Gojii as a "Risk-Adjusted Forecasting" Tool

The Association for Financial Professionals recently published a white paper on "Risk-Adjusted Forecasting". The paper discusses the state of the art in forecasting processes with a focus on dealing with the "risk" (what I would call uncertainty) inherent in those forecasts. Our tool Gojii is a risk adjusted forecasting tool as described in this white paper. While I don't agree with everything in the paper there are some great insights. Here's a quote from the white paper introduction:

...adjusting for risk is not about forecast accuracy. “Risk-adjusted forecasting is not about having a crystal ball. It’s about defining the relationship between financial performance and risk/drivers at a high level of precision . . .In many cases the relationships are not linear since multiple risks are integrated”
— Mark Pellerin, principal at Oliver Wyman

Our tool, Gojii, uses system simulation to calculate the impact of uncertainties at a "high level of precision" including feedback and non-linearities in the system. It integrates the simulation with the capability to visualize and analyze the results in a meaningful way.

Overall I like the paper a lot. Certainly, I agree that the concept of risk adjustment is "in front of the market".  And the observations of the disconnect between forecasting and ERP/ERM systems are right on.  

I'd take issue on a couple of things about the paper. Probably most importantly I think that the word "risk" is used to mean both "uncertainty" and "the adverse impact of uncertainty". They're both important but only the latter is actually risk. Sharpening up that language would lead to greater insight. 

Also, I would say that the approaches presented are overly reliant on estimated probabilities in the risk calculations. In Gojii, we take uncertain inputs to be deeply uncertain. By that we mean that different stakeholders will have different perspectives on the nature of the uncertainty and that no closed form probability distribution will adequately express the risk. You can watch the screencasts to see how that works.

And, while I think that the qualitative approaches described may be better than nothing there is overwhelming evidence that without tools to calculate the impact of non-linearities managers will usually get it wrong.

You can find the white paper at the AFP website (login is going to be required) or it appears that you can get it via an advertising linkup with "Wdesk". (Probably easier unless you are already an AFP member). 

Putting "Big Data" to Work Using System Simulation

Firms are making big investments in data but in many cases they are having trouble getting the return from it that they expect. Data, big or otherwise, has to be put to work if it is going to impact management policy and decision making.

In this screencast I describe the value proposition for data and how decision technology based on system simulation puts data to work.

Anylogic System Simulation at the Heart of Gojii

If you've read this blog post, watched the screencasts, or read this presentation then you know that Gojii, and all of our work at DecisioTech is based on system simulation models. System simulation is a type of "cause and effect" modeling that captures the time delays, non-linearities, and especially feedback effects that make the real world the unpredictable place that it is. System simulation can be contrasted with statistical or probabilistic approaches that are only loosely based on cause and effect structures.

The system simulation at the heart of Gojii describes the interactions of a firms operations with the marketplace. It captures the fact that a firms behavior affects the market and vice versa. This simulation is a real challenge to create because the modeling technology needed to model markets is quite different than the technology needed to model a manufacturing line.

The Anylogic Company provides the perfect tools for this situation -- their "multi-method" Anylogic simulation engine. This is the tool that DecisioTech uses in all of their projects. The Gojii simulation uses components based on discrete-event methods, system dynamics approaches, some agent elements, and bits and pieces of Java code. This all comes together seamlessly in the Anylogic simulation engine leaving the modeler free to worry about solving the problem rather than managing the simulation technology.

If you're interested in system simulation then you should be using Anylogic.

Gojii Screencasts

I've been introducing and demo-ing Gojii to a wide variety of audiences (most recently at The Anylogic Conference) and my audiences always tell me that the presentation and live demo tell the Gojii story far more clearly than what we've able accomplish so far with web pages and text description.

So, I've done the obvious thing, and captured the Gojii Introductory pitch and demonstration as screencasts! The introductory 'cast is just over 7 minutes long and the demo runs about 5 minutes 30 seconds. So, in 13 minutes you can get a capsule overview of what we're up to!

Find the screencasts on the Gojii Screencasts Page.