Near the end of last season, I made a post outlining a plan I had to rigorously test the quality and value of XBOT and ArchMage, the two leading services for botting an account in the Wild league.
One of the two seasons of testing is now over, and I want to write a brief interlude covering two main things. First, is some housekeeping I neglected to discuss last time. Second, is a quick overview of how XBOT fared in the first season and some thoughts on how I think ArchMage will fare next season and what the final outcome will be. This includes a specific battle where I was really impressed with how XBOT used some of the new rebellion cards.
Some Housekeeping
Last time, I went over what I thought were all the important points about the comparison, but I realize I left a few out, and want to make sure I state them aloud.
What is the point of this? Isn't XBOT the best since they have most of the leaderboard spots every season?
One of the critiques I got in setting this up was that this isn't really needed. The measure most used by the community is largely how many of the top leaderboard slots each service has, and XBOT has handily won this for months. So shouldn't we just say XBOT is the best?
The argument here is there are plenty of reasons why the leaderboard metric might not be ideal. You're comparing across, instead of within accounts. It's possible that ArchMage is actually better, but people with better collections use XBOT so XBOT has more leaderboard spots. (I don't think this is likely, but it's hard to rule out when solely looking at leaderboards).
More relevant, the leaderboard is only relevant if you have a pretty complete Wild deck (i.e., access to most every card available). People, like me, with a complete CL deck and a smattering of other cards, might not actually learn a lot of the bots' performance for our deck given the leaderboard. So this exercise will show me which service works best for my deck, which may not be the service that works best for the leaderboard.
So hopefully this shows why this approach, comparing both services on my own, very much not-leaderboard-quality, account.
Keeping staked SPS constant
One thing I've been doing for reinvestment purposes but realized is actually crucial to the fairness of this experiment, is keeping my staked SPS at the same level. The SPS rewards and how many chests I get are proportional to how much staked SPS I have. If I let it accumulate over a month, the service that goes second will look better because I have more SPS staked, not because it performed better.
So I've been unstaking SPS each week and going back down to 100k every Saturday, which will prevent this from being an issue.
Keeping the deck constant
This has been hard. I've wanted to buy some cards the past few weeks, but any changes in my deck will improve the performance, again benefitting the service that goes second. If you do this experiment yourself, make sure your deck is the same over the entire course of the experiment.
The Results of XBOT
I ran XBOT first, and saw pretty solid results. As a reminder, the outcomes I was interested in are: total rewards earned, total rewards netted (after bot fees), maximum rating, and performance with Rebellion cards.
XBOT Rewards Earned
During the season, I earned 457 SPS, 30 Diamond chests, and 98 Champion chests (including season chests). Using EV for the chests and SPS at its current value, this results in $47.67 total rewards. Taking out the 5% from SPS earnings and daily focus chests, the net rewards are $46.60. Out of 365 total battles, XBOT used Rebellion cards (mostly Grimbardum Smith) in 53 (15%) of them, winning 36 (66%) of those.
Although I ended at the very top of Champ II, I did make it into Champ I a few times during the last week.
Overall, XBOT performed well this season. It didn't blow me away, but I was really happy with it.
Showing off XBOT's position-based teambuilding
One thing I was particularly impressed by was the way XBOT built some teams. One focus of the Rebellion set has been positioning. Previously, outside of what card to put first and maybe second, there was not a ton of positioning needed for most battles. Abilities like weapons trainer and lookout have really changed the math on this. Importantly, this makes a bot's job harder, as it now not only needs to choose the correct team, but all the right positions, as well. In combinatorics terms, order now matter, and means the bot needs to choose from more possible teams.
This battle places a card with lookout (Toledo) adjacent to a card with taunt (Grimbardun smith with panda). Lookout (in addition to reducing ambush damage), reduces snipe, sneak, and opportunity damage by 1. Normally, those abilities might all target different cards. But with taunt, they all hit the taunting card, so it reduces all those attacks by 1. When the opponent goes in on a sneaky strategy, this can mitigate a ton of damage during the battle.
I was really impressed to see XBOT do this kind of lineup, and makes me pretty optimistic about the ability of bots to do quality teams even as the rules for splinterlands get more complicated.
My prediction
Based on this, and my previous experience with ArchMage (this is my first time using XBOT but I've used ArchMage the past year), I expect the two services will be similar in net rewards (and within whatever margin of error this experiment has). They seem to perform similarly (we'll see in more detail in two weeks how similarly), and because SPS is so inexpensive, the difference in fees doesn't equate to a huge difference in net earnings.
However, that is true for now. Any jump in SPS earnings will see the difference in fees matter much more, including all the SPS you earned in the past. So, if the two services are similar, I will likely default to XBOT as it is a more future-proof service if we do go through another bull cycle. I also like the layout and look of the XBOT config more than the ArchMage one. It's a stylistic preference, but I can't pretend it's not there.
Referral
Interested in the game? Want to try out a bot service? Use my referral link below, and I'll send you a few cards to get started.
Congratulations @badrag! You have completed the following achievement on the Hive blockchain And have been rewarded with New badge(s)
Your next target is to reach 80 posts.
You can view your badges on your board and compare yourself to others in the Ranking
If you no longer want to receive notifications, reply to this comment with the word
STOPTo support your work, I also upvoted your post!
Check out our last posts:
My collection is almost full RW, but only 50% leveld up in RB. and I think I got a win rate of 51% on average with Archmage.
Winrate unfortunately is a pretty useless metric. In a ranking system like this, you basically progress until you hit ~50% win rate. You'd get the same win rate whether you are in Bronze or champion. Over time, every account and every bot service will get to around 50% (if it was higher, you'd go up in ranks until it dropped back down to there). So I show my win rate mostly to show that it did as expected, but win rate itself is not really an outcome of interest.
With that I agree.
That's a very interesting project that you are undertaking. It will be interesting to compare the results with Archmage. I believe the best metric for comparaison should be the income from battles, since there is a random factor included in the quality of reward chests.
I have picked this post on behalf of the @OurPick project which will be highlighted in the next post!
Please consider voting for our Liotes HIVE Witness. Thank you!
Checkout our BDVoter Daily Hive Showcase & Participate into our Daily giveaway to win various prize.
Thanks for sharing! - @zallin
Good Work, thanks for sharing!