Observed math patterns throughout life
Since I was a kid, I was always super fond of math and numbers. Unlike many other school subjects such as biology or chemistry in which the concepts are usually too hard for a 14 year old to fully understand and deduct, math has a logical approach to it at which I could grasp much better. What is super cool about it is that you can find impressive yet super easy to explain patterns with it. I wanted to raise 2 of my main discoveries, which are obviously not ground breaking but I found them by independent observation and testing and are both super cool concepts:
Power difference property
When a number is powered to another number, it generates a sequence with non-linear gaps between these numbers; for example, if we use xˆ2: 2ˆ2 = 4, 3ˆ2 = 9, 4ˆ2 = 16, 5ˆ2 = 25; from here, we can observe: 9 - 4 = 5; 16 - 9 = 7; 25 - 19 = 9. Notice that the difference grows 2 by 2: 7-5 = 2, 9-7 = 2. This property is true to every single subsequent number of this xˆ2 sequence. What is most impressive about this, is that this rule applies to other powers as well. Let's use xˆ5 for our next example. I've built a table to view such sequence:
This is a simple demonstration of the Method of Finite Differences and a simple starting point to think about derivation; and notice that the final constant that was found is the power factorial: 5! = 120; 2! = 2. This is a simple, pattern recognized property that I noticed at some point in high school and it stuck with me until I learned more about it after raising it to a calculus professor (shoutout to Wagner!).
9 difference property
Pick a natural number; let's say you choose 8231 for this example. If we read it backwards, 1328 and then we subtract them we get 6903; notice that if, we sum up 6+9+0+3, we get 18? Does this property ring a bell to you? It demonstrates that this number is divisible by 9, and this property is true for any number when subtracted to it's backwards pair. Ok, cool; now pick a new number, scramble it, subtract it from itself and notice that it generates the same divisible by 9 property. For the last trick, pick a new number, sum up it's digits and then find a new number which has the same sum; let's use an absurd example: 890312, when all it's digits are summed, it returns 23; now, let's take 111.111.111.111.111.111.111.11 (yes,there are 23 1's here) and let's subtract them and then divide the result by 9: 1.234.567.901.234.567.802.311 is the final result, a no-remainder division. This was also a result I found by simply twitching around numbers in 2024, but it turns out to be a much simpler concept when compared to the first example.
What even is the purpose of all of this? How can I connect this with VC?
I wanted to raise a point that things that are so natural and trivial such as math present patterns which can extrapolate to multiple examples, and noticing them is crutial to simplify the process of understanding concepts and noticing other underlying patterns; notice that the Power difference example has the first property of finite difference, which then has an underlying property related to it which is the factorial result; the 9 difference also has a starting point which then tracks to the equal sum that is divisible by 9 property. This skill of pattern recognition is super useful to prevent mistakes (which, when similar to other mistakes, have a similar taxonomy and shape) but to also assert opportunities, and this is where VC kicks in as, potentially, the only asset class at which past performance of a fund has positive (perhaps, if you want to be optimistic, super positive) correlation to futures hits, given that this is a business much powered by sense and patterns recognized in previous hits.
Why is that?
At the end of the day, VC is, more or less, a game of combinatorial analysis with emotions and probabilities plugged into it, and that is where patterns kick in, specially thinking about tangible variables such as Founders background. Some patterns that are looked out for in founders, primarily are, for example: target schools, target previous companies at which the founders worked at a specific time window, not-first time founders,well-connected founders (introduction through network), backed by a specific set of angel investors founders, already Venture backed funds, especially high goodwill funds, etc. The variables here are basically infinite. We can, from here, analyze a VC fund's portfolio and try to find these patterns, and based on this and create a slightly overfitted equation to check for future founders which can be sourced to them. I'll take an example that is kind of easy to understand: Enter's founder and CEO, Mateus, is a GSB MBA dropout; David, co-founder and CEO of Nubank, is a GSB MBA; Marcos, co-founder of Mercado Libre is a GSB MBA; one apart from the other, there is a 13 year gap. To cohoborate with this, all of them attended Ivy league colleges at some point: Mateus is Master in Law from Harvard; David graduated from Stanford; Marcos graduated from Wharton. Going further, all of them were part of (arguably) challenging work enviroments; Mateus worked in product at a VC backed startup, Quansa; David worked in finance ranging from Goldman Sachs to Sequoia; Marcos worked in JPMorgan.
Does it mean that everyone that follows this path is going to be a successful founder? Obviously not! But does it mean that if you dont follow this path you are not going to thrive? Obviously not! But we can sum up some factors as "good indicators" for founders and from there we can abstract some kind of low risk, low volume deal sourcing. The flaw here is that this kind of sourcing for moon-shots (such as all the listed companies) are highly crowded; everyone is scouting Stanford students which dropout, get into Stealth and later create a startup. Everyone is tracking people in prestigious positions in challenging work environments.
So what?
I think the real value here is structured in 2 main actionable plans:
- Understand a VC fund's "index" or ideal profile of a founder + startup action area.
- Track the "underdog" successful founders and where they come from, so there is a higher chance of low volume, less crowded deal sourcing.
And some other points to consider:
- Looking at other not currently looked at indicators for future new founders. For example (this is just a thought): tracking academic outliers, since high school is a good way to find these deals; volume is much higher, much less crowded, but much less stable given age and extended time window.
- Enviroment surroundings are also a good way to track such potential founders; Mateus from Enter was pushed to go to Stanford because David from Nubank and Marcos from Mercado Libre encouraged him to do so. Obviously, this a reflex of the enviroment Mateus was/is inserted into.
Going further
I'm currently working on this as a side project; trying to find outlying future founders that match the index of some VC funds. It is a super scraping heavy, hard process, but when it is done, I will make another post dedicated to it.
Important disclaimer
Causality X correlation will be addressed in the future; false positives and negatives are expected and will also be addressed at some point when approaching this problem.
Sources
This was written based on 2 podcast episodes I recently watched that pushed me to think and analyze the VC and VC-backed startups with this new lens:
Mateus Costa-Ribeiro (Enter)'s first podcast episode with Filipe Lauar
Ramtin Naimi (Abstract Ventures)'s episode on Invest like the Best