The Tight Correlation Between Zodiac Signs and Sales Forecasting
Yes, Zodiac-sign forecasting is more precise than your current method.
Well, it might be time to toss the CRM, shut down the data warehouse, and hire an astrologer. Because when it comes to predicting business performance, Capricorn might be as good a forecaster as your CFO.
Why? Because all forecasts are wrong. Not some, not most—all. Nobel laureate Daniel Kahneman said as much: we live in an unpredictable world, and despite that, we still cling to forecasts like a toddler with a security blanket. Worse, we plan entire businesses around them. The irony is that the most honest—and accurate—five-year forecast would simply read: “We don’t know. But we’ll try our best.” Say that in your next board meeting and see how fast your badge stops working.
Kahneman’s research shows that algorithmic decision-making consistently outperforms human judgment. In one study, algorithmic diagnoses beat experienced doctors. Not because algorithms are brilliant—but because humans are predictably biased and lazy. Add office politics, sales incentives, and gut feelings into the mix and you get the predictive power of a broken compass.
Forecasting in Go-To-Market: The Theater of the Absurd
Anyone who’s worked at a mid-to-large company knows the five-year planning ritual. Months of spreadsheets, meetings, and slide decks—none of which resemble reality once the ink dries. And yet, sales leaders persist in assigning revenue probabilities like we’re playing Deal or No Deal: “This deal is at 70%.” Based on what? The rep made eye contact with the buyer on the Zoom call?
Even CFOs aren’t immune. Kahneman’s survey found that the correlation between CFOs’ forecasted and actual S&P 500 performance was zero. As in: they did no better than political pundits predicting elections based on lawn signs.
Still, companies continue forecasting revenue as single numbers instead of ranges, as if the universe cares about our spreadsheets. According to ChatGPT, in the last five years, roughly 50% of S&P 500 companies missed their forecast by more than ±5%. That’s basically a coin flip. So yes, you might as well outsource your forecast to a psychic—or better yet, the late Paul the Octopus, who correctly predicted 85% of World Cup match winners.
On the flip side, IBM’s Watson correctly predicted tennis match winners approximately 70% of the time using algorithmic models. Even the original electronic line-calling system in tennis didn’t detect the actual ball—it used statistical modeling to estimate where the ball would land. Let that sink in. Tennis has better forecasting than most companies.
So, What Is Algorithmic Forecasting?
It’s forecasting without gut feelings, sales bravado, or horoscopes. Instead, it uses statistically significant data, processed through a mathematical model that reflects your business dynamics. Yes, it requires math. No, it doesn’t care about your “hunch.”
Your Quick and Brutal Guide to Algorithmic Forecasting:
Determine your forecasting cycle.
Fast-moving businesses might need daily sales unit (sales/day). Selling airplanes? Try annual. The point is to pick a unit of time that reflects how often your business changes.
Choose a rolling window and normalize.
Depending on seasonality, you may want a 30/60/90-day rolling window. To normalize, use a little something called the Central Limit Theorem: take at least 30 random samples to stabilize your data. Yes, the same theorem that made you want to drop out of Stats 101.
Measure days to close.
This gives you a window of precision. If your average deal closes in 30 days, your 30-day forecast is based on actual pipeline. A 60-day forecast? You’re losing precisions, you’re estimating future pipeline.
Know your win rate (the clean version).
No, not the one where reps check boxes in Salesforce. Use this:
Win Rate = Total Sold Amount ÷ Total Quoted Amount.
It’s simple, clean, and doesn’t rely on human judgment—always a plus.
Avoid “deal aging” weighting.
It’s tempting to give more importance depending on deal age. Don’t. It complicates your model, incorporates mathematical fallacies and adds no meaningful accuracy.
Segment by vertical or cohort.
Each product line, region, or customer cohort has unique dynamics—days to close, win rates, etc. Treat them differently in your model, or prepare for wildly misleading forecasts. But don’t over do it, stick to meaningful segments with significantly different dynamics. More segments won’t necessarily equal higher precision.
Ban subjective inputs.
Sales reps assigning probabilities? No. Use a coin flip. Or a horoscope. Both are equally accurate—and faster.
Build, test, and iterate.
Use historical data to build a model and test its output against actuals. Tweak your parameters. This is where the magic (and reality) happens.
Calculate a forecast range.
Stop chasing single numbers. Here’s a sample formula, for a monthly forecast:
Average Monthly Forecast = Actual Sales + (Avg. Sales/Day × Business Days Left)
High Forecast = Average Monthly Forecast + (2 x standard deviation)
Low Forecast = Average Monthly Forecast - (2 x standard deviation)
In other words: forecast with confidence intervals. Not fairy tales.
The Closing Argument: Kill the Crystal Ball
Yes, this is dense and technical. But so is flying a plane—and you wouldn’t want the pilot using vibes and intuition, unless you are Spirit Airlines. Forecasting deserves the same rigor. It’s time to retire the “magic 8-ball” forecasts and hockey-stick graphs that magically curve up in Q4.
Imagine a world where go-to-market teams don’t set themselves up for failure every quarter. Where pipeline reviews don’t feel like improv night. Where reps don’t have to pretend they know the odds like a Vegas bookie. Imagine forecasts grounded in math, not hope.
And if that still feels like too much work, don’t worry—Leo season starts soon.