Introduction
A disruptive technology will often drive a paradigm shift in applications and deployment scenarios. Valuing the opportunity is a difficult task using analysis of existing markets because the output is predicated on the basis of the existing technology landscape. We seek to answer two questions.

How do we attempt to value opportunities with a high degree of uncertainty?

Subsequently, what development strategy do we pursue to allocate finite resource and maximise returns in light of uncertainty?
To answer the former we develop a Monte Carlo model to capture the key uncertainty drivers to produce a riskreturn profile for individual opportunities and the portfolio as a whole. The latter involves the development of a real option model to capture managerial flexibility through the lifetime of the project.
Uncertainty Drivers
The two principle uncertainties ‘Likelihood of market entry’– whether we can enter the market full stop with a product and subsequently ‘Market penetration and growth rate’ –the income over time for a product. The risks are subdivided into risks under our control (intrinsic risk factors) and those out of our control (extrinsic risk factors)
Likelihood of Market Entry 

This is a binary event in that we either enter the market or don’t 

Intrinsic Risk 
Extrinsic Risk 
Performance metrics Cost Development timeline Entry Strategy 
Market size Market price Industry trends Competitive landscape Regulation 
Uncertainty Variable 

Management would take an opinion on the likelihood of a successful market entry, which accounts for, and balances the range of risks. This can be codified into a likelihood estimate e.g. 90% chance of successful entry. This will be reviewed and revised over time as uncertainty is resolved e.g. successful outcome of a test program. In this way we have a common framework to link markets with a high degree of certainty with speculative applications and start to make a comparative analysis. 
Market Penetration and Growth Rates 

This describes the uptake of the product once we have entered the market 

Intrinsic Risk 
Extrinsic Risk 
Performance differentiation Cost Growth Strategy 
Market size – Market price Industry trends Competitive landscape Regulation 
Uncertainty Variables 

