Due to certain things related to Fandom, I have to use a different alias, but some people may know me as Wirxaw from Ludia Forums and Discord. This compendium is based on my old guide , which is no longer editable and is generally outdated. However, this is a disclaimer that I am not an impersonator and no data has been stolen.
Opening notes and edit comments
Throughout the guide I will use the following way of describing the dragons:
Xstar and X-star. Xstar means original dragon quality, like a 5star Toothless. X-star means trained level, like a 3-star Toothless.
Also, some abbreviations:
ATK - attack, DEF - defense, HP - health. AoE - team-wide damage\effect. DoT\HoT - Damage\healing over time.
Buff - positive effect. Debuff - negative effect. 

All of the dragon data is obtained from game, wiki and Szaddaj's spreadsheet .

A few words for the beginners

-I would not recommend to spend scales to hatch color-specific dragons, unless a duty asks you to. In the early game, you really should save the scales and spend them to only train the dragons you are certain to keep.

-Energy is your most valuable resource. Energy is XP, XP is progress, and your goal is to hit level 16 and unlock Breedery, and then to push further for more maximum energy and Hatchery\Breedery upgrades. Failing a very difficult quest, desperately trying to beat that story level would only set you back. 
However, you should try to push story Alphas whenever you can. This is because your repetitive Egg Shell quest will simply yield greater rewards, and chances are you are going to do it often.

-A lot of players used to ask me how to best spend their runes, and I've always given one answer: it's either gambling or convenience.
If you don't mind donating - you'll draft some dragons that could or could not be useful. In this case, I would at least recommend waiting for an event or a weekend specific draft. Anything that could increase the odds of getting the dragons you need.
However, if runes are limited, then saving up for that "10+1 4000 rune draft" is extremely unadvised. Because you can strike gold, or you can get 10 3star dragons and a 4star you don't need. If these 4k runes were your last - such bad luck could have a more adverse effect than even the best "good" luck. You can always invest into resets for event dragons, you can invest into revives to be able to do a level 4-5 scale quest, you can buy energy if you ever find yourself unable to do everything. You can speed up breeding or hatching if you have to go shortly and can't wait for some 30-40 minutes. You can revive on a very annoying story level. All kinds of conveniences that would keep your sanity and let you resume playing, instead of dropping the game in a fit of rage.
Or you can double-click on a Basic Draft when you get a free one and end up losing 1000 runes for a few worthless dragons. Yes, that actually happens, for a long time I believed that being careful is enough. It's not. Legends say that Ludia used to refund 500 runes ONCE without taking away the dragons(because they can't), but ... from my experience - they no longer bother. You have been warned.

Duty management
Micromanaging duties can yield you almost 80 runes a day, among other bonuses. Always pick the Hard chest, it gives the best rewards.

-Among the 4 duty characters, the most important ones are Hiccup and Astrid. They both give combat-related duties, so it's ideal to synchonize them so that you would get credit to both of them. 

-Hiccup's "Complete 3 duties" duty is especially interesting. If you are about to complete Hiccup's duty, the next one can always be "complete 3 duties". It can't happen twice in a row, but it can happen after any other duty. If you have 6\7 of any Astrid duty, or are able to do an easy Fishleg's quest or have Valka's breeding ready - do not always instantly pick the rewards, always keep them ready to be "collected" for Hiccup's duty.

-You CAN get credit towards Astrid's "Use 35 Special Tiles" duty in Clan Alpha fights. It won't be much, and you could probably farm the tiles with low-power dragons anyway, but anything that saves energy is worthwhile.
"Use X abilities of Y color" can also be done in Clan Alpha fights, but it's easier to just pick 3+ dragons of the same color and do some 3-energy first level of a repeatable quest.

-Daily reroll should be used primarily on Fishlegs' "Level 10 dragons" duty or "Hatch 10 eggs", depending on whether you have a shortage of fish or shells. However, if you are in a low-level location and your zone quest doesn't give enough shells\gold\fish, you could also reroll "Gather 15k gold\30k fish" duties if it takes too long until the next 4-hour resource quest.

