either i am lucky or you are unlucky, the revives on enemy TDS for me seem like rather the actual 25%, at least that impression my “biased” memory gives me
but then, i play with dragon soul 3rd trait locked (not using it) so maybe i am indeed lucky to compensate that? 
Happened to me too, quite recently. 
IK also recharges faster than TDS. And Abynissia can summon more than one!
…In fact I’d like my darling dragon to respawn a little more often…
That’s awesome tactics, only I don’t have anyone with stun in my current team.
There are 24 team slots for a reason, though.
The only place where you cannot change teams between battles is the Arena, and Dragon Soul doesn’t hunt there.
Please don’t think I’m being rude. Just genuinely curious how this is?
TDS uses less mana and it’s spell gives it mana back. Plus most teams that use TDS uses +2 Purple /+1 Red.
IK has a hire cost, doesn’t create its own mana and from what I’ve seen, isn’t normally helped by the banner.
As I said, not trying to be rude. Really want to know why you feel this way 
Abynissia already requires yellow and purple and usually has a demon team, so I’d be surprised to meet her under a banner that doesn’t give purple mana.
All the transforming IK does also creates quite a kinda explosion.
And it does fill fast
Thank you 
I don’t really use IK or face him.
Isn’t fun how many styles of playing there are, and how many more if we weren’t forced do play so much anti-meta?
I have won battles due to my TDS respawning multiple times… I also love killing TDS with kraken and he respawns only to immediately be devoured lol
Resurrection has a very high chance (over 95%) of being non-random.
Here’s a standard randomness disclaimer: you can always be subject to a rare “localized spike”.
It’s not impossible to lose a coin toss 4 or 5 times in a row. It’s why 50% odds aren’t really that good. That doesn’t make it feel good when you lose that coin toss “too many” times.
One time I had TDS respawn on me 3 times. That was infuriating. But I’ve never seen that happen again. The odds of it were very low. I don’t really know how often TDS is respawning, but aside from that one bout of “luck” I don’t feel it’s too far out of the 25% bounds.
I think there’s a lot wrong with TDS, but I don’t think the RNG has significant bias outside the 25% it should be hitting.
That said, hitting 3 respawns is so dang aggravating it wouldn’t upset me if the devs nerfed it such that each respawn halves the probability, or made it impossible to respawn twice. That might be fun on offense, but it’s a giant middle finger on defense.
Fun fact: Respawn is technically just “summon self with % chance.” To implement “memory” around resurrection chances would probably impact all summon-on-death traits, which are currently stateless (get random number, compare to threshold).
I did. On my subjective level, my DTS respawn less but I’ve noticed that the AI’s respawn quite more often than what I perceive as “usual”.
On the other side my Kraken devours a lot, respective as the previous “usual”.
On the other other side, looping now loops a lot under whatever I perceived recently. I don’t understand why we have armies of spammer troops and spamming is actively sabotaged. In my perception.
On quite anoter set of arms, TDS respawning can be a disaster as long as his team is still active. A TDS respawning alone has a very short life expectancy.
Mine very rarely respawn yet the ai often respawns his ds 2-5 times I find. It’s very annoying, especially in gw.
The test very clearly shows that there is something else there. IDK what it is but it’s there. It may be linked to some other non-random or even random event in the match. I can check with a simulation of two random dependent events. It still should be random but chance of it being non-random might be higher. Although it should not be THAT high.
The spikes of randomness have specific distribution of their own characterized by z-stat parameter of the Runs test. This is what the test had actually checked and the math behind it is quite simple. So, the test clearly shows that the hypothesis about random and independent resurrection is failing 95% of the time spikes and fluctuations included.
We’re just going to have to agree to disagree on this one, I’m afraid.
It is not about agreeing or disagreeing. It is about 2+2=4 or? 5? 3? 99? It is also about denying the obvious.
There should be some easy and simple explanation. Although even if it is found, it does not influence repercussions.
In my experience, the respawn is merely a nuisance late-game, only a serious problem early-game. However, I don’t play GW and I can see how those matches would be much closer, where a respawn in a close match means the difference between winning and losing.
I’m sorry but from my anecdotal experiences with gambling I believe 75-100 samples is small enough you can collect a data set that suggests bias where there is none. The larger you make the data set, the less skeptical I’ll become. At 1,000 I’d raise an eyebrow, and at 10,000 I’d be convinced.
For example, I played a coin toss game in Kingdom of Loathing and had lifetime losses in the millions. A few lucky streaks I lost 17 times in a row. I got so mad I wrote a simulator to show off just how rare that should be. I had stats on my total games, so it was easy to simulate my “life”. The only thing the simulation taught me is I was actually on the low-end of 17-loss streaks, and probably should’ve expected a couple more in the thousands of games I played.
But if I narrowed my window to just 100 of those games, I could get dramatically different results. One time I was on a 35-win streak. Another time I had 2 17-losses within 50 plays. If my 100-sample window included those oddities, I could conclude there was a strange damn bias in the game, with great stats showing how confident it made me.
This is how bad gamblers lose a lot of money. Instead of looking at the largest possible data sets, they fixate on small windows and try to find patterns.
So I really think you need to come back with a 5,000 observation sample set. It’s "not probable’ that 100 had a severe “bad luck” streak in it, but it is “possible” and it will screw up your data. The size of ‘not probable’ for 5,000 samples is dramatically smaller, and that change is far more significant than the jump from 10 to 100.
My opinions are that:
