(More amazing work from Delpez, and just one example of the depth and number of interesting decisions CoC has. The main comment I have is that more data is needed overall, which we will collect as more wars happen, but even with what we have here, our clan can use this info to improve, which is what I love most about the work done here and the game in general.)
At TH8 dragon attacks rule. Sure you see some Hog or Pekka attacks at the higher end, but for the most part dragon attacks determine who wins this crucial battlefield. Since I’ve reached TH8 I became more interested in the trends behind these attacks (if any), and I’ve been gathering data from the last two wars. There are a couple of caveats, however:
Unlike the overall war stats I collect from time to time, there is more subjectivity in these numbers. For instance, one of the parameters I’m tracking is whether a base was build against dragons specifically (Anti Dragon). In some instances this is very clear (see Nickio and Jonneh’s bases from the previous war), but there are also some bases where I had to use a judgement call.
Also, although I’m trying to reduce the dragon attacks to a number of parameters, the big missing is player skill. A skilful player can get better results with level 2 dragons than an ineffectual attack using level 3 dragons. These types of effects should cancel out with large enough data sets.
Finally, although I’m looking at a number of parameters, I’m not looking at combinations yet. For instance, balloons and rage spells in combination might have very different stats than when viewed in isolation. However, calculating all possible combinations is a huge undertaking and something I’m trying to tackle.
Also a word on the usefulness of these numbers. There is definitely an element of chance in these high level dragon attacks. Numerous attacks hinged on extremely small margins – sometimes an almost-dead AD will take out the last dragon attacking it, and the result is a 1-star. Other times the dragon takes out the AD and it’s a 3-star. No matter how perfectly you execute your attack, there are too many unknowns to be certain of the result: defending CC, location and level of traps, dragon (or balloon) pathing in 50/50 situations etc. Basically what I’m saying is that the aim of this analysis is not to determine a cast-in-iron strategy to succeed with dragon attacks, but to play the percentages. By focusing on tactics with better numbers, we will improve our margins, rather than solve high-end dragon attacks.