-Always try to have around 100 shells to be able to queue them for "Hatch 10 eggs" quest. In fact, if you are about to run into the shell cap - try to shuffle through Fishlegs' duties until you, hopefully, get the 10-egg hatching duty, or at least a specific-color hatching duty.

-"Level 10 dragons" duty does NOT require you to level 10 separate dragons. In fact, the easiest way to do this duty is to feed the lowest possible dragon as few dragons as possible and only enough to level up, and do it 10 times. You can later feed this dragon to any of your higher dragons for extra experience.

Dragon Training
While it is a well-known fact that if you consume as many dragons at once as possible, they all will cost only as much fish per dragon as the current level requires. However, in practice - you'll be saving... scraps, and you CAN run into a max level dragon with a non-maxed ability.

This isn't the end, you can still feed your dragon without worrying about XP bar and you'll still have the same %upgrade per dragon consumed, but the cost per dragon will be locked to your maxed out dragon. In other words - it can be VERY expensive. Especially if you are missing 3-4 ability levels.

The imporant part in training dragons - is upgrade% per XP, and the highest ratio comes from consuming 1star dragons. Well, technically the highest raio comes from feeding a copy, which always gives 25% upgrade chance and costs 50% of normal fish per dragon, but copies are scarce for 1-2star dragons, they are needed for 3-star training of 3star dragons and they are almost mythical for 4-5star dragons, so I wouldn't base my methods on anything unreliable.
The highest total %upgrade when leveling your dragon from 1-star 0 to 1-star 30 with 1star dragons is about 120%. That's 4 times feeding 10 1star dragons at once, 4x30%, each feeding gives you about 7.5 levels. Let me put it this way - you can fail every single time. You can fail if you feed 3 copies and have 75% chance. Or.. you can succeed even with 3% chance. What's the golden middle? About 10% chance per feeding session, which is about 2 levels trained at a time. In fact, it's suitable to train your dragons like that for "Level 10 dragons" duty. This was originally suggested by the Senior Rider ToothlessJake.

Obviously, you should only consume the 1star dragons of the SAME COLOR. Same color dragons give 3% for 1star, 4% for 2star, 6% for 3star, different color dragons give exactly HALF of that. Yes, that be rounded up or down when your first two 1star dragons of the different color give 2% upgrade total, and the third - 4% total.

If you feel like this method doesn't work, remember that it all comes down to RNG. You can be unlucky with most ideal training, or you can be lucky with the least ideal. The whole purpose of this method is to just... tweak the RNG in your favor, as much as it could be possible.

Breeding and Hybrids
Unfortunately, nobody knows the exact breeding percentages, not even the dataminers, and the developers are mum on the subject. The only more or less confirmed thing about breeding is that it's NOT affected by training level. A freshly trained hatchling has the same breeding capabilities as a maxed out dragon.

There is a player, who wanted to research this subject further, you can check out his thread if you want.

2star2star1star\2star\3star2hours4000 Fish
2star3star2star\3star3hours 6000 Fish
3star3star2star\3star\4star4hours 8000 Fish
3star4star2star\3star\4star10hours12000 Fish
4star4star2star\3star\4star\5star16hours16000 Fish
4star5star3star\4star\5star20hours20000 Fish
5star5star3star\4star\5star24hours24000 Fish

Breeding two standard dragons of the same species will only result in the star-matching dragons of these species.
Breeding two different suitable species will result in both of these species' star-matching dragons, as well as a SMALLER chance for any star-matching hybrid.
Breeding two hybrids of the same species will result in ONLY the star-matching hybrids, this is the only guaranteed way to get a higher chance to breed a hybrid.

What makes a hybrid different?
Hybrids are statistically more powerful than normal dragons. That is - they have marginally higher ATK\DEF\HP parameters. 
Hybrids have 2 effects per ability for 1star species and 3 effects for 3star species.  
Hybrids sometimes have unique abilities, like a chance to revive your fallen dragons, or stat-budget of their abilities can be higher, like up to 185% single target damage vs 180%.

