Video summary

The hidden formula behind Dota 2's most misunderstood feature

Main summary

Key takeaways

Gaming

Summary

The video investigates Dota 2’s quietly changed ranked MMR system, which Valve introduced without explaining how it works. The creator collected more than 1,000 ranked-game records across skill brackets, compared them with OpenDota data, and built simulations to explore the effects of the change.

How the system works

Before the change, wins and losses generally shifted MMR by about 25 points in either direction, with adjustments for factors such as rating confidence and team balance.

In the newer system, losses still cost about 25 MMR, but wins are worth more for players below roughly Legend 5. The win bonus declines as rank rises and returns to about 25 points at higher ranks.

The creator’s analysis suggests that MMR is unaffected by role, server, in-game performance statistics, play time, Dota Plus, or third-party data settings. Relevant factors appear to include rank and confidence before the match, the result, how closely matched the players are, and the balance between the teams.

What the change means for climbing

With a 50% win rate, lower-ranked players can gain MMR over time because their wins are worth more than their losses. The creator estimates that lower tiers can be traversed in roughly 100 games, while climbing through parts of Legend takes substantially longer. Reaching Ancient generally requires winning more than half of one’s games over a short period.

The analysis estimates that players may reach an equilibrium below a 50% win rate in some lower brackets—for example, around 43% at the bottom of Legend and about 40% at the bottom of Archon or Crusader. These figures are estimates from the creator’s analysis, not rules published by Valve.

Why the creator thinks the system may help

A ranked matchmaking simulation suggests that the old system can gradually become less effective at keeping players near their actual skill levels. Random match outcomes and uneven play frequency can create ranking errors that compound over time. The creator argues that awarding extra MMR for wins at lower ranks helps correct those errors more quickly. It may also prevent two under-ranked players from cancelling out each other’s progress when they repeatedly trade wins.

The creator describes the new system as a temporary sorting mechanism: it may improve players’ relative placement faster, even while inflating their displayed MMR. It could also move smurfs and account buyers toward their appropriate skill levels more quickly. However, the simulations are simplified and do not fully reflect real matchmaking.

Downsides and recommendations

  • The system can feel unfair because players may gain MMR without a positive win rate.
  • Rank inflation changes what MMR and medals represent. Over time, players may need higher MMR to maintain the same percentile.
  • Lower-ranked players who become stuck may lose more than half their games, potentially making ranked feel worse.
  • The creator thinks the change may help for a limited period, but Valve may eventually need to revert it, redesign ranked, or bring back seasons. If the system remains, medals should track percentile rather than fixed MMR thresholds.

The video’s central takeaway is that uneven MMR gains may look unfair while helping correct a matchmaking system that gradually mis-sorts players. The creator cautions that the analysis is based on incomplete game data and simplified simulations, not an exact reconstruction of Valve’s matchmaking.

Gamers and sources featured

  • Data sources: Ranked play-history submissions from the creator’s friends and followers; Dota Plus history readouts; OpenDota; Valve’s rank-distribution data.
  • Sponsor: RDY.GG, referred to as Ready.gg in the subtitles.
  • Named supporters acknowledged: Ayball, Rose Nightflower, Not Tat, Niko, Dota 2, PowerPoints, Where Did My Ants Go, Catalyn Stan, Anonymous Anna’s Hummingbird, Svenge, Male B, Friendly Fire, Adriel, Han’s Leak, and White Wind 664.

Rate this summary

Your feedback will help improve summaries.

Improve this summary

Reprocess with a stronger model when the summary feels incomplete or inaccurate.

Pro

Translate summary in another language

Pro

Ask questions to this video

Chat for follow-up questions, clarifications, and source-backed answers.

Coming soon

Share this summary

Original video