Project analysis – team, TVL, narratives

Introduction

In crypto 90% of projects die in a bear market. Not because the technology is bad, but because the team disappears, TVL was borrowed for farming, and the narrative is over. Fundamental analysis in crypto is not about reading a whitepaper full of buzzwords, but about verification: whether the team ships, whether TVL is organic, whether the narrative has fuel for the next 2 quarters.

This lesson gives you a project assessment checklist, the kind fund analysts use, but simplified for solo use. Without that checklist you are a puppet of Twitter.

Team and investors – who builds and who funds

Team

What you check in 10 minutes:

A founder with 10 years at FAANG + 2 crypto cycles is not a guarantee of success, but it reduces the risk of abandonment.

Investors and backing

Not every VC is equal. Check:

• Does the VC publish an investment thesis, or does it throw money everywhere? a16z, Paradigm and Polychain have a process, but also plenty of misses.
• Was the round a token warrant or equity? A token warrant means the VC gets tokens cheaply.
• Is the VC active in governance? If so, the risk of a quick sell-off is lower.
• Is a large CEX both an investor and a market maker? A conflict of interest, it can pump the listing.

Simple team scoring:
+2 built 2+ years in crypto with delivery
+1 public, technically active
-2 anonymous + large raise without an audit
-1 pivots the narrative every quarter (first AI, then DePIN, then RWA)

TVL and on-chain metrics – is the liquidity real?

TVL (Total Value Locked) is the sum of deposits in a contract. It is easy to manipulate.

How you measure real TVL:

1. DeFiLlama – TVL vs TVL without staking (No Staking) vs Borrowed TVL. Exclude double counting.
2. The 90-day trend, not the daily one. Is TVL rising while incentives fall? If TVL drops 50% after the rewards programme ends, that was mercenary capital.
3. TVL per chain – is 90% of TVL on a single chain whose bridge the team controls? Centralisation risk.
4. The ratio of TVL to users (Active Addresses). $1B TVL and 100 users = 2 whales.
5. Fees and Revenue. The DeFiLlama Fees dashboard. A project with $500M TVL and $100 a day in fees does not have product-market fit.

TVL quality indicator:
Quality Ratio = (Fees 30d * 365) / TVL = annualised yield from fees
If Quality Ratio <0.5% -> TVL generates no fees, it is dormant or borrowed
If >10% -> High usage, but check whether this is not wash farming

Product and audits – does it work?

• Audit: Who audited it? (Trail of Bits, OtterSec, Halborn, Certik – different tiers). Is the audit public and were the critical bugs fixed? 1 audit per 1,000 lines of code is the minimum.
• Bug bounty: Is there one on Immunefi with a pool >$100k? No bounty = the team does not believe in its own code or is cutting costs.
• Technical documentation: Is it clearly written how the system works, or is there only marketing diagrams?
• Dependencies: Does the project rely on 1 oracle (e.g. Chainlink only) with no fallback? What if the oracle fails?
• Admin keys: Does the contract have a proxy with an upgrade option and no timelock? Check “Oracle vs Admin” on DeFiLlama or the docs.

Red flag: a DeFi project with $100M TVL, 1 audit from an unknown firm, and an admin key in the hands of 1 person on an EOA with no timelock.

Narratives – the fuel of the cycle

A narrative is a story that explains why a token should rise. Narratives rotate every 2-4 months: L2, Restaking, AI x Crypto, DePIN, RWA, Memecoins, Perp DEX.

How to assess a narrative:

1. Does the narrative have revenue? AI tokens with no GPU usage are a narrative only. RWA with real bond volume is a narrative + cashflow.
2. Does the narrative have a calendar catalyst? E.g. the EigenLayer TGE, the Dencun upgrade for L2s = fuel.
3. Where are we in the narrative cycle?

Narrative phases:

Tools for tracking narratives: DeFiLlama Narratives, Kaito AI (mindshare), the Dune dashboard “Narratives”.