However, not all hybrids are absolutely superior, as I will explain in the further section.

Dragon abilities
It's important to understand what the dragons do, before learning about the species and combat. In this section I will try to explain all the currently existing effects in detail. Please refer to the opening notes for any abbreviation questions.
1. Damage
Damage can be single target or AoE, instant or over-time, ricochet, random-blasts or lifesteal.
Lifesteal damage gives a percentage of damage back as health, this effect can be AoE. 
Ricochet deals the SAME damage to two targets, but, if the ability has a single target effect, only the FIRST target is affected. 
Random-blasts damage is currently unique to two event dragons, Lightfury and Hookfang. It is indeed random, however, you can select a dragon to guarantee that at least one blast will hit it. Multiple blasts can hit the same target and these dragons are particularly useful for single target damage.

Single target damage is useful against Alphas and in the deep endgame. The highest percentage of single target damage is 325% with 5x65% blasts by Lightfury, second highest - 320% with 4x80% blasts by Hookfang, and third+ highest is 250% by Barf&Belch as the normal single target blast.

AoE damage is preferable for any mid-endgame content, particularly repeatable quests, because it allows you to clear numerous waves of dragons faster and\or easier.
Ricochet damage can go up to 380% as 2x190% by Toothless and Skullcrusher, with normal dragons doing up to 340% as 2x170%.
AoE damage can go up to 900%(5x180%) by Gritty Sawmaw, 1050%(5x210%) by the bugged Mythic Murmurquill, 850%(5x170%)DoT by Stormfly, or more commonly, up to 750%(5x150%) by direct AoE damage dragons.

Also, damage can be standard and ability-based. Standard damage is your tile damage and enemy blasts. It is affected by the color pattern that can be seen on your team's setup page. Strong-color damage is orange damage and it is the ONLY damage that CAN crit(I don't know the chance, but it seems to be around 15%). I am not absolutely positive about it, but I assume that strong-color damage deals 150% normal damage, while a crit deals 200%. Weak-color damage is the grey damage, and it deals about 50% of normal damage.
For example, a blue tile\dragon would deal anti-color damage damage to a red dragon. A red tile\dragon would deal weak-color damage to a blue dragon.
Ability damage is not affected by element colors. However, both types of damage are affected by ATK value of the attacking dragon and DEF value of the target dragon.

2. Healing can be single target or AoE, instant or over-time and it be a bonus health, which I'll call "shield".
Standard standard healing can be blocked by anti-heal debuff, while the "shield" can not. More on the importance of damage and healing in the next section.

3. Anti-heal debuff, as I said earlier, blocks the standard healing. For players - the only real value of anti-heal lies in deep-endgame levels, where enemy dragons can have 2x+ health of your dragons, therefore allowing them to execute almost any kind of heal will probably result in a failure. Until then - you are far better off optimizing your damage than trying to prevent healing, particularly because shield healing can not be blocked.

4. Ability block or "silence". It is the only way to prevent a shield... from basically being cast. Very few dragons have this ability and only one dragon, Shifty Murklurker, has it as AoE. All of these dragons are extremely valuable allies... and extremely dangerous opponents.

5. Strong Hit is an effect that causes all affected dragons to deal strong-color damage, which includes the ability to crit. By itself it's not very powerful, but when it is AoE and used with Counterattack or with numerous special tile activations\combos in the same turn - it can be devastating.

6. Counterattack
Counterattack or CA is a very peculiar effect that many players might misunderstand, because it differs greatly between being used by a player or by an opponent.
When an opponent  has a CA, it will shoot an extra standard blast at a random dragon if it was damaged by a tile during that turn. If enemies had AoE CA and multiple dragons were damaged - each will add an extra random blast to their normal damage rotation. Like all standard damage, CA blasts can deal strong or weak color damage and CAN crit. YOUR ability damage does NOT trigger enemy CA. Also, each enemy can only trigger CA ONCE per turn.
Essentially that means that abilities are the only safe way to deal with CA and that damaging multiple CA dragons with tiles is a great way to fail, whereas damaging multiple CA dragons that also have Strong Hit - is a great way to suicide.

