Valuation · 7 min · 2026-06-14

When I price a CS2 inventory, the single number I look at first is the float value. It is the difference between two listings that read identical on the surface and clear at prices that are anything but. I built fy_nance because most tools treat a skin as one ticker with one price, and that quietly hides real money. The float is where a lot of that money lives, so I want to walk through what the number is, how it maps to wear, and why the clean end of a band is worth paying for.

What the float value actually is

Every CS2 skin instance carries a float value, a number from 0.00 to 1.00. Lower means less wear, so a 0.01 looks crisp and a 0.90 looks beaten. It is generated once when the item is created and never changes, which is why I treat it as a permanent property of that specific item, not a market mood that swings week to week.

Valve buckets that continuous number into five named wear tiers. The tier is what most marketplaces show in big letters, and it maps to fixed float bands:

The tier label is a useful shorthand, but it is a lossy one. Field-Tested alone spans 0.15 to 0.38, a wide stretch, and two items wearing that same label can look and price very differently.

Float still matters inside a tier

Here is the part the tier label hides. Within a single wear tier, the float keeps moving the visible condition, and buyers pay for the clean end.

Take a Field-Tested skin. A 0.16 sits right against the Minimal Wear edge and reads almost clean, with scuffing you have to look for. A 0.37 sits at the far end of the same band, one tick from dropping into Well-Worn, and the wear is obvious. Same label, different item, different price. The market knows this, so demand clusters hard at the low end of every band.

I price this as a premium that scales with how close an item sits to the clean edge. As a working rule, the top quartile of a float band carries roughly a 10 to 15 percent premium over the band's midpoint for desirable skins. For a premium skin where collectors actively hunt low floats, I use about 14 percent for that top quartile. That is not a rounding error. On a four-figure skin it is real money that shows up only if you account for where the float sits inside the band.

A few things make the premium bigger or smaller:

• Desirability. A skin nobody chases barely moves on float. A grail moves a lot.
• Edge effects. Floats that sit just inside a tier boundary (a low Field-Tested near 0.15, a Factory New near 0.00) attract their own bidding because they are the best version of that label.
• Liquidity. A thin market means the premium is real on paper but slow to realize, which matters for what I say next.

My Howl example

I hold an M4A4 Howl, Field-Tested, in my book, and it is the cleanest illustration I have of why the label is not the price.

Picture two Howls, both stamped Field-Tested. Mine sits at a low float near the clean edge of the band. Another seller's sits near the high end. On a marketplace filter they appear in the same Field-Tested bucket, side by side, as if interchangeable. They are not. The low-float Howl clears higher, because the buyer who wants a Howl at all is exactly the buyer who cares which Howl.

The catch is timing. The float premium is real, but it is patient money. A seller who waits for the right buyer realizes the full low-float premium. A seller who needs out this week takes the best standing bid, and that bid mostly ignores the float and pays closer to the band midpoint. So the same item has two honest prices: a patient price that captures the float, and a fast-exit price that gives most of it back.

That gap is the whole reason I price float into realizable value rather than leaving it on paper. A number that says my Howl is worth its patient price is only true if I am willing to wait. If I model my inventory as if every item clears instantly at its best-case float premium, I am lying to myself about what the book is worth on a quick exit. fy_nance separates the two so the figure I look at reflects how I would actually sell, not the most flattering listing I could screenshot.

When the pattern matters more than the float

Float is the main axis, but it is not the only one, and I want to be precise rather than oversell it. For a class of skins the pattern index, the seed that decides how the texture lands, matters more than the float.

The clearest cases are blue-gem Case Hardened skins, where specific pattern seeds produce a large blue face and trade at multiples of an ordinary one, and Fades, where the fade percentage (how full the gradient is) drives the price. For those, two items at the same float can be worlds apart in value, and two items at the same pattern but different floats can be close. I treat pattern as a separate axis from float, because conflating the two would misprice exactly the items where the most money is at stake. Most skins are float-driven; a meaningful minority are pattern-driven; a few are both.

How fy_nance surfaces this

I wanted the float to be visible everywhere I make a decision, not buried in a detail page, so it shows up in three places:

• A float strip on every tile. Each item shows its exact float value and where it sits inside its band, so I can see at a glance whether something is clean-edge or bottom-of-band without opening it.
• A wear tier label. The familiar Factory New through Battle-Scarred tag stays, because it is how marketplaces filter and how buyers think, and I want my view to line up with theirs.
• A float repricing line in the realizable haircut. When fy_nance computes what an item is actually worth on exit, it adds an explicit line for the float adjustment. A clean-edge item gets credited its premium; a bottom-of-band item gets marked down. The line is itemized, so the number is not a black box. I can see exactly what wear is doing to the price.

The point of all three is the same. The tier label tells you which bucket an item is in. The float tells you what it is really worth inside that bucket, and what that wear will actually cost you when you sell. I would rather see that on every tile than discover it at the moment of exit.