Player Ratings – Part 2

You may have seen my post last week where I discussed calculating player ratings. I’ve had a number of further thoughts (and dead ends) since then!

Player Ratings

I’ve refined this ever so slightly, but I’m happy with the way these are now calculated for players. Player will score points at each GT they attend, the points available vary from faction to faction depending on their win rates.

How They’re Calculated
Each faction’s win rate is multiplied by 5 (the number of games at a GT) to give an expected result (for an average player) playing with that faction. This is then rounded to the nearest 0.5 (to represent draws):

These expected values are then divided by a value of 100 to give a points value for each win.

This would mean if that a player achieves their expected result, they would score 100 points. They score more for beating this and less for falling short. For example, a Gloomspite Gitz player scores 53 points for a win, while a Beasts of Chao player scores just 32.5 points. Meaning that if both the Gloomspite Gitz and Beasts of Chaos players both achieved 3 wins, then the Gloomspite Gitz player would score more points for achieving this with an underperforming faction.

Tournament Ratings
Some tournaments draw more players than others and so attract better players to the event. I had a few thoughts on how to reflect this, all of which didn’t really work.

My first thought was to weight each faction with a rating based on their win rate, total up these for a tournament and divide that by the number of players. Only this always kept coming back to roughly the same figure, as the factions more often than not cancelled each other out.

My next idea was to take the scores of each player and take an average for that event. This also didn’t work as in some cases a 9 man event had a higher average player score than events such as NOVA or the London GT.

In the end I settled on the ATP Tennis system (or close to it anyway). Where each tournament is given a multiplier based on its size. These multipliers wouldn’t be score breaking, but they would differentiate between attending a large event and small event. In the end the multipliers I settled on were as follows:

PlayersMultiplier
1-16 Players0.9
17-32 Players1.0
33-64 Players1.1
65-128 Players1.2
129+ Players1.3

Based on this a player who achieves exactly the same result with the same faction at two different sized events will score slightly more points for the larger events.

For events where there is a player off for the champion, those games are also awarded points in the same manner as the previous 5.

An Example Event

Here are the results of the Justice Series GT which took place this last weekend. These results are as presented on BCP, ranked with the number of wins and then battle points from their games:

Now, if we apply our faction scores calculated above for each win and then multiply each player score by a factor of 1.2 (as it’s an 82-player event) we get the following adjusted leader board:

This method should allow players to compete on an equal footing regardless of which faction they choose to play with at an event.

When is it calculated?

This is the important part. As you can’t calculate the results on the day of the event. Win rates, as we all know, are fluctuating all the time. The easiest method would be to lock the win rates down once a new FAQ, Battlescroll or Generals Handbook is released for the game. That way you have set periods of time where players are all competing using the same rules, once the next release is made, the win rates are locked in for that period and scores can be calculated for the events that have taken place.

At the moment, we’re still in the GHB release phase and are waiting for the new FAQ/Battlescroll drop in the next week. Once that happens, we can lock down the scores for the tournaments that have taken place since the start of July.

Players will gain scores for their best four events during the six months under each Generals Handbook season. This would have been five or six normally, but as the seasons are dropping to six months, the number of events a player can attend in that time is different.

Current Top 20 Players

Using the method described above, I’ve calculated each players performance at each GT and have taken their top four scores:

This is only an example at the minute, as the win rates for this period are still fluctuating.

What do you think?

Let me know below or on our Discord server!