Hello Battlemages!
So my clickbait title lured you in? Hopefully I didn't put my face with a weird expression on the cover, I will keep that for youtube. Stay with me and extract the substantive marrow, for the content is actually valuable, you haven't been clickbaited for nothing!
In my quest for maniacal optimization of everything, it was long overdue that I take a look at how ECR works in combination with Rshares.
The formula goes with:
Obviously we want more Rshares. It seems intuitive that the higher the ECR, the higher the Rshare. More ECR, more Rshares, more chests.
So I got in mind that in order to maximize earnings, it is better to play at high ECR. With the hypothetical perfect player playing one game every 59 minutes. For sure it maximizes earnings per game. But is it really the best at the end of the day?
During normal play
What is normal play do you ask?
Let's say: when you are earning maximal Rshares. I guess it's the same for most players: after the reset at the start of season, we climb back until we made it to "cruise" rating, when we are limited (by choice or by Collection Power) in a league. Then it is not our actual rating that counts for Rshares calculation, but the maximum rating of said league.
What matters here is that we have 25% ECR to spend per day, no matter what. There has been lot of discussion on how to play with that.
- Play the 25% in one go and call it a day
- Play two sessions of 12-13% ECR
- Or three sessions, or as many as possible during the day
- Or play one game every 59 minutes because you're more maniacal than me
First, let's strip the Rshare formula from the non essential. We will not consider bonuses and starter cards. Rating is fixed as we have reached our maximum. Let's say 1000, because it's easy, it's top of Bronze 1, and a rating of 1000 gives 1000 Rshares, how convenient! Then the rewards Rshares are just function of ECR.
Our maniacal man-bot who plays every 59 minutes will play 25 games with 100% ECR, for a maximal earning of 25 000 Rshares, if he wins all.
Someone playing a session from 100 to 75 ECR will play 29 matches, for a potential maximal Rshares of 25 030. He would have spent 25.03 ECR to do so.
Playing two sessions from 100 to 87 ECR every 12 hours will require to play 27 matches, for a potential maximal Rshares of 25 103, for 25.1 ECR spent.
Playing one session from 80 to 55 ECR will require to play 38 matches, for a potential maximal Rshares of 25 396, with 25.39 ECR spent.
Did you get the pattern?
It's all the same, reported to the ECR spent and our daily 25 allowance. You can play 1, 2, 3, 25 sessions: it's the same. Low ECR, high ECR: it's the same. Wanna play 100-50 ECR every 2 days? Be my guest, it's the same!
Well, as long as you're above 50% ECR and a match only costs you 1% of your current ECR.
Spending 0.99 ECR for 99% of the reward or spending 0.51 ECR for 51% of the reward: in the end it's the same. It's just more or less matches to consume our daily 25% ECR.
"Great", you think, "I've been clickbaited to something that doesn't matter!"
No, no, stay with me! Here comes the good stuff.
At the start of season
If ECR doesn't matter when rating is maxed out, how about when it's not?
We know Rshares depends on cubed rating, and at the start of the season we all want to climb as fast as possible to get better rewards.
Now, the rating in SPlinterlands is based on a modified ELO formula. Essentially, it's not a 0 sum ELO, but additionnal rating points are added, when the difference between opponents is significative or for every win streak. That means that you rating is very likely to increase, even with a 50% win rate or even slightly lower.
We will take here for the sake of simulation an average rating increase of 5 per match. Because it's what I observed on my account, with a winrate between 50 and 60%, my rating increased of between 5 and 7 per match. Anyway, it could be 1 per match, it doesn't matter: the tendency stays true, only the impact will be more or less important according to how you perform.
Here I come with my crazy excel tables: I simulated several scenarios of play to spend 25% ECR. For example:
- 1 daily session of 25% (100=>75% ECR, 90=>65%, 80=>55%)
- 2 daily sessions of 12.5% each (100=>87%, 90=>77%, and so on)
- 3 daily sessions of 8% each
- ultimate man-botting for the extremes (1 game when reaching 100% or 51% ECR)
Some tables for the example, hopefully they will be readable.
I won't put the data for all scenarios, it's not interesting, it's just numbers.
And for each scenario I will get the total potential Rshares won (if 100% winrate) divided by the total ECR spent. We will then be able to compare the "rentability" of ECR in each scenario.
Here come the results, presented as potential Rshares per ECR, by play sessions and by initial rating.
We can see that it is better, at the start of the season, to play with the lowest ECR possible. The most maniacal player should play once ECR reach 50,01%, every 29 minutes or so. It actually makes sense: by playing more games with less percentage of reward, your rating increases faster, and as it's cubed, it's worth it.
We can also notice that the lower your rating, the more impactful it is. The difference between the two extreme types of play is 40% with an initial rating of 300, and quickly diminishes with higher ratings. It's only 8% for a starting rating of 2000.
Keep also in mind this is the difference between the two extremes. In real life it will be inferior, but well, it's still 5-10% more to take. And you have to add the bonuses on top of that. Definitely a couple of chests more by the end of the day.
I hope you found this little study helpful. At least I was surprised with the result, but it makes sense when you think about it.
So I'm transitionning right now from high ECR play to low ECR play. The good news about it is that I won't have to worry about reaching 100% and wasting some ECR along the way. Like if I slept too much. Which happened more than once.
Any question, if something is not clear or you want to check my formulas, do not hesitate to comment!
That being said, see you on the battlefield!