Why Point Forecasts Are Almost Always Worthless

There's something unsettling about a number with false precision attached to it. "Revenue will be $47.3 million next quarter." "GDP growth will be 2.6%." "Oil prices will average $85 per barrel." The specificity creates confidence. It implies thought. It implies the forecaster has reason to believe 47.2 or 47.4 would be wrong.

Almost all of the time, that confidence is fiction.

The Corporate Annual Forecast: Effort Without Accuracy

Walk into any large corporation in September. The finance team is deep into what I call the "annual forecast ritual." Spreadsheets are built. Assumptions are debated. Division heads propose revenue targets. Committees meet. The numbers are refined, adjusted, aligned with strategic goals. The process is serious. The effort is real.

By November, a polished deck emerges. Revenue targets. EBITDA projections. Growth rates by segment. Margins by product line. These numbers are approved by senior leadership and communicated to the board and shareholders with apparent confidence.

Then the year happens. Actual results come in. And the forecast is wrong — usually by significant margins, almost always in the direction of excessive optimism.

Here's what matters: the process that generated the wrong forecast continues unchanged. Next year, the same team, same methodology, same apparent rigor, produces another forecast with the same properties — systematic bias toward optimism and no acknowledgment of uncertainty.

The scandal is not the error. The scandal is that the error teaches nothing. The forecasting process is an artifact of organizational necessity, not a genuine attempt at prediction. The organization needs budgets. The budgets need numbers. And so numbers are produced, reliable or not.

The Oil Price Consensus: $150 to $34

In July 2008, with oil trading in the $130–$145 range, the consensus forecast from the oil industry was remarkably confident: oil would trade between $150 and $200 per barrel for the remainder of the year.

The reasoning was straightforward. Demand was strong. Supply growth was constrained. Long-term fundamentals supported high prices. The forecast looked reasonable given the data available in July.

Then August and September arrived. The global financial crisis deepened. Credit froze. Companies cut spending. Demand collapsed. OPEC maintained supply. By December, oil was at $34 per barrel.

Think about what that means. The consensus forecast missed by more than 300%. Not 30%. Not 300 basis points. More than 300%. The range of $150–$200 proved to be three to six times higher than the actual outcome.

And notice something else: the consensus forecast included no error bars. It was not "we believe oil will be between $150–$200, with a 20% chance it could be as low as $100 or as high as $250." It was a clean range with no acknowledgment of deeper uncertainty.

The Problem: Numbers Masquerading as Information

When you separate the number from the uncertainty, you create a false sense of precision. A statement like "oil will be $150–$200" implies the forecasters have thought through this domain carefully and arrived at a bounded range. It implies that prices outside this range would surprise them. It anchors listeners to the center of the range.

The unspoken reality is usually something like: "We have no idea what will happen. Oil could go to $50 if the world enters recession. Oil could spike to $300 if geopolitical crisis disrupts supply. Most likely, between $100 and $200, but all of these are guesses."

That is a very different statement. It is honest about uncertainty. It is also politically useless for a corporation or a government or an industry group trying to communicate confidence.

And so the precise-sounding forecast — the one without error bars — is what gets produced and circulated.

What the Market Does With Worthless Numbers

Here's where it gets dangerous. Once a forecast is published, it acquires an institutional life independent of its accuracy.

Traders position based on it. Companies build capital plans around it. Policy makers rely on it. Investors adjust their expectations based on it. The consensus forecast becomes a self-fulfilling prophecy not because it is accurate but because so many actors are positioning based on it.

This is fine when the world behaves roughly as expected. But when a Black Swan arrives — when the actual generator of events produces something outside the consensus range — the actors who were positioned based on the worthless forecast experience catastrophic surprise.

They positioned for oil between $150 and $200. Oil went to $34. The capital that was allocated expecting returns at $150+ oil is now negative. The positions that were hedged for a different scenario are now exposed. The markets move violently as everyone tries to reposition at once.

The forecast was wrong. But more importantly, the forecast had no uncertainty bands, so there was no indication that the $34 scenario was possible. The market was not positioned to absorb it.

How This Pattern Repeats

The oil consensus is not anomalous. Similar patterns appear across every domain where point forecasts are produced:

In each case, the forecasting process continues unchanged. In each case, the next year's forecast is produced with the same methodology and the same confidence despite the previous year's failure.

The Pattern: Artifacts of Demand, Not Accuracy

This persistence is the tell that point forecasts serve an organizational function independent of their accuracy. They are not produced because they work. They are produced because organizations need numbers, and the pressure to provide confident numbers outweighs the pressure to acknowledge uncertainty.

A CFO who says "we will meet our targets" gets promoted. A CFO who says "we have no idea, there's a 30% chance we miss by more than 20%" is seen as uninformed and uncommitted.

And so the industry of confident, imprecise, systematically wrong point forecasts persists.

What You Should Do Instead

If point forecasts are worthless, what replaces them?

Ask for ranges and confidence intervals. If someone says "oil will be $85," ask "what range contains the true price 90% of the time?" The answer will be far wider than the point estimate, and that width is closer to the truth.

Specify in advance what would falsify the forecast. If the forecast cannot be wrong, it is unfalsifiable and therefore not knowledge. A useful forecast should have clear conditions under which you would reject it.

Focus on the tails, not the center. The consensus forecast for the middle is worse than useless — it anchors you to a false center. But exploring the ways things can go catastrophically wrong is valuable. What scenarios are the consensus forecasters missing?

Design strategies that work across multiple scenarios. Rather than optimizing for a single predicted path, build positions that survive and profit across a range of plausible futures.

A point forecast without an error bar is not information. It is confidence dressed as precision. Treat it as such.