CS2 Skin Market Report: August 2026

Here is the whole window in one view. Each item's change is its lowest listing on 22 August against its lowest listing on 31 July, across the marketplaces the feed returned (CSFloat, Skinport, CS.MONEY and market.csgo at the time).

The size of the move matters less than how uniform it was. A median of −5.6% with 47 of 56 items lower means the drift was broad: 84% of a basket that spans $0.10 cases and $11,000 rifles moved the same way in the same three weeks. When that many unrelated items fall together, the likeliest cause is something market-wide: fewer buyers, more sellers, or both. A problem with any single skin would not move the whole basket. Nine data points per item cannot tell me which, and I am not going to pretend otherwise.

Three weeks is also short. I would not call this a trend on its own. What makes it worth writing down is that it gives the next report a fixed baseline to compare against, which is exactly how the October report used it.

Every category except cases had a median fall between 4% and 7%. Cases had the smallest median move and the largest single drop, which tells you medians and averages describe cases badly.

Gloves look worst, but three items is too few to call a category result, so I read that row as "all three gloves I track fell". Pistols and rifles are the more meaningful numbers: 25 items, 22 of them lower. The Desert Eagle Blaze (Factory New) going from $652.99 to $561.12 was the largest fall of any gun in the window.

Cases need a closer look. The Kilowatt Case's −33.8% is real in the sense that the snapshot recorded it, but the whole fall arrived in the final sample: its floor sat between $0.132 and $0.142 for eight straight snapshots, then read $0.09 on 22 August. Three weeks later, on 13 September, the feed had it back at $0.125. One low listing at the end of a window can set the headline for a whole category, which is why I do not quote a case "trend" from a single endpoint. The Chroma 3 Case, at the other end of the price range, held between $2.78 and $2.89 the entire time. The CS2 case prices guide tracks the case ladder live; at the latest refresh (13 October 2026) the Kilowatt Case's lowest listing was $0.08.

Sorting the same 56 items by their 31 July price gives a cleaner picture than sorting by category:

The middle band fell hardest and most consistently: 26 of 27 items lower. The top band barely moved; the AWP Dragon Lore (Minimal Wear) went from $8,400.00 to $8,428.40 and the M4A4 Howl (Factory New) from $6,515.52 to $6,650.00. The cheap band falls in between on the median, but it contains the most extreme single readings in both directions, because a two-cent change on a 27-cent case is a 7% move.

My explanation for the middle band, which is opinion: four-figure items have few buyers but patient ones, and there is usually a standing offer not far below the cheapest listing. Mid-tier skins have many sellers who are relaxed about price, so when demand softens the floor simply drifts. I set out that reasoning, and its limits, in why CS2 skin prices drop. Treat it as an argument, not a measurement. My data shows the ordering, not the cause.

Direction is half of a market report; the other half is how bumpy the path was. For each item I took the gap between its highest and lowest snapshot inside the window, then the median of that by category:

Knives were the calmest group I track, which surprised me given their prices. Cases were the jumpiest by a wide margin. The Recoil Case started and finished the window at $0.27 but touched $0.12 on 7 and 10 August and $0.29 on 19 August. That is a 141% range around a flat result. At those prices, percentages are the wrong unit; I look at cents instead.

The practical use of this table is setting alerts. Six of the eight knives stayed inside a 7.4% range for the whole window, so a 10% knife move deserves a look (the two that exceeded it, the Gut Knife and Butterfly Knife Dopplers, are exactly the ones I would have wanted flagged); a 10% move on a case was routine. If you run a tracker with one threshold for everything, cases will flood it with noise and knife moves will be the ones you miss. The CS2 skin price tracker guide explains why a single threshold fires constantly on cheap items and never on expensive ones.

I do not have trade volumes, so I cannot measure liquidity directly. What the feed does record is each marketplace's lowest asking price for the same item at the same moment, and the gap between venues is a rough stand-in for how efficiently an item trades. To keep it like-for-like I compare market.csgo against CSFloat, the two venues present for almost every item at every snapshot:

By category on 22 August, cases had by far the tightest gap at 4.7%, while knives sat at 14.1%, AWP at 15.0%, rifles at 15.4% and pistols at 14.5%. The three gloves showed 20.3%, again too few items to lean on. Cases being tightest fits the idea that very cheap, heavily traded items get arbitraged quickly. Many sellers, near-identical goods, and a few cents of difference is enough to move buyers.

Where was the cheapest listing? On 22 August CSFloat held the floor for 34 of 56 items, Skinport for 13 and CS.MONEY for 9. If you are buying, that is a reason to check more than one venue: the cheapest venue changed from item to item, and the median gap of 13% is real money on anything over $50. The price checker shows every venue's live price side by side. The marketplace comparison covers the fees and cash-out rules that the asking prices leave out.

One caveat: these are asking prices. A wide gap can mean the dearer venue is simply slow, not that anyone pays its price. It is a liquidity hint, not a liquidity measurement.

On the 13 August snapshot three unrelated items printed their window low at the same time: the AWP Asiimov (Field-Tested) at $81.67 (from $108.02 on 31 July), the Glock-18 Water Elemental (Field-Tested) at $14.39 (from $18.67), and the Souvenir AWP Dragon Lore (Field-Tested) at $5,950. By 16 August the Asiimov was back at $104.69 and the Glock at $18.42.

I cannot tell whether that was a soft afternoon across several venues or my sampler catching a handful of mispriced listings at once, and nine data points will not settle it. I am recording it because a report that only describes the smooth parts of a dataset is misleading. If you see a spike in your own charts that reverses within one or two samples, suspect a single listing before a market event. The full walk-through of this cluster and the other artefacts is in CS2 skin price history.

This is the list I wrote on 17 August of things I wanted to watch through the autumn. It is not a list of recommendations. I have kept the first and third columns exactly as published, because the October report grades them against those words. I have rewritten the middle column, which originally leaned on claims I have now removed.

Two of these rows were badly chosen, and I would rather say so here than have you find out later. My data starts on 31 July 2026, so the first row's test against 2025 was never something I could run. And my feed records asking prices, not volume, so the second and fifth rows' tests cannot be checked either. The October report grades what it can and says so for the rest.

For transparency, this is what the original 17 August version said that is no longer on the page:

• A price spreadsheet on "400+ skins since 2023". The dataset behind this site starts on 31 July 2026 and covers 56 items. The report now uses only that.
• A 2024–2026 market history in "three acts", including a case index that "roughly tripled" and a specific AK-47 Redline run in summer 2025. None of it came from data published here.
• A "spring report" and a scorecard of its predictions. No spring report exists on this site.
• A category table of "roughly +10–20%" style bands against a 2025 peak. It is replaced by the measured 22-day table above.
• Market-size, owner-count and daily-volume figures attributed to a third-party report. I could not retrieve that report on 14 October 2026 to confirm the numbers, so they are gone rather than repeated unchecked.
• A personal selling anecdote, fee arithmetic and dated claims about late-2025 volatility that I had no record to support.

The result is a narrower report, and I think a more useful one. It describes three weeks of 56 items, and it describes them exactly. If you want the method behind the numbers, it is laid out in the investing hub. The live prices for every tracked item are on the CS2 skin prices homepage.