I analysed 25,000 matches to show how bad being contested is
This is a TL;DW version of my video, link is at the bottom of the post.
We know that it is bad for you to get contested, so I analysed 25000 matches to show how bad it actually is.
I took the A and S tier comps from patch 17.1 from MetaTFT and group them based on their playstyle: 1,2, and 3 cost rerolls, and Fast 8.
For The 1 cost rerolls, these are Teemo Leona, TF, Veigar
2 cost rerolls just includes Belveth
For the fast 8s we have Xayah, Space Groove, Kindred, Corki, and Leblanc
3 cost rerolls we have Kaisa Fizz, Urgot, and Viktor.
Overall, these comps averaged a 4.06 AVP and 56% top 4 rate. But when they are contested, the AVP becomes 4.66 and a top 4 below 50%.
If you are uncontested, your AVP becomes 3.96 and your top 4 is now at 59%.
Some comp types are more affected.
| Comp | AVP Cont | AVP Uncont | Top 4% Cont | Top 4% Uncont |
|---|---|---|---|---|
| 1 Cost RR | 4.87 | 4.04 | 44% | 60% |
| 2 Cost RR | 4.94 | 4.11 | 42% | 59% |
| 3 Cost RR | 4.7 | 3.8 | 52% | 64% |
| Fast 8 | 4.53 | 4.14 | 48% | 56% |
Basically the data shows that A tier comps can perform as good as an S tier comp, while an S tier comps can be borderline B-C tier (<50% Top 4, 4.5 AVP).
This makes interpreting Master Yi comp in 17.2 really interesting. It has a >1 play rate in MetaTFT, and its an S tier comp with 4.22 AVP and 56% Top 4. The data is basically saying that on average the comp is contested and its that good, so if its uncontested this is basically the best comp right now.
In my video I talked more about the math on the impact of being contested and why 1 cost rerolls are more affected than say a Fast 8.
Another interesting point is the holding hand effect. Across all comp types, 50-60% only one player is at the Top 4, you are equally likely to hold hands together to the Top 4 or Bot 4 (20-25%).
Video is here