Reflection after book "The signal and the noise"

First reflection is on my daily job of financial forecast of the company.

The financial forecast focus on profitability of the next 12 months. EBITDA (Earning Before Interest, Tax and Depreciation) is considered the most important output. The forecast is based on sales forecast from customers or sales. And operation team builds the plan according to foretasted revenue and cost is calculated based on that.

The book "The signal and the noise" mentioned about three basic rules of good forecast

• Rule 1: Provide forecast and range
• Rule 2: Always validate forecast and results and adjust model accordingly
• Rule 3: Look for consensus

If we use these rules to examine my forecast work.

Rule 1:

Yes. we do provide a range of EBITDA or risks & opportunities after providing EBITDA forecast

But, 

1 .the range is not based on statistics, but on risks & opportunities (scenario analysis)

2. scenario considered is very limited. Most of time, it focuses on revenue only. not on operation side.

Rule 2:

Yes. we do compare forecast and actual monthly. And we look in details about where the variance comes from.

But, 

1. the range of EBITDA is never validated

2. error of the forecast might be understood in P&L detail level, but model is not adjusted

-  For material, no purchase is foretasted, so, PPV is hardly predictable. And inventory adjustment is seldom forecast.

-  For labor, no labor schedule is foretasted. so, labor usage is forecast poorly

-  For direct cost, no procurement policy on factory supplies and no validation of efficiency

-  For OH/SGA labor, it's hard to tell when new hire will join and the turnover rate is high

-  For OH/SGA expense, a big error comes from repair and maintenance. no real tracking on repair and maintenance.

Rule 3:

Sort of. The consensus comes from if the forecast are aligned with historical trend.

But, we never/seldom look outside of the company to see performance of other companies.