DOTA 2 · Match case study
Why do three lane estimates still leave the winner uncertain?
Recalculate the Dawn Bulls vs Synapse Game 3 lineup from EPL Masters III using historical matchup samples of 7, 9 and 10 games. Separate assumed roles, economy estimates and the recorded result.
Verifiable match facts
- Event and stage
- EPL Masters III · Group B
- Match date
- Game
- 3 · BO3
- Map result
- Synapse won · 56:07
- Final kills (Radiant : Dire)
- Dawn Bulls 42 : 40 Synapse
Calculation: read sample size alongside the estimate
All differences are from Dawn Bulls’ perspective. The estimate is the historical mean of 0.35 × cumulative earned-gold difference at minute 5 + 0.65 × that difference at minute 8, in gold. It is neither net-worth difference nor a measurement from this particular match.
Sample scope: professional games in the 7.41 series; current subset: 7.41f. This describes the dataset, not independent verification of the target match’s subpatch.
| Statistical matchup | Series / current subset | Mean weighted gold difference | Confidence |
|---|---|---|---|
| Spectre / DoomCores only; supports uncontrolled | 7 / 2 | +291.5 | Low confidence · fewer than 12 games |
| Necrophos / Earth SpiritMid-lane matchup | 9 / 4 | +624.8 | Low confidence · fewer than 12 games |
| Beastmaster / Nature's ProphetCores only; supports uncontrolled | 10 / 8 | -236.4 | Low confidence · fewer than 12 games |
1. Separate match facts from role assumptions
The match page confirms factions, ten heroes and the result. This case does not verify actual lanes minute by minute from a replay. The calculation uses Bladeform’s automatic role assignment, with no position locks or team/player hero pools. These roles are analysis assumptions, even when they happen to match the order of the player list.
2. The largest positive estimate can still be weak evidence
For Necrophos versus Earth Spirit, the historical weighted economy difference averages about +624.8, but the selected 7.41 professional-series sample contains only 9 games; its 7.41f subset contains 4. Nine is below the 12-game confidence threshold, so the tool still marks the evidence as low confidence. This is not the target match’s eight-minute gold difference or a win probability for Dawn Bulls.
3. The side-lane samples cover core matchups only
Spectre versus Doom and Beastmaster versus Nature’s Prophet use 7 and 10 core-lane samples respectively, without controlling the support pairing. These records cannot isolate the effects of Mirana, Rubick, Windranger or Dazzle. The 2-game and 8-game 7.41f subsets are already included in the series samples; adding them would double-count games.
4. Use the result to check claims, not invent causes
Dawn Bulls finished ahead in kills, 42–40, yet Synapse won the map. Final kills, hero picks and historical matchup aggregates do not establish which fight, item or decision caused the outcome. This case shows why local matchup tendencies, final kills and the match result describe different things and cannot substitute for one another.
5. How to inspect the lineup in the tool
Copy the lineup below and import it through Lineup Overview → Paste both lineups after signing in. To compare against this case, clear player hero pools, restore automatic roles and check the data dates. Later snapshots or manual role changes may change the figures. The downloadable record preserves this calculation’s inputs, settings, data dates and results.
Copy the lineup and inspect the assumptions
Only this map’s picked heroes are included. Radiant maps to A and Dire to B. Listing order does not establish actual lanes.
Radiant (Dawn Bulls): Spectre, Necrophos, Beastmaster, Rubick, Mirana
Dire (Synapse): Nature's Prophet, Earth Spirit, Doom, Windranger, Dazzle