← All posts
2 min readsteam-apitechnicaltroubleshooting

CS2 Stats Not Showing? Why Steam's API Returns Nothing for 75% of Accounts

CS2Scan shows N/A for Precision and Accuracy on most profiles. It's not a bug — Valve's CS2 stats endpoint is unreliable. Here's what happens and how to fix it.

If you've used CS2Scan and seen N/A for Precision, Accuracy, and Efficiency, you hit the biggest limitation in the entire CS2 stat-checker space.

Roughly 75% of accounts return no CS2 stats

Valve has two stats systems:

  1. Legacy stats endpoint (GetUserStatsForGame with AppID 730) — built for CS:GO. Often empty for CS2 accounts created after the CS2 transition, or accounts that never played CS:GO.
  2. CS2 game stats — not exposed publicly via Steam Web API in a reliable way.

Result: even with a public profile, game details set to public, and 1,000 hours in CS2, the endpoint can return {}. This is documented by every checker: Leetify, FACEIT Finder, and TrackBans all have the same problem.

What CS2Scan does when this happens

Instead of inventing numbers, we do what an honest tool should do:

  • The engine that needs that data reports N/A
  • It drops out of the weighted vote entirely
  • Confidence is recalculated and shown as Low / Medium / High

You will see this on the result page: "CS2 statistics unavailable — Precision, Accuracy, Efficiency engines offline. Score based on Network + History only."

That is intentional. A guess would be worse than no data.

How to make your own stats visible (if you want to be checked)

  1. Steam > Profile > Privacy Settings > My profile: Public, Game details: Public
  2. Steam > Profile > Edit > Privacy > Inventory: Public (helps Opacity score)
  3. Play one competitive match after changing settings — Valve's cache can take 24h to update
  4. Don't use the "Hide my game details from friends" toggle

Even then, ~30% of fully public accounts still return nothing due to Valve's side. That's Valve, not us.

What this means for detection

This is why we weighted Network and History higher. They are available on >95% of public friend lists, while aim stats are available on <25%. Any model that claims 95% accuracy on stats alone was tested on the biased subset that does have stats — not the real population.

If you are building your own checker, handle N/A as first-class, not as zero.

Want to see live handling? Check any profile at cs2scan.org and look for the confidence badge.