How the SkinScope Skin Index beats raw platform prices
Why a weighted-median fair-value across 34 CS2 marketplaces gives traders a better reference price than any single venue. Methodology, examples, and why outlier filtering matters.
A trader opens SteamAnalyst, then Buff163, then CSFloat, then Skinport. Four tabs, four prices, four fee structures. Question: what does this skin actually cost?
The answer everyone reaches for is “median”. Take the middle price across platforms. It feels fair. But the median is fragile: if one venue spikes because of a thin order book, or Steam wallet-locked pricing pulls the whole distribution down, the median moves with it. That is a feature of the number, not a bug in the market — and for serious traders, it is the wrong reference.
The SkinScope Skin Index (SSI) is our answer. It is a weighted median with outlier filtering, inspired by how CME computes the COMEX gold settlement price. One number per skin, trustable enough to trade against.
The three problems SSI solves
A good reference price has to be resistant to three predictable failure modes of the CS2 market.
1. Thin-listing spikes
A Souvenir SCAR-20 with a single €2 listing on CSFloat and a €0.84 listing on Skinport is not a 138% arbitrage. It is a data anomaly. Plain mean or median would happily average those two values and produce a misleading fair value.
SSI applies a Modified Z-score outlier filter using MAD (median absolute deviation). Any price more than 3.5 MAD-units from the median gets excluded before computing the final number. Single-listing spikes and fat-finger mistakes never reach the index.
2. Wallet-locked Steam prices
Steam Community Market is structurally cheaper than Western venues because the seller can only cash out into Steam Wallet (not EUR). The €24 listing on Steam is not €24 cash; it’s €24 “Steam credit,” which is worth roughly €17–€19 depending on your willingness to buy games.
SSI weights Steam at 0.3 — included for breadth, but never dominant. Buff163, CSFloat and Skinport (the main cash-settled venues) weight at 1.0.
3. Cross-platform noise at different capture times
Skinport refreshes every 5 minutes, Steam is rate-limited to roughly one item per 4 seconds, Pricempire aggregates on its own cron. If you average prices that were captured at different moments, you are averaging different markets.
SSI only admits snapshots from the last 6 hours. Anything older is ignored. If coverage drops below 2 platforms fresh, we emit a null fair value rather than fabricate a number.
The weighted median, step by step
Plain English: sort the surviving prices ascending, walk up them while accumulating platform weights, and take the value at which you cross half of the total weight.
Why not the weighted mean? Mean is sensitive to extremes —even after outlier filtering, a mean gives too much pull to whichever venue happens to be at the extreme of the surviving distribution. Weighted median is the robust choice for skewed distributions, and CS2 price distributions are always skewed: Buff is structurally cheapest, CSFloat is middle, Skinport and the rest pile up above.
A worked example — AK-47 | Redline (FT)
Say we have the following latest snapshots:
| Platform | Price (EUR) | Weight | Survives filter? |
|---|---|---|---|
| Buff163 | 27.89 | 1.0 | Yes |
| CSFloat | 28.04 | 1.0 | Yes |
| Skinport | 28.50 | 1.0 | Yes |
| Waxpeer | 28.62 | 0.7 | Yes |
| DMarket | 29.12 | 0.7 | Yes |
| Steam | 24.30 | 0.3 | No — flagged as outlier |
Median of fresh prices: 28.27. MAD: 0.41. Steam’s €24.30 sits 9.7 MAD-units below median — flagged and dropped.
Weighted median across the five survivors: €28.14. That is the SSI.
Raw arithmetic mean of the six prices would have been €27.75, dragged down 1.4% by Steam’s wallet-locked outlier. A 1.4% drift on your entry price is the difference between a flip being profitable or barely breakeven after fees.
Confidence, not just a number
Every SSI comes with a 0–100 confidence score that blends:
- Coverage — how many platforms fed into the index. 6+ venues is full coverage; 2 is sparse.
- Tightness — how disagreeable the venues are. If surviving prices span €27–€30, confidence is high. If they span €24–€35 even after outlier filtering, the market is disagreeing and confidence drops.
- Platform diversity — 4+ distinct venues earns a small bonus. Prevents a rogue single marketplace from dominating.
A high-confidence SSI is a price you can execute against. A sparse SSI is a hint, not a decision.
Why this matters for your flips
When you see SkinScope show a +18% margin on a deal, the margin is computed against the fee-adjusted SSI — not against the cheapest listing you happen to have open. That matters because:
- Fat-finger mistakes that briefly create 50% “arbitrage”don’t surface as signal. You won’t chase ghosts.
- Steam wallet-locked prices don’t make flips look artificially attractive.
- Prices from different capture times don’t mix. You aren’t arbitraging against yesterday.
The SSI is visible on every skin detail page next to the platform price table. When you’re deciding whether a flip is real, it is the first number to check.
See it live: open any skin, for example AK-47 | Redline (Field-Tested), and look for the SSI card under the hero. The per-platform breakdown shows which venues fed into the number and which were filtered as outliers.