Suppose you wanted to guess the winner of this Sunday’s Super Bowl. Which of these methods do you think would work best?

(a) Flip a coin

(b) Make an educated guess

(c) Gather data and conduct an informed analysis

In a classic study, researchers Paul Slovic and Bernard Corrigan attempted to answer this question. Instead of football, they looked at horse racing, but the results are equally applicable.

In their study, Slovic and Corrigan asked expert horse racing handicappers to make predictions using varying amounts of data about the horses in a race. Some of the handicappers received just five statistics on each horse, while others received ten, twenty or forty. The study produced two interesting findings:

As handicappers received more data, they became more confident. Those who received forty statistics about a horse were far more confident in their predictions than those who received twenty. They, in turn, were more confident than those who received ten, and so forth.

Despite increasing confidence, more information didn’t necessarily lead to better results. While those with the least information ended up making the worst predictions, there appeared to be a point of diminishing returns as each handicapper received more information. In fact, beyond a certain point, more data actually resulted in worse predictive accuracy.

Taken together, these findings provide an important lesson for individual investors: When it comes to making predictions, you definitely want to gather some information, but not too much. No question, you can’t simply flip a coin. At the same time, it’s counter-productive to dig too deeply into the data. An educated guess, it turns out, has the best chance of being right.

I mention this because it’s so counter-intuitive. Especially when it comes to financial decisions, everyone wants to feel that they have “done their homework” and conducted a thorough analysis. But, as the authors of this study explain, the problem with having too much data is that we end up drowning in it. We overthink things, we become distracted by outlier cases and we generally miss the forest for the trees. And yet we don’t even realize it. We think we’re becoming more expert when, in fact, we’re just getting ourselves tangled up.

Before I conclude, I want to draw an important distinction. The findings in the horse racing study apply only to questions that involve predictions. Examples include:

“How much of my portfolio should I invest in international stocks?” (requires predicting future market returns and exchange rates)

“Should I take advantage of the Roth option in my 401(k)?” (requires predicting future income and tax law changes)

“If I have a lump sum to invest, how should I time my investments?” (requires predicting future market returns)

“When my long-term care contract renews, should I pay the premium increase or accept reduced coverage?” (requires predicting health and longevity)

“Is it safe to invest in long-term bonds?” (requires predicting interest rates)

In all of these cases, because they require some element of prediction, I believe the horse racing study applies. This is when you want to make an educated guess.

But there are lots of other cases that do not involve any element of prediction. In those cases, you definitely do not want to guess. Life insurance is a good example: Yes, your needs might evolve over time, but you don’t need to make any predictions to calculate how much insurance you should carry today. To answer a question like that, don’t guess; instead, just sharpen your pencil.