Avoiding the Traps of Overfitting and Underfitting Your Investment Thesis

Union Square Ventures (USV) is the poster child for on-thesis investments working out. Their disciplined, thematic approach has led to some of the most successful venture bets in history.

But then you’ll hear plenty of stories like this one, where a non-thesis investment ends up being a fund’s best result.

So which is right? Stick only to your thesis? Or invest opportunistically as deals come?

There’s clearly no universal answer, but the tension between on-thesis and off-thesis investing can be thought of in terms of mathematical model fitting.

So what can we learn from math?

What is Underfitting?

Underfitting occurs when a model is too simplistic—it fails to capture meaningful patterns in the data.

In investing, this is akin to having a shallow thesis that doesn’t allow for nuance or learning.

If your investing model is too broad or generic (e.g., "we invest in great companies"), it doesn’t give enough structure to make informed decisions.

You end up missing important lessons from past investments because you aren’t building a strong enough foundation to recognize recurring patterns.

The result? A scattered, unfocused portfolio that doesn’t prepare you for future opportunities.

What is Overfitting?

Overfitting is the opposite problem: a model that describes past data so precisely that it starts capturing noise instead of true signal.

Why does this happen? Measured data is never perfect—it contains randomness, inaccuracies, and external influences. If you insist on creating a model that perfectly fits every data point, you must also fit the random noise present in that data. The result is a model that does not generalize well—it works only on the given dataset but fails on new, unseen data.

The classic marketing adage applies here: 50% of marketing spend works—we just don’t know which half. Overfitted models trick us into believing we do know which half.

From a more qualitative perspective, human thinking is bound by our past experiences. We rely on it, yet it can also lead us astray. If we perfectly describe what has come before, we run the risk of explaining noise just as well as signal.

Opportunity Cost: Tradeoff between Accuracy and Generalizability

Mathematicians find best fit by using regularization techniques—balancing complexity with predictive power. But how do you know when you've overfitted? Or when additional refinement simply won't yield a commensurate return?

In investing, more thesis refinement and deeper analysis may initially improve decision-making, but eventually, the marginal benefit declines.

Like in least squares regression, adding more complexity can reduce residuals but may not translate into better future predictions, because the model may memorizing the dataset rather than learning its underlying principles.

Recognizing when more effort ceases to yield real improvements is key to avoiding paralysis and opportunity cost tradeoffs. The most effective investors strike a balance between rigor and flexibility, knowing when to stop refining and start acting.

What Does This Mean For Investors?

The best investors strike a balance between overfitting and underfitting by identifying the most critical variables that drive success.

At The Healthcare Syndicate, we “fit” the world of early-stage investing by:

1. Focusing on healthcare – A broad investment scope that is focused enough to allow for building up experiences.

2. Prioritizing founder-market fit – A variable that captures a large portion of what a startup essentially is (people tackling a market opportunity).

3. Demanding strong go-to-market strategies – Going from 0 to 1 requires a strong sense of how the company will uncover and generate market demand.

For everything else? We recognize the risk, measure it to the best of our ability or identify how to mitigate it through our network of investors and founders, and ensure that the potential reward is large enough to justify taking it.

Conclusion

Venture investing, like mathematical modeling, is about finding the right balance. Underfit your strategy, and you won’t learn from past mistakes. Overfit your strategy, and you’ll misinterpret past successes as a roadmap for future wins.

The best investors build models that generalize well while leaving room for the unexpected. In early-stage investing, that’s where the real magic happens.

Please subscribe to our newsletter if you haven’t, and share our newsletter with a friend. Stay tuned to our newsletter for more insights into healthcare innovation!

Join us at The Healthcare Syndicate as we back the most ambitious founders 10Xing the standard of healthcare!