When you use CA, there are are few differences:
-Enemy abilities CAN trigger CA, but they have to be direct. DoT abilities will not trigger CA. Also, while standard blasts do trigger CA, enemy CA blasts do NOT trigger your own CA.
-Your CA triggers only when a dragon with CA is hit. Which makes single-target CA fairly worthless, and only AoE CA to be worth considering.
-Your CA blast damage is... much weaker than enemy's. I don't have concrete values, but if, say, your dragon's ATK was equal to the minimum 3-tile combo hitting the same enemy dragon, then your CA blast is about 1-tile worth of damage.
-AoE damage triggers CA from all of your affected dragons.
-Your CA can trigger for EVERY enemy dragon that hits your corresponding dragon. Which means that, coupled with the previous point, if multiple enemies use AoE abilities against your team in the same turn - all your affected dragons will launch their CA blasts multiple times, however, they will only target the dragon that used the ability. In other words, if enemy CA blasts are random, your dragons use CAs directly against enemies that attacked them.
-While this also pertains to the enemy CA dragons: FALLEN DRAGONS DO NOT CA. So if you manage to down an enemy while he had CA, or an enemy finishes off your CA dragon - that dragon will not counterattack.

In a nutshell, this means that enemy CA is one of the most dangerous effects to fight against, though not as dangerous as healing\ability block, while your own CA is fairly low on damage values even when its used on your entire team. The niche situation where you reflect multiple enemy AoE abilities is.... still hardly better than a dedicated AoE ability, and it's generally bad for your health to allow numerous enemies to use AoE abilities.

7. Accuracy debuff affects only your tiles and enemy blasts. However, accuracy debuff can prevent strong-color damage, crits and CA blasts, therefore it's a great way to reduce the potential sniping or the pressure of a 5 dragon wave.
One thing to note is that your "missed" tiles will still give YOU the spirit, will still give ENEMIES the spirit and will STILL trigger enemy CA. Likewise, enemy missed blasts will still trigger your own CA.

8. Spirit modifiers can be instant and a buff\debuff over time. When they are instant, they either give your dragon\s the spirit immeidately and enable their abilities, or they drain enemy spirit, preventing them from using their ability next turn. Warcry is a good example of a tactical dragon that utilizes both.
Over-time spirit modifiers affect your tile spirit gain and enemy spirit generation from damage. They will not increase the effect of spirit drained from an instant offensive modifier, nor will they increase the spirit gained from your own instant spirit. In other words, these modifiers temporarely change the "charging rate" of an affected dragon. If Moderate is 100, then Very Slow is about 50% and Very Fast is 150%. Therefore a 25% modifier to a moderate dragon will likely make it charge similarly to a Fast\Slow charging dragon.

9. Damage done\taken modifiers are extremely important, because they can stack with each other and affect all types of damage. An AoE damage buff basically increases ALL your damage, all tiles, all dragons, for the duration. Likewise, if a single enemy target is debuffed, or the entire team is debuffed - all your damage is further amplified.
The % values amplify the damage numbers after ATK-DEF calculation and they stack multiplicatively. For example, if the effective damage value is 100, then 30% increased damage done would make it 130%. A 40% increased damage taken would further make it 182. More details on ATK-DEF will be added in the Combat Section.

10. Status removal effects can be both friendly(allies "clease") and hostile(foes "lose"), it shouldn't be too hard to distinguish them. The hostile effect can ONLY be AoE, however, the friendly effect can very often be single target. Some people make that mistake in case of Obsurdian, who has AoE hostile effect and only personal friendly effect. Skrillcrusher, on the other hand, has both AoE effects.
Some dragons can "steal" effects, which removes the effect from the target dragon and gives it to you. The duration of the stolen effect is refreshed to the duration that your "steal effect" shows("This Dragon Steals Beneficial Effects FromTarget Foe For 5 Turns"). This effect can ONLY be single target.