Narrative assessment procedure:
1. Check mindshare (Kaito, Twitter) - is it rising for 14 days?
2. Check the narrative's TVL (e.g. total restaking TVL) - is it rising?
3. Check the number of projects in the narrative - if >15 in 2 months, it is overcrowded
4. Check whether the narrative leader has a catalyst in <45 days
Missing 2 of 4 points = weak narrative

Project assessment model – the scorecard

Build your own card. Example:

Weighted total >1.5 = a project for further analysis. <1.0 = reject.

This is not a valuation, only a quality filter.

When the plan says do not trade

1. When a project has $500M TVL but 70% of it is 1 address (team multisig). That is not TVL, that is a team deposit.
2. When the team is anonymous, with no audit, and asks for a deposit with no timelock. That is a casino, not DeFi.
3. When the narrative is overcrowded and every new fork rises 3x in 2 days. You are buying the peak of attention.
4. When a project changes tokenomics a week before the TGE (increases team allocation). No respect for the community = a future dump.
5. When you cannot explain in 2 sentences what the project does and where its revenue comes from. If you do not understand it, you have no edge.

The mistakes that cost the most

1. Trusting TVL from an aggregator without checking double counting. DeFiLlama has a “Double Count” toggle – many projects count the same dollar twice (e.g. LST + LP).
2. Ignoring admin keys. “The project has an audit, so it is safe” – an audit does not protect you from a malicious admin who drains the funds through an upgrade.
3. Chasing a narrative without checking the leaders. You buy 5 copies of the leader while the leader is already falling. Capital rotates to the leader first, then to the copies, then back to the leader.
4. No update to the score. You rated a project 1.8 three months ago and you hold. In the meantime the team changed, TVL fell 70%, and you did not update.
5. Overrating VC backing. VCs are wrong on 80% of projects. Backing is not a price guarantee, only a guarantee that the project had money to start.

Measurable exercise

A task on 3 projects from one narrative (e.g. Restaking):

Pick 3 projects (e.g. Ether.fi, Renzo, Puffer). For each one, fill in the scorecard and the table:

Then track their performance vs ETH for 30 days and compare it with your score. Did a higher score correlate with a smaller downside deviation during declines?

Journal entry:

Assessment date: ...
Project: ... Score: ... Thesis: "I think X is better than Y because..."
After 30 days: price of X vs ETH, price of Y vs ETH, did the thesis hold? What did I miss?

The goal is not prediction, but calibration of the quality filter.

Check yourself

1. Name 3 red flags when assessing a team.
2. Why can TVL be misleading and how do you check its quality?
3. What is the difference between Fees and Revenue in DeFiLlama?
4. Describe the 4 phases of a narrative.
5. Why do you need a timelock in a DeFi contract?

Frequently asked questions

Is an anonymous team always a scam? No. Bitcoin is anonymous. But in DeFi with large TVL, anonymity increases risk. If the team is anonymous, require at least: audits from top firms, a bug bounty, a multisig with public signers, a timelock, and more than 1 year of operation. The more anonymity, the larger the safety margin you need.

How do I check admin keys quickly? DeFiLlama has a “Hacks” and a “DeFi” tab with information about the multisig. You can also check the contract on Etherscan – the Contract -> Read as Proxy tab -> whether there is an owner and whether the owner is a Gnosis Safe with multiple signers. If the owner is an EOA (an ordinary wallet) – red flag.

Is TVL the most important metric? No. The most important thing is the combination of TVL + Fees + user retention. TVL alone with no fees is a parking lot for capital, not a business. Hype alone with no TVL is a narrative with no foundation.

How many projects should I watch at the same time? 5-10 in depth is better than 50 superficially. Rotate narratives quarterly. Your watchlist should have a score and a date for the next review.

What if a project has low TVL but a strong team? That may be an early-stage opportunity, but it requires a larger position of knowledge and a smaller position of capital. Apply the rule: early stage = smaller position, more GitHub and testnet verification.

---