We have the game stats of all professional LoL tournaments for the last years. The historical data is our teacher — using it we trained AI based on machine learning models to predict the future. In this project we are focused on predicting results on the World 2020, check it out.
FINAL Worlds 2020 PREDICTION
SEMIFINALS Worlds 2020 PREDICTIONS
FINALS Worlds 2020 PREDICTIONS
2020-10-11 Worlds 2020 PREDICTIONS
2020-10-10 Worlds 2020 PREDICTIONS
2020-10-09 Worlds 2020 PREDICTIONS
2020-10-08 Worlds 2020 PREDICTIONS
2020-10-06 Worlds 2020 PREDICTIONS
2020-10-05 Worlds 2020 PREDICTIONS
2020-10-04 Worlds 2020 PREDICTIONS
As we wrote in the last post, we are just starting our experiments with predictions from League of Legends. The play-in stage was the first step in testing our model. Out of 24 matches, our AI gave 13 correct predictions and 11 incorrect ones. This brings the AI win rate to around 55 percent.
We looked at our model - we fixed some incorrect calculations - and are ready to share our predictions for the group stage and the first day of this stage. As you can see from our predictions, the most illogical is AI's bet on a team from the CIS region - Unicorn of Love. We checked the calculations several times, but the AI produces this result over and over again.
Here we can only say one thing - while making predictions for DOTA 2 matches, we faced a similar situation several times. And very often the bet on the underdog worked. Let's see how it will turn out this time. And if needed we will correct the model further.
As always, your comments are welcome - they can be incredibly useful, when working with our AI models.
2020-10-03 Worlds 2020 PREDICTIONS
2020-09-30 Worlds 2020 PREDICTIONS
2020-09-29 Worlds 2020 PREDICTIONS
2020-09-28 Worlds 2020 PREDICTIONS
2020-09-27 Worlds 2020 PREDICTIONS
2020-09-26 Worlds 2020 PREDICTIONS
2020-09-25 Worlds 2020 PREDICTIONS
Win Prediction and Tournament Simulation
In order to make predictions we have to predict how many matches the player will play next day and which of them will be winning. Here comes our supporting machine learning model that gets information about any pair of teams and estimates their win probability. For instance, now the model predicts the chance that PSG Talon will beat Rainbow7 as 65%, which means that if PSG Talon could play versus Rainbow7 1000 matches in a row and their skills, fatigue and strats will not change during all those matches, then about 650 games will be won by PSG Talon.
Having predictions and a known tournament grid we make a simulation of 1 million possible outcomes of World 2020. Favorites of our model to play-in stage are Mad Lions, LGD Gaming, Team Liquid and PSG Talon. But, the probability 15% of winning the tournament is still not very high. So, be rational if you decide to make a bet ;)
Stay tuned and subscribe to our Facebook and Twitter! Suggest us what else to predict.