An important fact is that many parameters are only known ex post, when all uncertainty has been resolved (e.g. we have good estimates of cost at high volume but won’t know the exact figure until we get there). The basic principle is to use a random distributions to describe variables such as cost, price, volume sold, aggregate market size. A Monte Carlo NPV cash flow model simulates the impact of all uncertain variables. This provides us with an estimate of return and the associated risk. In speculative applications the uncertainty is captured in specifying the distribution. 
Timing – This is the other key variable to affect the aggregate value of an opportunity. We discount cash flows to account for the time value of money, often the discount rate is large as a way to try and capture risk. We want to capture uncertainty in an explicit manner through the distribution of possible outcomes and Monte Carlo analysis, which means we don’t need to use an inflated discount rate to account for risk. If we used a high hurdle rate we would be double counting risk. This principle it outlined in Stefan Scholtes’ work and his approach to option analysis through Monte Carlo methods.
Capturing and Representing Portfolio Opportunity and Uncertainty
The ‘bubble’ chart is a good conceptual way to grasp the value, risk and timing associated with a range of applications. The challenge is to embed this into a revenue model that logically captures the structure and information contained within the ‘bubble’ chart.
One advantage of this representation is that we can conduct top down and bottom up analysis and have an ‘apples for apples’ comparison. There are multiple layers of detail that we can use when valuing an opportunity and if you are looking across different industries chances are there will be significant gaps in knowledge. In an ideal world we would have accurate bottom up estimates e.g. pricing, volumes, costs, timing, sales against line itemed customers; this aggregates to an NPV for the opportunity that we depict as a single bubble in the chart. Note the bubble is placed on the time when a product is released even though it will generate revenues over a period of time. You could look at more complicated representations but I think this do the job without over complicating matters. If we do not have accurate low level data we can work with general economic indicators, broader trends and high level analysis for calculations of market potential e.g. assumed market penetration rates. VCs hate the “if we get x% of some arbitrarily large market”, however we usually have to work with imperfect information and if you use multiple methods to get to answers that are the same order of magnitude then you are doing well.
Interactive Portfolio Dashboard – Click here to launch the portfolio dashboard
A large degree of intuition about a portfolio is developed as a result of building the model; the challenge becomes one of involving the executive team and making the model accessible so they can individually add their knowledge and build a deeper understanding. Even a well designed Excel model can be difficult to navigate; we can use dashboard tools which capture all the data but allow the user to navigate and drill down to the most salient areas of interest. One of the best dashboard programs is Crystal Xcelcius, which ‘sits on top’ on Excel, and extracts values from the worksheets for graphing / visualisation and allows users to change values in the model with sliders etc
With this portfolio tool executives can control a range of input parameters and immediately see the effect on the portfolio. The bottom section allows the user to select different product proposals and explore key uncertainty drivers and decisions e.g. should we develop product, when do we start, what happens if development costs are exceeded, first year volumes are lower than expected, cost of goods are higher etc. There is a standalone product income statement, which feeds into the portfolio income statement and cash model. As the portfolio decision space is explored the executive can monitor cash position, total NPV and adjust other parameters such as level of capital investment. The entire management team participates to input their knowledge e.g. sales may have a good idea of the first year volume potential, manufacturing can provide bill of material cost spreads. The results are displayed instantaneously and the team becomes sensitised to how their project fits within the context of the portfolio or company performance, which can help to reduce ‘silo’ mentality.
The intent is not to necessarily find the optimal product mix but rather to provide a first pass ‘best guess’ that has companywide buy in. We can then use refined models to build in uncertainty with Monte Carlo analysis and managerial flexibility with real options.
Monte Carlo Model – Download the excel model here
In the Monte Carlo model we assign distributions to uncertain cells. We then simulate thousand of scenarios, where we sample the uncertain variables and feed them into the revenue model and look at the impact on total revenue.
The model illustrated below is logically correct; however the revenues figures are arbitrary numbers. It is useful to illustrate the Monte Carlo approach.
Likelihood of market entry – We have assigned a likelihood of entry estimate as a percentage. The model generates a random number between 0100% and compares it against the likelihood number. This determines whether the initial market entry condition is met.
Growth Rates – The growth rate for each market is simply a random variable. One sample ‘realises’ a growth figure, which is used to calculate the total return from the opportunity. We sample thousands of growth rates to achieve a distribution of return. This is a simplistic approach; a more complete model would look at the underlying variables at a more granular level i.e. volumes, pricing, cost etc. It is straight forward build up increasing levels of accuracy.
The diagrams below illustrate three different simulated outcomes. It is clear to see that different markets are entered and different growth rates and revenues are realised. This view represents the top level summary of all the markets. In practice more detailed models would lie behind this.
Sample 1: Entry in industrial, military and environmental markets. Total revenue $346
Sample 2: Entry in industrial, military, consumer and medical markets. Total revenue $679
Sample 3: Entry in all markets. Total revenue $785.1
Range of Income across Portfolio
When we simulate a few thousand outcomes with @Risk we can add confidence limits into the income statement for the entire portfolio; we get a better idea returns and their associates likelihood. This more useful that a single point estimate of NPV as it provides a better understanding of how the ‘dice are loaded’.
Portfolio Value at Risk (VAR)
The Value at Risk chart gives the probability of the realised revenue being greater or less than the revenue value on the horizontal axis. The chart below shoes the VAR for a single product, it is clear that there can be significant variance in the returns, which in reality would be compounded farther by the fact that a product with unsuccessful market entry would yield an overall loss.
As more opportunities are added the average return is increased and the downside risk is reduced (fewer eggs in one basket). In the diagram below the lower blue line in the return profile for military and industrial markets only, and the purple line is return profile for all the markets simultaneously. The challenge we haven’t yet addressed is allocation of finite resource to achieve the optimum returns.
[…] diminished. For example if the ‘fit metric’ was 20% we could use this directly in our Monte Carlo portfolio analysis as the threshold for market entry. You can also spot requirement trends that can have a strong […]
[…] of Interactive Dashboards Interactive Portfolio Model – Posting / […]