The battle avg number is completely useless during regular war weeks

Currently the river race shows 3 stats – battles played, avg points/battle, projected daily war fames. The number in the middle, the battle avg number, is completely useless during regular war weeks as it is calculated using accumulative numbers. During war week, a clan could have 180 on first day, but drop to 170 next day. The accumulative battle avg does not reflect such swings.

It is only useful for the colo week.

I now have to install other bots to help me calculate this daily battle avg, instead of accumulative avg. Please, change it to daily calculations. As a war clan leader, I don’t really care about yesterday’s win rate. I need to know whether we can beat the other clans TODAY.

See above example. This is taken today. It gives you an illusion the two clans are tied, but in reality:

OiaForce is at 189, Indo is at 181, a huge difference. Not tied at all.

Hi, I’m not sure I understand your concerns.

If you want to see which clan is doing better you should look at the projected score.

Average daily scores would be relevant if all your clan members had a similar skill level, but our research suggest this is an incorrect assumption for most clans. Our average has proven to be significantly more reliable as a predictor, and that’s why we use it.

And to be clear, there isn’t a single calculation method that’s better for all cases, but the one we are using was better more often, and really close when it wasn’t. In contrast, daily averages were extremely inconsistent. Again, this may not be the case for your specific clan, but it is for the vast majority.

The projected score is also pointless as it’s using accumulative points from day 1 of the week.

My point was whether it’s battle avg or projected score, it has no bearing on whether my clan is leading today or not, because all previously day’s performance was calculated as well into these numbers.

As you might know, clan performance swings high/low day by day. Yesterday’s battle avg could be 180, but today it could be 170, and tomorrow it could be 190. By using accumulative war points throughout the week, the battle avg (or the projected score) does not provide an accurate picture which clan is going to win TODAY. As you know, river race is played one day at a time, not by the week (only colo week is accumulative). Each day, clan performance is reset to 0. Therefore, any accumulative calculations are pointless in TODAY’s clan performance.

Hope you can understand my point now. It may be a good indication of an overall clan performance, but who cares about that during a given war day? Like today is Saturday, why do I care how well the clan did on Thursday or Friday? I need to know how well we are doing TODAY to see whether we can beat the other clans.

Here’s why you should care, precisely because players only play once each day. So players having a good winrate on Saturday morning will have absolutely no bearing in those who play Saturday afternoon.

And that’s why aggregate average is better: it uses stats from your whole clan to predict performance, instead of just a few samples from whoever played early each day.

At least that’s the reasoning that I’m using, which may be right or not.

What is not debatable is that this projection method is better; as I said earlier, we have tested several approaches and this was the most accurate.

“So players having a good winrate on Saturday morning will have absolutely no bearing in those who play Saturday afternoon.” This is absolutely correct! Then why does a player’s yesterday’s win rate has any bearing on today’s win rate? You are assuming players win rate is a fairly consistent and predictable stat, but most of the times it’s not the case. There are 50 players in the clan, sure 10 of them, maybe 20, could be consistently getting 800 avg/day, but the other 40 could swing from 900 to 500 from day to day. Those swings make your accumulative calculation not a good indicator whether your clan is going to win today or not.

Otherwise, how would you explain the two screenshots in my original post? Your API says our two clans are tied but in reality, for that day, we are way ahead. Your accumulative calculation is only good for Thursday (since it’s first day, accumulative is same as single day) and colo week. For any other day, I have to manually calculate the fames divided by battles played to see how well our clan is doing on any particular day. And then also manually calculate on other clans to compare.

This is, until I installed CW2 discord bot (my second screenshot), which precisely calculate that day’s battle avg, without considering previous days’ performance.

Previous day performance is a decent predictor because it includes the players that haven’t played today (yet).

In contrast to current day average, which is using stats from players who have played to predict those who haven’t. The intersection with those 2 groups is the lowest available, null, so it should not be considered a good predictor.

And again, specific cases are anecdotal, but the screenshots you posted go against what you are suggesting: your clan lost on Day 3 despite having a significantly higher daily average.

This is the simplest way I can put it:

  • We know A&B stats of a few days
  • We know A stats from today
  • How will B perform today?

Using A&B will at least have some correlation, predicting B performance from A isn’t warranted.

Just to give you another example.

Took the screenshots a few hour before cutoff. Your stat shows OiaForce falling behind Indo but in reality we lead by like 5 points margin and eventually won.

Your accumulative stat become more and more useless towards the end of week. As I mentioned before clan performance swings from day to day. This is not an isolated incident. It happens so often I simply cannot use your tool anymore and installed a different bot to calculate win rate on current day only.

Ok, well, as I said, we can’t base our decisions on specific cases. We’ll keep showing what works better most of the time.

Btw, based on your screenshot I see that there is a bug in the projection calculation; projected score should always be higher or equal to current score, it should say 36279 instead of 35785.

Can you show two numbers? Weekly and today’s win rate?