CS2 Skin Arbitrage: Real Profit, Not Price Gaps

The same skin trades at different prices on different marketplaces — that part takes a minute to see. What almost nobody writes about starts right after: a price gap and a profit are two different numbers, and between them stand the selling fee, a seven-day hold and liquidity. Here is how to run the numbers so that your own arithmetic is not the thing that costs you money.

What CS2 skin arbitrage is

Arbitrage means buying an item where it is cheap and selling it where it is expensive. It works because the skin market is fragmented: there is no single exchange, and every marketplace has its own pool of buyers, its own fees, its own currency and its own withdrawal speed. The same AK-47 | Redline (Field-Tested) lives on a dozen marketplaces at once, and their prices are not synchronized.

AK-47 | Redline (Field-Tested)best price: $28.97 · whitemarket

On a liquid item the spread between the cheapest and the priciest cashable marketplace is usually 5–20%. Wider gaps exist, but they are almost always thin items where the expensive side is an ask nobody is hitting — a 40% spread is more often a liquidity warning than an opportunity. The catch is that you measure the spread before deductions and live with it afterwards.

The spread is not the profit

One formula covers it, as long as you keep every cost inside it:

net = sell_price × (1 − sell_fee) − buy_price × (1 + deposit_cost) − withdrawal_cost
profit % = net / buy_price × 100

Steam sell leg: use sell_price / 1.15, the fee is charged on your proceeds

On most marketplaces the buyer pays no fee — the seller carries it. So the whole calculation hinges on the fee of the leg where you sell.

Take a gap that looks great: bought at $10.00, sells at $11.50. Gross spread: 15%.

The identical gap pays anywhere from nothing at all to +$1.27, and the only variable is where you close the trade. The Steam row is not a rounding accident: Steam charges its 15% on the seller's proceeds, so a $11.50 tag pays out exactly $11.50 / 1.15 = $10.00. A 15% gross spread is precisely the Steam break-even point — anything narrower is a straight loss, and even break-even converts spendable dollars into wallet funds you cannot withdraw.

Hence the first rule of thumb: a flip does not start with "it is cheaper here", it starts with "what is the fee on the sell leg, and does the gap cover it".

Three traps the spread does not show

Steam money stays in Steam

The Steam Community Market takes 15% (5% Steam plus 10% game-specific for CS2) and, more importantly, does not pay out. You sell higher, you get wallet balance that can only be spent inside Steam. For someone growing a gaming balance that is fine. For someone who wants money on a card, "buy cheap, sell on Steam" is not arbitrage at all: it converts spendable money into money you cannot withdraw.

That gives a split between marketplaces that matters more than the size of the fee:

• Cashable — money can be withdrawn as fiat or crypto: Skinport, CSFloat, CS.Money, Market.CSGO, Waxpeer, CS.Deals, Buff163, White.Market and others.
• Non-cashable — Steam (wallet) and trade bots: LOOT.Farm, Tradeit.gg, SkinsMonkey, SkinSwap, CS.trade, Swap.gg. Their fee often looks like zero because there is no fee in the usual sense: you swap items for items, and your "income" arrives as another skin.

A trade bot's zero fee is not a discount — it simply pays you in items instead of money. Profit held in items becomes money only after one more sale — with a fee attached.

The 7-day hold

An item you buy or trade for goes into a 168-hour trade hold. This is not a formality but a real risk: for a week your capital is locked inside an item whose price can drift down. Since June 2026 Steam Market purchases unlock at the exact time of purchase instead of a shared daily slot, which smoothed out the price swings around unlock — but the week of waiting is still there.

Put a number on that. If a thin position routinely moves 10% in a week, a flip that nets 3% is a coin toss paying a third of its own risk. A usable threshold: net after fees should be at least the size of that item's typical weekly move — roughly 5–8% on liquid positions and considerably more on thin ones. The hold does not forbid small flips, it prices them out.

Illiquidity: a big percentage as a symptom

The fattest percentages almost always sit on items nobody buys. A 40% gap on a niche gun does not mean you struck gold, it means nothing trades at that price on this marketplace. Paper profit then freezes in your inventory for months.

That is why sales volume is a hard filter and not a nice-to-have. Look at two things together: what percentage the flip pays and how many trades a day go through the item. Liquid and modest beats fat and dead almost every time.

Depth, not price, is what caps this strategy. Before buying a fifth copy of anything, look at how many sales a day the item does and how many listings sit within 2% of the price you plan to sell at. Once your position is larger than a day of volume you are no longer a participant in that price — you are the reason it moves.

What to price from: median vs lowest ask

The second most common mistake after ignoring fees is measuring the spread from the lowest listing.

The lowest ask is somebody's asking price, not a completed trade. Often it is a single cheap listing, a pricing error, or an item that is about to be bought. Building a calculation on it means planning to sell at a price the market does not offer.

Price layers, from optimistic to realistic:

1. Highest ask — a wish, not a price.
2. Lowest ask — the ceiling for your own sale, sensitive to outliers.
3. Median of sales — what the item actually goes for. The working fair price.
4. Highest bid (buy order) — what someone pays right now. The floor of your exit.

Price the two legs on different bases, because you meet the market on opposite sides of the book. The buy leg is the ask you can actually hit right now — and on a real deal that ask sits below the venue's own median, which is the entire point of the trade. The sell leg is the median of real sales for a fair estimate, or the highest bid for a conservative one. Median-to-median flatters the entry, because nobody sells you the item at the median: that is what makes it the median.

The four layers are not a fixed ranking either. An underpriced listing sits below the median, and that inversion is the deal. What is fixed is only the frame: the highest bid is the least you can get today, and the highest ask is the most anyone is asking.

That is why our item card shows the median of real sales next to the marketplace price: an ask below the median is highlighted, and that — not "the lowest number in the row" — is the trade candidate.

Live deals right now

Below is the top of the current table: net is already computed after the fee on the sell leg, not as a gross gap.

Pre-trade checklist

1. The sell leg and its fee — which marketplace closes the trade and how much it takes.
2. Cashable or not — will the money reach your card or stay inside an ecosystem.
3. Net in money, not in percent — 12% of $3 is 36 cents, which does not pay for the time it took.
4. Sales volume for the item — is there anyone to sell to at your assumed price.
5. The hold — are you willing to carry the position for a week of market risk.
6. Price from the median, not from one lonely cheap listing.
7. Cross-listing — if the item is listed on several marketplaces, pull the other listings the moment it sells: a double sale ends in a ban.

FAQ

How much money do I need to start? What matters is net in dollars, not the percentage. Below roughly $200-300 of working capital a good flip pays cents and one mistake wipes out a week of them. The work starts making sense where the net per trade covers the time the trade takes.

Why did my flip pay less than I calculated? Three usual reasons: you counted from the lowest ask but sold at the median; you ignored the fee on the sell leg; the price moved while the hold ran.

Is Steam better than third-party marketplaces? As a place to sell — almost never: 15% and locked money. As a price source — yes, Steam gives a reference median and volume that everything else can be measured against.

Can you live off arbitrage? It is a volume business, not free money. Earnings are capped by liquidity: the number of items you can really buy and sell in a day without pushing the price down is finite.

What about trade bots? Use them as a source of cheap items and as a way to reposition quickly, but do not treat the difference you gain there as money until the item has been sold for fiat.

What next

Fees in detail — a full table by marketplace and what reaches your card out of $100 — are in a separate breakdown: marketplace fees. If you care less about the gap between marketplaces and more about underpriced copies inside one of them, look at float and pattern: the profit mechanics there are different.