jcorvino / wows-ranked

This project estimates the number of battles needed to get to rank 1 in World of Warships.

Description:

This project estimates the number of battles needed to get to rank 1 in World of Warships.

Example Usage:

python wows-ranked.py -f 0.15 -m 3000 -s 20000 -o ./output

This will run the program with the following settings:

• a first place rate of 0.15 (15% chance of getting first place in a battle),
• a maximum of 3,000 battles per simulation (to prevent infinite loops),
• 20,000 simulation runs for each win rate.
• each win rate from 0.48 to 1.0 will be simulated separately (48% to 100% chance of winning a battle),
• One plot for each win rate, and the summary plot, will be saved to a folder called "./output"

Adding a --sprint flag will simulate a ranked sprint season instead of the regular ranked season.

Run python wows-ranked.py -h for the full help menu.

Example Output:

The program will save a histogram for each win rate simulated. It will also save a summary plot of win rate vs median battles.

Regular Season:

python wows-ranked.py -f 0.15 -m 3000 -s 20000 -o ./example-regular-season

Example Summary Plot:

Example Histogram:

Sprint Season

python wows-ranked.py -f 0.15 -m 3000 -s 20000 --sprint -o ./example-sprint-season

Example Summary Plot:

Requirements:

The project uses Python 3.8 and matplotlib. See requirements.txt for details.

Credits:

Original version by player Terror_Tost from the EU server.

wows-ranked's Issues

Update README.md

Latest commit changed how code works, but README.md was not update.

Add RNG seed

Add the ability for users to specify a RNG seed. This will allow for exact reproduction of graphs / simulations.

Add confidence intervals

Add 95% confidence intervals to the summary plot. Consider using the arch package and the BCa method to estimate confidence intervals of skewed distributions (note most of our distributions have a very high positive skew).