← Analysis
9 min read

Your portfolio is worse than you think

Your inventory has two values. There is the number Steam shows you, and there is the amount of real money you would end up with if you sold everything and moved it to your bank account. Those two numbers used to be close. They are not close any more.

This post walks through the evidence for that, why it is happening, how far it is likely to go, and what I’m going to do because of it.

The value of Steam dollars is dropping

By treating Steam dollars and CSFloat dollars as two different currencies, you can measure the exchange rate between them. You can measure it directly by dividing the Steam price of an item by its CSFloat price. It is exactly the same as a currency conversion, as it is pricing in the Steam fee, the CSFloat seller fee and other withdrawal fees. But since those are fixed over the period we are looking at, I ignored them.

Obviously tracking a single skin or a single item would be prone to noise and volatility. So here are two different indexes: one which tracks all cases, and one which tracks a fixed basket of 673 liquid skins (the ones present in at least 90 percent of months out of the top 1,000).

How much more an item costs in Steam dollars

Above 1.0 means Steam wallet money buys less than real cash does.

Median Steam price / CSFloat price across the fixed basket. Skins end at 1.42x, cases at 1.62x — a Steam dollar is worth about 70.6 cents.

In early 2023 the rate was about 1.13, and it even dipped through parity for a moment. Today it takes 1.42 Steam dollars to buy a skin that costs 1.00 real dollars on CSFloat, which means a Steam dollar is now worth about 71 cents.

Cases are worse at 1.62. Cases are the cleanest way to move money out of Steam, because they are all identical, they are all liquid, and there is no float or pattern to argue about. Money pools where it is easiest to move, so the premium is highest there.

The trend is steady rather than sudden. Fitting a line to the skin series gives 6.7 percent per year, and it is about as statistically solid as this kind of thing gets (p below 3e-8, R squared 0.54). This means if you have been holding skins and valuing them in Steam dollars, you have been losing roughly 7% to the Steam equivalent of inflation and debasement.

I checked what the value of the debasement would be before the trade-up update. Still, you were losing roughly 5.5% a year over 2.7 years (R² 0.33, p = 5e-4). However, cases in that window are only 3.4%/yr and are not a solid fit (R² 0.12, p = 0.07).

Steam-dollar debasement before the Oct 2025 break

Log-linear fit on months before October 2025. 673-skin basket; full sample +6.7%/yr.

Skins +5.5%/yr over 2.7 years (R² 0.33, p = 5e-4). Cases +3.4%/yr is not a solid fit (R² 0.12, p = 0.07).

I checked the cases to see which ones devalued in terms of Steam dollars and which ones increased. Every single one moved the same direction; Steam dollars down. Therefore it is not a few outliers moving the index, it is actually the index shifting as a whole.

All 40 cases moved the same way

Change in the Steam premium from 2025-07 to 2025-09 versus 2026-06 onwards.

Median change +25.6%. Every case moved in the same direction, so this is the whole index shifting rather than a few outliers dragging it.

Why are CSFloat and Steam prices drifting apart?

There are a couple of main mechanisms which will affect the exchange rate between CSFloat and Steam.

And to give you a crash course in basic common sense:

As the price of Steam dollars went down, we are looking for a situation which would have either caused:

A supply side shock?

Hypothesis
During the trade-up update fiasco, there was a massive amount of deposits onto Steam causing an oversupply of Steam dollars. And a couple of months later they manifested themselves as people trying to get out of the system by cashing out via CSFloat.

Case trading volume

Same 40 cases, units sold per month on each venue.

CSFloat history starts mid-2023. July 2026 on Steam is an incomplete month and is left out rather than drawn as a fall.

Verdict
Looks like at the start of 2025 we can see stronger demand for cases on CSFloat, but not at the expense of the Steam market. This is inconclusive, but it suggests more people are moving cases through CSFloat without pulling volume off Steam.

Trade-ups?

Hypothesis
Trade-ups rely on CSFloat now and are driving CSFloat demand.

I am into trade-ups and have been tracking them for a while. Because trade-up margins have been compressed (for reasons which are interesting and I might make a post about later), trade-up outputs are best sold through CSFloat to avoid ~13% in sales tax.

Top 1000 skins, units traded

Two panels with separate scales. Steam's monthly volume is roughly twenty times CSFloat's, so a single shared axis would flatten the lower series to nothing.

