Honest takes · 9 min · 2026-06-14

The most valuable strategic decision I made this year was not something I built. It was something I refused to build. For a few weeks earlier this year I was convinced that skin-backed lending was the obvious Phase 2 for fy_nance, and I am genuinely glad I talked myself out of it before I wrote a line of code. This is the long version of that reasoning, written down honestly so that future me (and anyone weighing the same idea) has the full argument in one place.

The temptation was real

I already have the hard part. fy_nance is a CS2 valuation and tax tool, and the engine at the center of it answers one expensive question well: what is this inventory actually worth, right now, if you had to turn it into money. Once you can price an inventory with confidence, the adjacent business almost suggests itself. Lend cash against a player's inventory, take the skins as collateral, charge interest, and unwind the position if the borrower repays or liquidate it if they do not.

The demand is not hypothetical either. Traders are chronically illiquid in exactly the way that makes lending attractive. They hold five and six figure inventories and they want cash without selling, because selling means eating fees, missing the next move, or realizing a gain they would rather defer. "Liquidity without selling" is the oldest pitch in finance, and here was a market full of people who clearly wanted it. The total addressable market looked big, the demand looked obvious, and I already owned the pricing primitive everyone else would have to build from scratch.

So I took it seriously. And the more seriously I took it, the more clearly I saw that the version I was excited about, custodial lending where I hold the collateral myself, is a trap.

Why custodial lending does not work

Custodial means I take the skins as collateral and hold them while the loan is outstanding. That is the version that feels like a real lending business, and it is the version I am saying no to. Here is the actual reasoning, not the hand-wavy version.

1. The collateral cannot be perfected or liquidated on a lender's timeline

This is the one that ends the conversation by itself. A lender's entire model rests on one assumption: if the borrower defaults, I can seize the collateral and sell it fast enough to make myself whole. Skins break that assumption at the protocol level.

Steam imposes trade holds, the 7-day trade protection window tied to Steam Guard, on items that move between accounts. That means the moment I would most need to move collateral, right after taking possession or right when liquidating it, I am frozen for days. Markets move a lot in seven days. Worse, the items can be frozen for other reasons, the account can be VAC banned or locked, and Valve can revoke items or reverse trades under the Steam Subscriber Agreement essentially whenever it wants. I cannot reliably seize a skin, and I cannot reliably sell one on a deadline. Collateral I cannot perfect and cannot liquidate on demand is not collateral. It is a hope.

2. "Possession" of a skin is really control of someone else's account

There is no clean way to hold a skin the way you hold a stock certificate or a car title. A skin lives inside a Steam account, and the terms of that account belong to Valve, not to me and not to the borrower. So what I would actually be taking as security is not an item, it is control of an account, governed by an agreement I am not a party to and that explicitly reserves Valve's right to change the rules.

You do not own the collateral, you do not control the platform it lives on, and the platform owner can rewrite the terms unilaterally. That is not a security interest a real lender can rely on.

No serious lender builds on top of an asset whose custody is defined and revocable by a third party who owes them nothing.

3. The regulatory surface is a minefield for a solo founder

Lending money is regulated, and heavily. Depending on structure and state, that can mean lending licenses, usury caps, disclosure requirements, and consumer-protection rules. Now layer on top of that the fact that my collateral is a volatile, semi-liquid game item, and that I am a solo US founder without a compliance department.

The failure modes are not gentle. One forced liquidation that a borrower disputes, one complaint shaped the way a consumer-protection regulator likes to see complaints, and I am defending the entire company over a loan book that was never the core business. The asymmetry is brutal: the upside is interest margin, and the downside is existential legal liability. I am not going to bet fy_nance on getting lending compliance exactly right while also building a valuation product.

4. Custody makes me a honeypot and a fiduciary

The day I hold other people's five and six figure inventories, two things become true at once. I become a target, because a concentrated pile of high-value, transferable items is exactly what attackers go after. And I become a fiduciary, responsible for assets that are not mine, with the obligations and the unbounded downside that come with that role. A single breach, a single compromised account, a single mistaken transfer, and I have lost customer property I can never make whole. The downside has no ceiling, and no amount of interest income justifies sitting on a liability like that.

The hard line

So let me write it plainly, because the value of this decision is in how unambiguous it is.

Custodial skin lending is a NO-GO for fy_nance. Not "not yet," not "after we raise," not "once we are bigger." It is a deliberate, considered no. Every blocker above is structural, not a matter of effort or timing. Steam's terms do not loosen because I grew. Lending regulation does not exempt me because my engine is good. Holding the bag does not get safer at scale, it gets more dangerous.

What I will build instead

Saying no to a business is only useful if it sharpens what I say yes to. The skins lending exercise did exactly that, because it surfaced the part of the opportunity that is genuinely mine to take.

I am not the lender. I am the valuation and underwriting layer the lender plugs into.

A licensed lender who wants to do skin-backed loans has the one thing I lack (the license, the regulatory standing, the appetite to custody and collect) and lacks the one thing I have (a defensible, fair-value, risk-aware price for any given inventory). I can be the trusted oracle in that relationship. I price the inventory, I score its liquidity and its volatility, and I hand a licensed counterparty a number and a confidence band. I never touch the collateral. I never touch the loan. I never enter a regulated lending relationship with a borrower.

That split keeps the part I am good at and that is clearly legal for me to do, and it hands off the part that is a trap. It is the rare case where the safe move and the focused move are the same move.

Why the core engine is already the right primitive

The thing that makes this credible is the principle the product was built on in the first place: realizable value, not paper value. You do not lend against a screenshot of an inventory at sticker prices. You lend against what the inventory would actually fetch, net of fees, accounting for how fast it can be sold and how much the price could swing while you sell it. That realizable number, with liquidity and volatility attached, is precisely what a lender needs to set a loan-to-value ratio. I have been building that number for tax and valuation reasons all along. Underwriting is the same primitive pointed at a different question.

The lesson I want to keep

The honest takeaway is that the flashy adjacent business was a distraction dressed as an opportunity, and the discipline to not build it is what protected the focused one. If I had chased custodial lending, I would have spent this year on Steam's edge cases, lending licenses, and security posture for assets I should never have been holding, instead of on the valuation engine that is actually my moat.

I will revisit this exactly once, and only in one shape: the non-custodial underwriting layer, and only if a licensed lender shows up with real demand and a real license. If that day comes, I plug my number into their loan and I keep my hands off the collateral. Until then, the answer is no, and writing down why is how I make sure I do not have to relitigate it the next time the idea looks shiny.