Why venture capital should be treated as a science not as an art

According to common business wisdom, every great success story starts with a why. It therefore seems fitting for our first publication to explore the question of why we founded MorphAIs. Luckily, the answer is pretty straight forward: Like many others, we are convinced that the current model of venture capital is broken. Let us tell you why and how we’re going to fix it.

‍

The brokenness of the asset class is probably most evident when one looks at the shockingly low long-term performance of VC funds across all phases. According to a study by Kauffman Fellows, 84% of surveyed VC firms failed to exceed returns expected by investors with only 5% being able to achieve more than 3x return rates. Without going into too much detail here, it becomes obvious that the industry undeniably has an accuracy problem. While there are certainly a variety of drivers behind this issue, three in particular stand out: Lack of innovation, gut-driven decision making and biases. Let’s start with lack of innovation.

‍

For the fact that VCs have significantly influenced the technological developments of the past years by backing technology-driven business models, it is astonishing how technology averse most VCs are when it comes to their own operations. Just take deal sourcing and deal screening as an example. To this day, both processes are largely based on manual labor, which on the one hand makes them incredibly time-consuming and on the other not scalable, as the deployable man-hours always act as the limiting factor. Technology aversion is however not only affecting the efficiency of internal processes but also the judgement of investment decisions from a risk management perspective. This brings us to the other two drivers of poor investment decisions: Gut driven decision making and biases.

‍

Weak data basis reinforces herding and FoMo effects

Since there is often insufficient data to base decisions on, VCs all too often rely on their “experience”, not to say their gut feeling when picking a company to invest in. Repeatedly you hear sentences from VCs like “Over time, you will develop a strong feeling about who is a good founder and who isn’t” or “I’d go skiing with the team to find out whether we are a good match”. Especially at the earlier stage, VCs might therefore favor those entrepreneurs who resemble themselves most in terms of educational, professional, and also ethnic background. It doesn’t take a behavioral scientist to figure out that a decision-making process based on such unquantifiable metrics is prone to all sorts of biases that ultimately lead to negative financial consequences as a growing body of research shows. And, coming back to efficiency once more, a business model based on ski trips with founders is hardly a scalable one.

‍

The big elephant in the room seems to be that for decades, unlike any other asset classes that are exposed to investment risks, VCs seemed to have gotten away with managing theirs based on intuition and experience rather than mathematical models i.e., data. Given the vast amount of money that is invested into start-ups via VCs every year (in 2020, €37bn were deployed in Europe alone, a 17% increase from 2019), it seems astonishing how little is actually talked about mitigating downside risk. Just for the sake of the argument imagine an insurance company operating on similar metrics. The outcome would most likely be disastrous. For some reason however, VCs were able to justify their approach with merely telling their LPs it’s an inherently risky asset class — as if there is nothing one can do about it.

‍

How we outperform traditional VCs by 11 times

Having said that, we do however believe that the shortcomings of the current venture capital model can be overcome with technology that combines human experience with machine precision and scalability. Using AI technology in particular, hundreds of thousands of data points and traction signals can be monitored to provide a clearer picture of which companies to invest in and which ones to rather pass up. The underlying potential is game-changing: De-biasing investment decisions through technology can make venture capital a significantly more capital-efficient asset class, while also increasing decision transparency to LPs. Through a data-driven understanding of patterns of success and failure, we can reduce the amount of capital being misallocated from prematurely rushing into investments because of herding or FoMo effects. At the same time, a technology-driven investment approach has the potential to open up the asset class to a larger pool of profitable investment opportunities, i.e. talent, that was previously overlooked due to biases influenced by gender or similarity to the investor. Undoubtably, those who further reject to harness the power of such technology will soon fall behind.

This conviction was the starting point for the development of our AI-driven technology that allows us to predict the probability of success for early start-ups. After a year of intensive R&D the result is a highly accurate predictive technology with which we were able to achieve 16x return rates in investment simulations, outperforming the average early-stage fund by 11x. By treating the VC investment process as a science rather than an art, our technology introduces three components that the current VC model needs most: efficiency, accuracy and scalability.

‍

Considering data-generated investment opportunities, the current model of venture capital is surprisingly inefficient. According to a study published in HBR, the average VC spends almost a third of his or her time on sourcing and screening potential deals. In contrast, our technology lets us handle hundred thousand of data points within seconds and preselects only the most promising companies for our investment team to screen. This way we can obtain a larger pool of pre-screened investment opportunities in significantly less time than traditional VCs. The accuracy is ensured thanks to training runs on millions of data points that identified the relevant success factors. The resulting predictive power manifests itself both in model tests and in the real world, where high-scoring firms consistently receive higher evaluations or achieve greater funding success.

‍

“VC doesn’t scale” is a long-held belief amongst observers and members of the VC space. However, this is only the case if success is based on a particularly capable partner or team, i.e. if it is human-centered. If one introduces technology to the equation, it becomes evident that certain aspects of the investment value chain can very well be scaled. With technology at the core of the sourcing and screening processes for instance, deployable man hours are no longer a limiting factor, and the freed-up time can be instead be spent more productively on value-adding post-investment activities.

‍

Tech in venture capital is here to stay

The developments of the recent years have already given us a taste of the emerging role tech and data-driven approaches will play for VCs in the coming future. To a certain extent, one can draw parallels here to the mid-1980s, when the mathematical and statistical investment approaches of the emerging quant funds were met with skepticism by more traditional investors. Today, this then novel approach has taken over the entire industry with nearly all trades being quantified and automated to some degree. Venture capital most likely won’t follow that same path as it certainly is a different kind of asset class, one that at its core is centered around humans. This said, we can observe a flickering light on the horizon as the understanding for the need of innovation and data-driven decision-making processes seems to be growing, at least slowly, among VCs. After all, an industry that is so fundamental to the innovative capacity of our society should stop clinging so rigidly to the status quo. It is long overdue that the sector seizes the opportunities a more tech driven investment approach has to offer, benefiting everyone involved.