February 2026, March 2026, July 2026 Steam counts are thin scrapes (fewer days than neighbouring months), not missing months. CSFloat averaged 107k sales a month before the Oct 2025 update and 449k after.

BucketSteamCSFloatExchange-rate change
Weapons (364)+160%+105%+27%
Knives and gloves (309)-40.5%-50.4%+20%

Steam and CSFloat are each bucket’s cumulative price change since October 2025. Exchange-rate change is not Steam minus CSFloat. It is how much the Steam/CSFloat ratio moved: (1 + Steam) / (1 + CSFloat) − 1.

Verdict
CSFloat is definitely seeing higher volume in skins trading volume, but this was not caused by the trade-up update. It has been a persistent rise for years now. This, along with the fact that cases (which cannot be traded up to) also gained against Steam, points elsewhere.

People are just using CSFloat and third-party markets more

I think this is the most likely cause. A multi-year venue shift is occurring. We already have evidence for this:

Share of sales happening on the cash market

Percent of all sales by count, per quarter.

Cases are the cleanest way to convert out, and they stopped migrating in 2025 — flat near 9% for five straight quarters while skins kept climbing to 22%.

Cases are the cleanest way out, so they migrated first, then stopped. CSFloat market share of case sales has sat near 9% for five straight quarters. Conversion volume has been flat since mid-2025.

What this is telling me is that people are just accepting the fact that to move money out of Steam they will pay more.

What this does to your portfolio

Skins were the best spot over the past 2 years. $1,000 of cash in January 2023, allocated three ways:

What $1,000 became

Each line starts as $1,000 of cash in Jan 2023. Skins are marked at CSFloat; the wallet is marked at the cash-out rate.

Held in skins: $2,189. Left in a Steam wallet: $796. Held as cash: $1,000.

Held in skins, that $1,000 is about $2,190 in cash today. Left in a Steam wallet it is about $800. Skin holders made money. Wallet holders lost it to the exchange rate.

If you have been valuing your items via Steam credit: I’m sorry. Right now is the largest discrepancy in history. The blue line is the total value of a broad basket of skins at Steam prices. The orange line is the same basket converted into what it would actually fetch via CSFloat.

Your portfolio, measured two ways

The shaded gap is value on the Steam price tag that is not in your bank account.

Indexed to January 2023 = 100. Steam ends at 275 against 219 in cash — a gap of 20% of the stated portfolio value.

The shaded area between them is roughly 20 percent of your stated portfolio value. And if you haven’t accounted for that yet, your loss in value over the past 2–3 months is going to look a whole lot worse.

What does the future hold?

The divide between third-party marketplaces does have ceilings and floors. You can buy games on Steam, plus Steam Machines and Steam Decks. These offer a small safety valve on a runaway Steam debasement.

However, the Steam Deck has been on open sale since 2022. If wallet to hardware to cash were a workable conversion at scale, arbitrageurs would have pinned the rate near 1.18 (assuming they resell ~85% value) for the last four years. Instead the rate climbed from 1.13 to 1.42 over exactly that period. It moved away from the supposed floor, not toward it. It is unclear where the floor is, but it has to be there. Steam hardware and games are currently the only mechanism preventing Steam credits from hyperinflating.

Steam has a strong incentive to prevent this type of arbitrage. They see almost none of the profits, and take most of the loss.

But at the same time, if Valve continues to pursue and prevent arbitrageurs from reselling Steam hardware — their only current mechanism pinning Steam credit prices — they will encourage a larger price deficit between Steam credits and third-party marketplaces.

Thus, ironically, encouraging more people to run the gauntlet of reselling hardware as the exchange rate continues to divide.

Allow hardware arbitrageurs: lose money on a product. Don’t allow them: postpone the problem and the problem gets bigger.

However, Steam holds all the cards. And if it becomes a problem they won’t let it affect their bottom line. They can:

My predictions

Bar interventions from Steam, I am not confident that the market will return to 2025-era exchange rates. The existing mechanisms (games, hardware) didn’t work previously, so they won’t work now.

But I am slightly more confident Steam dollars will continue to debase. Steam has consistently debased 5% per annum if not more. The current system just favours third-party marketplaces. They are gaining share of the market and nothing seems to be preventing them from growing. They let you withdraw, Steam doesn’t. They charge 2–5%, Steam charges 13%.

My two cents, no one else’s.

Please roast me. I’d love to hear counter opinions, especially if they come with data.


Method and caveats

Data snapshot 20260813T013830Z, covering 2023-01 to 2026-07.