From the original Elo system, which is used to rank chess players, I derived a formula to fit into CS:GO, which takes into account round difference, overtime, roster changes, etc. I use the formula to rate CS:GO teams from the collected data of ~4000 matches and ~8000 maps.

The Elo system can give a team's probability to win against the other. With this probability and Kelly criterion, we can calculate how much to bet on a team. We have done simulations to test out the profitability of our system working with Kelly criterion.


No high/med/low bets.
The amount to bet for is calculated in percent, and it is usually between 2% - 10% of your bankroll.

Lots of underdog bets.
I'll be honest here, I think 90% of the bets will be on underdogs. Why? The community often overhype the overdog more than the underdog, resulting in profitable odds if a bet is placed on the underdog.

Lose a lot, but win a lot.
I've simulated betting with this method. Taking CSGL odds for matches since 2.5 months ago, and here is the result. If we strictly stick to the system, we would have made 275% in profit. However, the win rate is only 37.5%. That's why I call it "Lose a lot, but win a lot."

The spreadsheet can be found here: https://docs.google.com/spreadsheets/d/1YHJEOIU0RUzXtLVucnv8If31ITrOJ1T3AxYaHDWauEk/edit?usp=sharing


If you are interested in this group and method. We have more instruction posts in the General Discussion forum. And we also have a spreadsheet to keep track on this method and post probability of matches each day.

• Spreadsheet: https://docs.google.com/spreadsheets/d/1uVAzEkx_PaV-DzHFQ7ai0smPdUAcAj1yuj7D3EAt0uo/edit?usp=sharing
  
  
• Simulation Spreadsheet: https://docs.google.com/spreadsheets/d/1YHJEOIU0RUzXtLVucnv8If31ITrOJ1T3AxYaHDWauEk/edit?usp=sharing