How We Bring Data into Decisions
Using computers to make smarter decisions
If you’re any class of thinkboi, you’ve likely read Ray Dalio’s ‘Principles.’ For those of you who haven’t yet made the (progressive) transition of living with your head completely up your own arse, let me provide a brief background on Dalio.
Dalio is probably the most famous investment manager in the world. He founded the world's largest hedge fund, Bridgewater Associates. They are the world's largest hedge fund because they repeatedly deliver above market returns to investors.
In his book ‘Principles’, Dalio outlines the systems, processes, and principles that he employs both at Bridgestone and in his personal life. The principles that he attributes to his success.
Computers and Humans, working in Harmony
The underlying narrative throughout principles is Dalio’s early and aggressive adoption of computing as a way to inform decisions. Dalio argues that computers, being completely rational, and able to process far larger data sets than humans, are an indispensable tool in the investor's arsenal.
Dalio, from the very early stages at Bridgestone, employed computers to analyse data points he thought were important, weigh the impact of those data points, and, in a simplistic sense, spit out a risk-reward profile for an investment class. An example of this would be when deciding to buy corn futures. Dalio would get his computers to analyse weather patterns, make predictions for droughts, take market data of supply and demand, and also factor risks for war and other rare events. Over time, these systems would be updated, refined and tweaked, providing Dalio with a huge competitive advantage.
Hedge funds v Venture Capital
Hedge funds are inherently very different beasts to Venture Capital funds. Hedge funds make bets on much more established asset classes, they diversify their holding to offer investors risk profiles that suit them. Hedge funds look for consistent market-beating returns, they are diversification over specialisation.
Although the aim of Venture Capital is broadly similar to hedge funds (to make money) the approach is completely different. Venture Capital is a much higher risk asset class, you are betting on early-stage companies, often before they’ve even proven they can deliver revenue. You are making a bet on an idea, a team and the timing more than any ‘fundamentals’ such as P/E ratios.
While the hedge fund model involves diversifying in safer assets, Venture involves specialising in much riskier classes. The model factors in the high failure rate of startups. Most Venture Capital firms operate on the premise that of 10 investments, a few will fail, a few will break even or be mildly successful, and one will be so wildly successful it will cancel out the losses of the others (and then some).
Data and Venture Capital
Some would argue that data doesn’t really have much place in Venture Capital. What matters is getting to know the founding team, understanding the vision, understanding how much specific knowledge they hold and their competitive advantage, and then making an intelligent bet, based on the macro-factors and trends that you see driving their product. The data that exists in the public domain to make bets on established docs just doesn’t exist for Venture Capital.
To an extent, we agree with this viewpoint. It’s true that the team and the timing matter far more than any underlying fundamentals. Most successes at Y-Combinator resulted from a complete product pivot — which raises the interesting question of whether the idea that a founding team bring to the table even matters.
Having said this, whether we like to think so or not, there are still a number of factors that come into the decision-making process of backing an early-stage company, each of these factors have certain weightings, and the result of these weightings should influence our decisions.
What if we could codify these factors? Using computers to provide a rational output to support our decision-making process?
Why Startups Succeed
A few years ago, Bill Gross and his team did an analysis of a number of startups, applying a methodical rating system to each startup at its inception, based on a number of factors.
Bill and his team arrived at the below weightings. With timing being the most important element, and funding being the least.
We like this model. It’s lightweight, it doesn’t fall into the fallacy of ‘more data points = better decisions’ it focuses on a few key metrics, that can be measured (albeing often subjectively) by the Lithium team. Here’s how we’ll be using it.
We've built a lightweight Framework that uses Bill Gross’s findings as a kicking off point for decision-making.
The process works as follows:
- After the initial project interview process, at least three of the Lithium team will fill out the scorecard above for the project.
- The scorecard will then spit out an overall score, based on the weighting and the inputs. As a general guideline, an average score of 12 out of 15.6 shouldn’t cause major concerns.
- The scorecard will also identify any potential red flags. We added a ‘Legal Barriers’ section after the AGV launch, which was cancelled due to the app store legal issues they experienced.
- A decision will then be made among the decision makers, the scorecard will just be a single data point, aimed at informing the decision process, not leading it.
- Every 6 months we will do a retrospective on all projects, and update the metrics and weighting if we feel this is required. We may see that projects with strong teams performed particularly well, and up the weighting on that.
As well as helping us create a more robust decision-making framework, this scorecard approach could be a valuable resource tool should we wish to pivot to a DAO model. For example, we could show the results of the scorecards on the project selection page, and then allow token holders to vote on whether they want to launch the project.
Tom, Team Lithium x