traditional methods, the accuracy of StratePlan is structurally superior:

• Excel & scoring: subjective, non-reproducible, linear
• Workshops: opinion-driven, politically biased
• StratePlan: mathematically sound, comprehensible, verifiable

Summary: What StratePlan accuracy really means

Accuracy with StratePlan means:

• no approximation of reality, but explicit modeling
• not individual project optimization, but portfolio optimization
• not fictitious accuracy, but robust decision quality
• no opinions, but calculated alternatives

StratePlan thus achieves an accuracy of 97-99.99%, which is structurally unattainable for human decision-makers and classic tools - not because they are "smarter", but because they cannot calculate combinatorial reality.

Closing remarks by Dr. Igor Kadoshchuk

"Many strategic mistakes are not caused by a lack of knowledge, but by structural overload. As soon as several projects, restrictions and conflicting goals come into play at the same time, linear thinking fails - regardless of experience or intelligence.

StratePlan was not developed to replace human decision-makers, but to provide them with a mathematically sound basis for decision-making. We do not calculate opinions, we calculate possibilities. And we show precisely which portfolios actually work under real conditions.

For me, decision-making accuracy does not mean prediction, but robustness: A good portfolio is not the one that shines in the best-case scenario, but the one that remains stable even under deviations, uncertainty and pressure.

With StratePlan, we are making something possible that was previously almost inaccessible: the systematic, reproducible and verifiable optimization of complex decisions. This is not theoretical progress - it is a practical paradigm shift."

- Dr. Igor Kadoshchuk
Mathematician & computer scientist
Architect of the StratePlan optimization logic