Ludia Forums

A Bag of Beans: retsamerol’s guide to the Test of Might (December 2019)

**Word of caution: The current event trophy win loss algorithm makes it very difficult to complete this challenge if you try to get a good win rate for your first few matches. The key is losing your first 10 or more games. When I was winning the first few matches, I went 35-50 (spent 480 gems) and gained over 1,000 event trophies from 3,000 to ~4,000 trophies, where I ended up facing level 20 bots and the top players in the game. When I lost 10 matches immediately at the start, I ended up going 30-12. Lose early lose often is the winning strategy here. **

If you want to subject yourself to this silly exercise, I have the following observations to share.

Event Matchmaking Rating (MMR): This is a hidden value that can only be estimated based on the difficulty of the bots and their stated trophy level. This rating is what is used to determine matchmaking.

AI will be within 200 trophies of your current rating for the event. At 3k-3.5k, AI will start at level 16; at 3.5k-4k level 18, and 4k+ level 19/20.

Your initial matchmaking rating is determined by your highest trophy count in the current season.

Neither your own nor a human opponent’s stated trophy count is informative because it only reflects their current Battle mode trophy count.

Note: for heroic adventures, the trophy count for the event resets between different stages.

Trophy gain/loss: Based on extensive testing, I have determined that while matchmaking is based on event trophy count, the event trophy win/loss calculation is based on the visible Battle mode trophy counts.

This necessarily creates two positive feedback loops:

  • Harder feedback loop: If you face opponents with a displayed battle mode trophy of 200 or higher than your displayed battle mode trophy count, then your event trophy count will gain 50 points on a win and 10 points on a loss. This makes it very easy to increase your event trophy count and difficult to decrease it, which will just make your matches difficult and will cause you to have a terrible win rate.
  • Easier feedback loop: If you face opponents with a displayed battle mode trophy of 200 or fewer than your displayed battle mode trophy count, then your event trophy count will gain 50 points on a win and 10 points on a loss. This makes it very easy to decrease your event trophy count and difficult to increase it, which will make your matches easier and will cause you to have a decent win rate overall.

Assuming your current battle mode trophy count is the same as your season max trophy count, earlier wins and losses have a huge impact on how the remainder of your Test of Might run goes:

  • If you win a lot early, you push yourself to the Harder feedback loop and will hurt your overall win rate.
  • If you lose a lot early, you push yourself into the Easier feedback loop, and this will counterintuitively improve your win rate for the remainder of the run.

This formula is especially punitive to players who have dropped 400 or more trophies in battle mode. You will start the event at your max season trophy, be matched against other players at that trophy count, and if they are more than 200 battle mode trophies higher than you, then you will be losing 10 trophies per loss and gaining 50 trophies per win, setting you up to have a very difficult time.

EDIT: the following should be struck out --there is no reason to think that the event uses a different algorithm than normal to calculate trophy gain/loss, i.e. 30 for same level, at least 10 and at most 50, depending on trophy count differential. There does seem to be losses, as I was bumped from the 4k+ category to the 3.5k category after a long losing streak.

The problem is that this formula leads to perverse results when you have limited pools of players to match against, which is especially at higher levels. Consider, for example, the simple situation of two players, being 100 trophies apart, but with different tiers of gear quality, such that one player will win 80% of the time. It would take 5 to 7 losses to separate the two players before they are outside the 200 trophy match. If the lower tier player actually wins one, then they gain a lot of trophies and gets pushed up within range again.

For this event, a 40% win rate shouldn’t translate into a 1,000 trophy gain. Rather it should be approximately a 450 point loss. Trophies are not your friend because they push you into harder AI and player territory.

Unless the developers publicly announce a change to how Test of Might is run in the future, I recommend avoiding the event.–

Modified matchmaking: The mode does seem to direct players to AI after two losses in a row. The AI is where I picked up most of my wins, although they were usually against a level 18 group.

Similarly, after two wins/loss against the same player the matchmaking algorithm prevents a third rematch.

Interactions with Battle mode: There’s some interesting interactions worth mentioning.

Arena and Test of Might players are in the same pool: the matchmaking algorithm takes a Test of Might’s player’s event trophy count, and will attempt to match against another player with 200 trophies in either who has queued in battle mode (based on their season trophy count) or Test of Might (based on hidden event trophy count).

This means that even if you queue into battle mode, you may encounter a player who is more than 200 trophies higher or lower than you, even though you didn’t opt to wait. That means their hidden event trophy count is within 200 of your current battle rank. Consider conceding because a loss costs the Test of Might player 10 gems and their time, whereas a loss only costs you time.

