Last week looked like a slowdown. This week looked more like the market remembered the original assignment.
Across 33,456 comparable references after excluding Sticker Slabs and Cologne Ranked stickers, the median move from August 14 to August 21 was about -0.82%. That is still not a market-wide collapse. It is, however, a clear step weaker than last week's -0.20% clean median.
Breadth moved the same way. Using the same plus-or-minus 0.25% noise band as last week, 17,720 references moved lower, 10,725 moved higher, and 5,011 stayed roughly flat. The decline was not only deeper at the middle; it covered more of the clean comparison set.
The shape was familiar. Gloves were again the weakest reliable broad category. StatTrak weapons, knives, general weapons and items, containers, and original Cologne stickers all moved lower at the median. Non-Cologne stickers, souvenirs, and charms were flat at the median, but still produced enough extreme outliers to remind us why a market average is a dangerous narrator.
This is not financial advice. CSBeacon is not a marketplace, broker, or trading advisor. These notes are meant to help CS2 inventory owners understand what changed and how much confidence the available data deserves.
The clean categories leaned red
The most useful weekly comparison is still the boring one: ignore the obviously noisy products, compare only positive references on both dates, and ask where the median moved.
Among comparable references from August 14 to August 21:
- Gloves had a median move of about -2.93%; 395 references fell beyond the noise band, while 57 rose.
- StatTrak weapons had a median move of about -1.88%; 2,091 moved lower and 981 moved higher.
- Knives had a median move of about -1.66%; 3,334 moved lower and 1,446 moved higher.
- General weapons and items had a median move of about -1.36%.
- Containers had a median move of about -1.21%.
- Original Cologne 2026 stickers had a median move of about -1.53%.
- Non-Cologne stickers, souvenirs, and charms were flat at the median, though each had meaningful movement underneath.
Gloves remain the cleanest bearish signal because the median and breadth agree so strongly. Nearly seven glove references moved lower for every one that moved higher. That is hard to explain away as a few odd listings.
Knives and StatTrak weapons were less severe, but their breadth was also negative. These are the categories that matter most to many inventory totals, so a broad headline near -0.82% can understate the felt move for anyone concentrated in premium wearables or StatTrak skins.
The other important point is that this was not just another sticker-data problem. The clean set excludes Sticker Slabs and Cologne Ranked stickers. The weakness still shows up.
Armory reds kept repricing lower
The four Season 5 red weapon finishes are now several weeks removed from launch, and the price-discovery process is still leaning down.
Across the 20 standard wear references for AK-47 AUTOEXEC, AWP Sovereign Flame, Glock-18 Ghost Protocol, and AK-47 Consequence of the Jinn, 19 moved lower from August 14 to August 21. The only exception was Glock-18 Ghost Protocol Field-Tested, which rose about 2.7%.
Factory New remained the easiest place to see the premium unwind:
- AK-47 | AUTOEXEC fell from about $747.38 to $623.54, roughly -17%.
- AK-47 | Consequence of the Jinn fell from about $678.44 to $602.46, roughly -11%.
- AWP | Sovereign Flame fell from about $499.94 to $455.34, roughly -9%.
- Glock-18 | Ghost Protocol fell from about $619.92 to $582.46, roughly -6%.
That does not mean the skins failed. It means the first month of pricing still looks like launch premium being negotiated downward.
AUTOEXEC is the most obvious example this week: every wear fell, with Factory New down the most in dollars and percentage terms. Sovereign Flame also declined in all five standard wears. Ghost Protocol had the only green wear reference, but its Factory New, Minimal Wear, Well-Worn, and Battle-Scarred references all moved lower.
The practical inventory lesson is still the same as it was earlier in August: new high-rarity skins can be desirable and expensive while still losing launch premium. Those facts do not contradict each other.
Cologne originals softened while Ranked stayed messy
Original Cologne 2026 stickers were not the most dramatic group this week, but they were reliably negative. Excluding Ranked stickers and Sticker Slabs, the original Cologne median was about -1.53%.
By finish:
- Original Foils had a median move of about -4.23%.
- Original Holos had a median move of about -1.74%.
- Original Golds had a median move of about -1.17%.
- Other original Cologne sticker references were flat at the median.
The composition is nearly even: 199 comparable Foils, 199 Golds, 199 Holos, and 198 other original references. The combined median sits between the Holo and Gold medians because the full distribution, not the average of four subgroup medians, determines the midpoint.
That is a cleaner signal than the Ranked Series. Ranked references again produced impossible-looking individual moves in both directions. The combined Ranked median was about -9.97%, Foils were down about 31% at the median, Holos about 10%, and Golds were flat at the median. But the same data still includes individual Gold and Holo references moving from single digits into hundreds of dollars, and others moving from hundreds back to a few dollars.
