Week 24: Two and Twenty
Decisions, decisions. No, literally: decisions.
Housekeeping
Welcome to the FIRST week of Two and Twenty. Killer name, I know. We’re here to fill the void that exists between emerging fund managers.
Our new domain is 2and20.vc. This newsletter is getting more dialed in as time goes on. The Slack has been nuked — way too complex & distracting. Instead, I’ve added a “chat” feature here so I can send news & shitposts I see on X.
As always, please hit me back with feedback and comments—I’m constantly seeking ways to make this newsletter a more valuable read.
Diving right in and keeping things brief:
On my radar:
Version two of the Distilled Intelligence website is live – would love your feedback. Version three will be coming soon.
We’re hiring a Distilled Intelligence Intern to help me with finding awesome companies to attend. We’re paying $20-ish an hour, referrals would be much appreciated. You can share this application.
DI is going to pivot a little bit. Rather than have the event be a pitch competition, we’ve decided to widen the aperture and make it more of an ecosystem summit. It’ll allow us to have more founders who are there to broadly meet investors and learn from each other, rather than making it a competitive event to raise money.
Apply to Distilled Intelligence here. Be sure to mention that NNR sent you!
Whoop 5.0 tomorrow at 10am ET. LFG.
The Main Idea
Decision Trees
This week’s topic: decision trees. I totally tuned this out while sitting in 311 at Marshall… turns out it was actually important info, so I’ve spent some time re-learning it. Hopefully my little explainer will serve as a good refresher, or will teach you from scratch so you can take advantage of this very useful tool.
Context
Venture capital involves making decisions under conditions of uncertainty. In most cases, you’re operating with incomplete data. You only have the current snapshot, not the full trajectory. Still, you have to deploy capital, pass on opportunities, choose which founders to support, and determine where to allocate your next dollar.
Many investors lean on intuition or pattern recognition when facing complex decisions. But there’s a practical way to add structure and clarity: decision trees. They help break a decision into parts, quantify the possible outcomes, and surface the option with the strongest expected value.
A little intro to decision trees
A decision tree is a tool used to analyze decisions that involve risk or uncertainty. It looks like a branching diagram—like an upside-down tree—where each branch represents a possible path your decision could take. You start with a specific decision (like whether or not to invest in a startup). Each action leads to uncertain outcomes. You estimate the chance of each outcome happening, and you assign a financial value to each outcome. Then, using a simple formula, you calculate which decision gives you the best expected result.
There are three building blocks used to create a decision tree:
A square represents a point where you have to make a decision. This is called a “decision node.”
A circle represents a point where something happens outside of your control. This is called a “chance node.”
A triangle marks the final outcome at the end of a branch. This is called a “terminal node,” and it shows the payoff (how much money you’ll make or lose).
The basic math behind decision trees uses a formula called expected monetary value (EMV). This tells you the average amount you’d expect to win or lose if you could repeat the same decision many times under the same conditions.
The formula is super simple, but kind of hard to explain…
EMV = (Probability of Outcome A × Payoff A) + (Probability of Outcome B × Payoff B) + …
You’ll see what I mean by this as we dive in further. You’re just multiplying the dollar amount yield (either positive or negative) of a decision, by the chance that the outcome actually happens. You then add them all together for each note.
Once you’ve assigned EMVs to all possible outcomes, you work backward through the tree to see which initial decision leads to the best expected result. This process is called rollback analysis.
Building decision trees
The best way to explain how to build a decision tree is with an example. In this case, our example will be deciding whether or not to participate in a follow-on round for one of your seed investments.
Step 1: Draw a square to represent the decision you need to make. This is where your tree begins. In our example, the decision is:
Should you invest $1.5 million to maintain your pro-rata in a company raising a Series A?
This square is the decision node. From it, draw two branches, one for each action you could take.
Step 2: List out each action you’re considering. In this example, you have two options:
Option 1: Invest the $1.5M to maintain your 7% ownership.
Option 2: Do not invest, and accept dilution to 3.5%.
Label these clearly as you branch out from the square.
Step 3: Now think about what might happen next under each action. These are events you can’t control. For example, if you invest, the company might:
Exit at $500 million
Exit at $100 million
Fail completely and return nothing
Draw a circle at the end of each action branch. From each circle, draw three branches, one for each outcome. These are your chance events.
Repeat this for the “Do not invest” action. The outcomes might be the same, but the payoffs will be different because your ownership stake will be lower.
Step 4: For each outcome under each chance node, assign a probability. These should reflect your best guess at how likely each scenario is. Your probabilities must add up to 100% at each chance node.
For example, let’s say you believe the following:
If you invest:
40% chance of a $500M exit
30% chance of a $100M exit
30% chance of failure
If you don’t invest:
20% chance of a $500M exit
50% chance of a $100M exit
30% chance of failure
Write these probabilities next to the corresponding outcome branches.