Win/loss streaks carry over between modes: If you have a win streak (4+ wins) in battle mode and then switch to Test of Might, if you meet an AI, it will be several levels higher. If you have a losing streak (4+ losses) in battle mode and then switch to a Test of Might, if you meet an AI, it will be several levels lower. This was the only way I could make any headway once I got to 4k trophies for the event.

Recommended approach: skip the event.

If you have to complete it: Before you start your event, make sure your battle mode trophy count is close to your season max. When a stage requires 20 or more battles to complete, tank your first 8-10 matches in the event (this will cost you gems/gold). I call this the “Lose Early; Lose Often” strategy. I have successfully used this strategy for two Heroic Adventure Test of Might’s stage 3 and it has been very effective.

4 Likes

Thanks for the analysis.

Given you’ve found its not worth it - do you think it’s not even worth it to get to the first gem prize, or to some other milestone?

Ps love the topic of your guides in general, but this one in particular :slight_smile:

1 Like

I think the 5,000 gold breakpoint was okay because I think I spent 60 or 90 gems to get there. After that it’s a pretty miserable effort to reward ratio.

1 Like

Assuming that the trophy gain/loss issue is not resolved, and that gain/loss for the event mode is based on Battle mode trophy score alone, I will be using the following strategy for the Test of Might in the upcoming Heroic Adventure:

  • Play battle mode until I reach my current season max (3,000)
  • Play one match in test of might to lock in my event trophies
  • Play battle mode until I reach my previous season max (3,800)
  • Play Test of Might mode.

If my assumption is correct, I would minimize the amount of trophies per win in the Test of Might event mode, allowing me to play additional matches at lower levels.

Unfortunately, I don’t know whether the two Test of Might sub events use independent trophy counts.

Just for a different perspective: I don’t have lots of gems but I do have a lifestyle that allows plenty of act-and-wait type activities. So - me hitting the “wait” button instead of fighting a bot is very reasonable while I make a cuppa or do something else.

Across the ToM event period, I waited for 19 humans, lost 2 and won 17. I had 2 matches against players who were far weaker, and 4 matches against players much stronger. The majority were considered “fair” to me - I just didn’t have enough time in the day to keep going. I’m not sure if I’d have spent gems if I’d lost the 3rd match.

None of this experience conflicts with your experience at all, particularly as my record keeping was abysmal - I do not have any meaningful data to present beyond the above anecdotal observations - but it does suggest that avoiding bots, if you’ve got the time, might be less frustrating than what you’ve been through.

Thanks for your input Franam. I found my early record for Test of Might similar to yours. My record was initially 10 wins to 6 losses.

The problem was that as I progressed in the event, the difficulty increased extremely fast and my win rate for the remainder of the event ended up being 25:44.

It looks like the Heroic Adventure event carries over trophy counts between different Tests of Might.

Moreover, the first Test of Might provides you with unlimited tries, whereas the later ones require gem purchases for 3 failures.

So the optimal strategy is to tank a few hundred matches of the first Tests of Might before continuing on.

Yeah, looks like tanking fights the first stage is the way to go if you actually plan on doing Heroic Adventures or Test of Might in the future. This is kind of the last straw for me. The time investment alone isn’t worth it for the PvP events. Now that they implemented a way to game the new trophy system, it’s probably in my best interest to completely ignore PvP events.

We will see another Test of Might on December 24.

If you intend to finish the event, Ludia hasn’t fixed their algorithm and don’t mind paying gems, I recommend deliberately losing your first 7-8 games.

This will lock you into the win 10 trophies, lose 50 trophies range and the remainder of the event should be a breeze even with an exceptional win rate.

Note: tanking your Battle mode trophy count will make the event extremely difficult for yourself because you will be earning 50 trophies for a win and lose 10 trophies for a loss.

1 Like

only need to do this in stage 1?
@retsamerol

Doesn’t matter, the rewards are so insignificant that I can’t imagine anyone spending the time necessary to actually do it.

1 Like

Lol “fixed their algorithm”

Tested it between each stage the tanking reset. It did work. But second stage I was back with high ranked

Based on my observations from the previous event, your event trophy count carries over between stages.

I don’t know if they updated this.

Just tested it tanked a lot on first stage I started with 3500 opponent at the end of first stage was 2500 second stage bam 3500 again

And as I a wrote this i faced you in tom

Thanks for testing that out Krom. It’s useful to know.

Auto correct screwed me had to fix a few words