That is not mature price discovery. It is a young, thinly listed series producing unstable aggregate references. The useful conclusion is not that Ranked stickers crashed by exactly one number. It is that the series still deserves a lower confidence label than ordinary, better-covered items.
Original Cologne stickers, by contrast, looked more like a normal weak market: not every reference down, but enough median and breadth pressure to treat the move as real.
The lenticulars mostly drifted lower
The Season 5 lenticular stickers gave back more ground, but not as one clean block.
This section returns to the full eight-item launch lenticular set highlighted on August 7. Last week's shorter discussion focused on the six most useful movers, so the comparison set is broader here.
Poo Cocktail Supreme was the weakest tracked lenticular this week, falling from about $1.84 to $1.54, roughly -16%. Pepper Gauge fell about 9%. Pineapple On Pizza, Electric Avenue, Burnout, Digital Dashboard, and NITRO! all declined between roughly 2% and 5%.
Partyboy was the exception, rising from about $3.18 to $3.60, roughly +13%.
Digital Dashboard remains the price leader in this small group at about $10.73, but it has now drifted lower for three consecutive Friday checks after peaking around $14.09 on July 31. The lead is still there. The acceleration is not.
The player-count backdrop is still uncomfortable
The price move is happening while the public player-count chart remains soft. Steam Charts showed CS2 at about 821,000 average players in the latest 30-day window when this post was verified, down about 3.75% from July. July itself was down about 6.92% from June.
That does not make CS2 dead. Steam Charts still showed a 24-hour peak above 1.1 million players, and CS2 remains one of Steam's largest games. But the direction matters because market confidence does not live in the item table alone.
CSMarketCap showed the estimated CS2 skin-market value around $6.11 billion, down about 3.4% over seven days and 8.9% over 30 days. That is an estimate of repricing, not cash leaving the market, but it matches the broader pressure visible in this week's stored price references.
We went deeper on the player-count argument in Is CS2 Dying?. The short version for these notes is that player count is useful context, but it is not enough to explain every item move by itself.
It is tempting to turn the player slide into a neat explanation for every skin-price decline. That is too easy. Valve's economy changes, new supply, summer seasonality, cash-market liquidity, and ordinary risk-off behavior can all move prices at the same time. Player count is context, not a complete cause.
For inventory owners, the best practical question is narrower: is your inventory exposed to the categories showing reliable pressure? This week, that means gloves, knives, StatTrak weapons, the new Armory reds, and original Cologne stickers more than it means the whole catalog equally.
Closing read
This week was weaker than last week, and the weakness was broad enough to matter.
The clean comparable set fell about 0.82% at the median, with far more fallers than risers. Gloves again supplied the strongest bearish category signal. StatTrak weapons and knives followed. General items and containers leaned lower too.
The new Armory reds kept losing launch premium: 19 of 20 standard wear references declined, and all four Factory New references fell. Original Cologne stickers softened, especially Foils. Ranked stickers remained too unstable to use as a normal market signal, even though their raw medians were red.
The most honest summary is not panic. It is renewed pressure. After one week where the selloff looked shallower, the August 21 snapshot says premium inventory still needs item-level attention.
Source and data notes
Market movement compares CSBeacon's stored global pricing snapshots from August 14 and August 21, 2026. Broad category medians include only items with positive comparable aggregate references on both dates. Moves within plus or minus 0.25% are treated as roughly flat for breadth counts.
Sticker Slabs were excluded from all sticker calculations and highlighted movement because they are separate, thinly listed items that can distort sticker comparisons. Cologne Ranked stickers were excluded from broad direction claims because their current references remain internally inconsistent and too unstable to treat as mature prices.
Valve's Cologne Ranked Series announcement is the primary source for the Ranked Series and its lower fixed-price structure. Valve's July 29 update confirms that Cologne shop items remain available through September 29. Valve's Season 5 announcement confirms the Armory weapon and sticker collections discussed here.
Steam Charts player-count data and CSMarketCap's market-cap estimate were checked as public context on August 21, 2026. CSMarketCap displayed an estimated market cap near $6.11 billion, 29.9% below its modeled all-time high, with a seven-day move of about -3.4%. Those public datasets are context, not proof of causation for the weekly price movement.
Prices are aggregate market-listing references processed by CSBeacon, not confirmed sale prices or guaranteed cash values. Thin listings, unusual variants, naming aliases, limited liquidity, and changing source coverage can create extreme apparent movement.