League of Legends Stats, Builds, & Win Predictions

How do you come up with your build suggestions?

We analyze the information from millions of League matches each patch. That information is dissected using state-of-the-art statistical methods to determine which items will provide you the highest probability of winning against different team compositions. Don't just rely on one build suggestion from other guides. Use the build that has been proven to be effective against your enemy team comp.

How often do you update your data?

We update our data often, approximately one to two times a week. This schedule allows us to acquire enough match history to make quality predictions and provide detailed statistics while keeping our pages up to date with the current meta.

What is machine learning?

Machine learning is a branch of computer science that develops algorithms that allow computers to discover trends and make predictions without being explicitly programmed how to do so. This is the same field that has allowed Alexa to understand you and has taught your Google Assistant to recognize you in a new image.

How do you come up with your tier list?

We have developed a proprietary algorithm that weights each of our numerous factors differently for each role. For instance, we recognize that utility is much more important for a support than for an ADC, while mid often needs more damage than top lane. Furthermore, it is often up to the top and jungle champions to try to split push while the support stays with his/her carries; our algorithm takes this into account as well.

Our tier list makes use of the same massive collection of data that we have put together for our stats and win predictions, so you can rest assured it is accurate and up-to-date with the current meta. Check out our champion tier list.

How does your match predictor work?

Given the number of unique champions available in League of Legends, there are over 4.5 sextillion (4.5E21) possible number of match compositions, not including different roles. We use modern machine learning algorithms to teach our Match Predictor to predict the outcomes of future matches. Using our proprietary method that takes into account much more than just lane counters or pair-wise win rates, we are able to anticipate the outcome of matches that have never been played before.