Important: You don’t need perfect accuracy here. The point is not to predict the future, it’s to force yourself to clearly state what you believe will happen. This process makes your assumptions explicit and gives you a way to test different scenarios later.
Step 5: Now assign a payoff value to each outcome. This is how much money you (or your fund) would make under each result.
Let’s walk through this for both options.
If you invest and maintain 7%:
At a $500M exit: 7% × $500M = $35M
At a $100M exit: 7% × $100M = $7M
At a failure: $0
If you don’t invest and get diluted to 3.5%:
At a $500M exit: 3.5% × $500M = $17.5M
At a $100M exit: 3.5% × $100M = $3.5M
At a failure: $0
Write these next to the terminal nodes (the triangles at the end of each path).
Step 6: Now perform the actual calculation to determine the expected value of each choice.
For the “Invest” option:
(40% × $35M) + (30% × $7M) + (30% × $0)
= $14M + $2.1M + $0
EMV = $16.1M
For the “Don’t Invest” option:
(20% × $17.5M) + (40% × $3.5M) + (30% × $0)
= $3.5M + $1.4M + $0
EMV = $4.9M
Step 7: Now you have a quantitative comparison. The final step is to compare the EMVs of your options and choose the one with the highest expected return. Based purely on expected value, investing is the better option.
But don’t stop there—consider other factors, like how much capital you have left in reserves, how this fits into your overall portfolio, or whether new information might change your assumptions. Still, this method gives you a clear baseline for decision-making.
By the end, your decision tree should look something like this (please excuse my messy handwriting):
Factoring in additional info
Once you’ve built your initial tree, you may get new data that changes how you see the future. This is where posterior probabilities and sample information become useful.
A posterior probability is an updated probability estimate that incorporates new evidence. Suppose your original guess was that the company had a 40% chance of having a big exit if you invested. After seeing a huge customer onboard and a new round led by a top-tier firm, you might revise that estimate to 60%. That new number is your posterior.
Sample information is the evidence that helped you update that belief. In venture, this includes customer interviews, product metrics, investor interest, or changes in the competitive landscape. The more relevant and high-quality the information, the more reliable your updated tree becomes.
You can plug these posterior probabilities into your original tree and re-run the EMV calculations to reflect the new reality.
Where they’re most helpful
Follow-On Investments: This is the cleanest use case. You can compare the expected returns of exercising pro-rata vs. walking away. Here, decision trees will account for dilution, stage of company, outcome probabilities, and payoff size. It makes follow-on allocation more rational. Filing a decision tree in your CYA folder or investment memo can be really helpful to give LPs insight into why you re-upped if the company failed.
Initial Investments: Early-stage deals are highly uncertain and driven by long-tail outcomes. A decision tree helps you map the risk/reward balance. Even if the chance of success is small, if the upside is large enough, the EMV might justify the investment.
Portfolio Construction: You can build a model of your whole fund using trees for each company. This helps you understand your expected fund-level DPI, identify whether you’re overexposed to one type of outcome, and adjust capital reserves accordingly.
Exit Timing and Liquidity: Trees can model different exit timelines and sizes, showing how choices today impact future distributions and fund pacing.
Pre-Mortem Analysis: Use the tree in reverse: what paths could lead to failure? You can model the probability of each and use that to proactively manage risk.
Dank Tweets
“Red Bull just launched a €200M venture fund. Red Bull Ventures will focus on early-stage investments in sports tech, sustainability, media, and human performance. The fund marks a strategic shift as the brand moves deeper into innovation and venture investing — with plans to back founders who share its edge.” Fitt Insider
“NewLimit, a biotech startup co-founded by Coinbase CEO Brian Armstrong, Blake Byers, and stem cell expert Jacob Kimmel, has raised $130 million in a Series B round led by Kleiner Perkins to advance its mission of developing age-reversing treatments through epigenetic reprogramming.” - from TradedVC
Some cool stuff on my radar
Here’s this week’s pocket dump. It’s been rainy in NYC:
Magnesium Taurate has changed my life:
Got my first order of Create gummies today. They arrived less than perfect, so haven’t tried them, but 5-star customer support!
Copped this sick Stetson hat. Commence summer.
Craighill made a cool sunglass lanyard. Still dorky, but cool.
The M4 Pro is basically the same as the M1 Pro. I learned that the hard way.
My second favorite pencil is on sale.
Pelican made a carry on bag and it looks SICK.
Closing
Thanks for taking time out of your Wednesday to read.
As always, you can find me on X and LinkedIn, and I’d love to hear from you via email. Whether it’s talking startups or just shooting the shit, I’m always happy to connect.
Onto the next!
//Eli
The new name 🔥