Mid coaching is dominated by CS discussions. 80 creeps by minute 10. Deny count above 20. The benchmarks feel rigorous and specific. They’re almost entirely disconnected from whether you win.
We ran full correlation analysis on 404 mid-lane records from 403 professional matches. Here’s what predicts winning.
The Full Correlation Ranking
Kill participation at r = 0.665. Assists at 0.581. XPM at 0.569. GPM at 0.541. Tower damage at 0.450. Then: CS at 20 min (0.04), CS at 15 min (0.03), CS at 10 min (0.02).
The three CS metrics are statistically indistinguishable from noise. Every other metric in the top five has correlation above 0.45.
Why Pro Mids Adapt Their Farm Pattern
A mid who goes 3-0 in lane has lower CS than a 0-0 safe farmer. CS doesn’t know the game state. XPM does — it captures kills, camps, and CS combined. A mid with three kills, four camps cleared, and 70 CS has high XPM. A mid with 85 CS and 0 kills might have lower XPM.
Pro mids adapt their farm pattern to the game. If a kill is available, they take the kill. CS benchmarks penalize this. XPM rewards it.
What You Should Track Instead
- XPM. The single best individual farm metric. Captures the full picture across 20+ minutes.
- Kill participation %. r = 0.665 — the strongest signal in the dataset. Mids who create for their team win more than pure KDA farmers.
- Rune pickups. 10+ runes = 68.9% win rate. One of the strongest threshold predictors in the entire dataset.
- Assists (r = 0.581).The best mids in this dataset didn’t top the CS chart. They topped kill participation and assists.
The Coaching Implication
Stop giving mids CS feedback. Ask instead:
- What was your XPM at 20 and 30 minutes?
- How many runes did you secure?
- Were you creating for your team or just farming?
- What was your kill participation percentage?
These questions correlate with winning. CS@10 does not.
The Data Behind This
All findings derived from DOTApulse’s 403-match pro dataset across patch 7.41.
Read the full research →Related articles