As far as the practical value of these effects go, it should be obvious that cleansing a damage taken or accuracy debuff, removing enemy healing or counterattack - are all good measures, if the fight has them and you can time your abilities.
Two things, however, need to be remembered:
-You CAN cleanse hostile damage-over-time effects, because they are a visible debuff.
-You can NOT remove shields with this effect, because shields are NOT a visible buff. That's right, shields ignore both the healing debuff, and cleansing debuff.

11. Two dragons, Knock-Rocket and Chestnut Knight, have a revival ability. I have neither dragons and from my experience fighting these dragons, I can say the following:
-It's random. KR can revive the entire team and CK can revive a single dragon.
-The health that the enemy dragons are revived with - is set as their maximum health. KR's value is 20%, CK - 50%. It may be a visual bug, and I haven't had the chance to revive my dragons, but if you see a full HP enemy - click it and be reassured that it's absolutely not full HP.
-Enemies, who were queued to attack during a turn when you kill them with tiles, and who are revived by this effect - still use their attack. However, they are revived with no spirit.

Dragon Species
While everyone starts with various combinations of dragons, whether it's normal 3-star dragons or drafted 4-5star, if you want to know what should you breed and what you shouldn't - this section is for you.

1. Snafflefangs
Snafflefangs are standalone species, which means that they can't breed with anyone else for a hybrid species. This species is the primary source of HoT effect, which most Snafflefangs have as AoE, only the 1star Slough Snafflefang has personal-only HoT.
Despite the importance of healing, Snafflefangs are not an ideal breeding goal. There is only one good Snafflefang - Brute-Wurst, and that's a 3star dragon, which means that you can get it from basic drafts, or you can realistically breed it with 2x Stocky Snafflefangs.
The 4star Fendmender seemingly improves on Brute-Wurst, losing the spirit recharge for AoE DoT and a shield for the weakest ally, but Fendmender fails at the most crucial part of what makes a healer - consistency. The Very Slow charge can result in him not charging in time, or not sustaining the the healing often enough, and this can lose you the fights that would otherwise be winnable with a lower Brute-Wurst. It also has the lowest %\sec healing value(3.875%) of ALL the healers.
The 5star Frosty Sparguard would be a great dragon, as it shares the highest healing\sec value with Brute-Wurst(5%) and it even has an instant spirit recharge. But.. the spirit recharge is single target, and 35% single vs 4 targets with 20% each - is definitely an undergrade. The added 25% attack vs Red dragons is a very situational bonus that doesn't even remotely justify the fact that this is a 5star dragon.
Verdict: If you have drafted Sparguard - be happy and use it. If you have drafted Fendmender - use it to breed for Brute-Wurst, because you aren't likely to get a copy of Fendmender again. And if you have Brute-Wurst - max and it be happy.

2. Typhoomerang
This other standalone species doesn't specialize on any effect, but it has two dragons worth of mention: 3star Torch and 5star Radiant Skyglow. Torch is one of the best 3star dragons, because it's the only 4> dragon that has AoE damage buff, and Radiant Skyglow has both an AoE damage taken reduction buff(and of whooping 45% value) and 25% AoE spirit recharge, plus a hefty AoE CA. Unfortunately, between these two outstanding dragons liees Skywarden, a rather mediocre dragon with slow charge, spirit buff(over time) for your dragons, AoE CA and AoE Strong Hit. Now, this is a very tactical dragon that can go far, but that can also go nowhere if you don't use it properly, and it occupies Brute-Wurst's yellow spot. In other words, if you don't benefit from spirit buff(that lasts only for 3 turns) and Strong Hit - this dragon will do you no good at all.
Verdict: Keep Torch around for as long as you can, if you draft Skywarden - you can try to breed for another copy and then breed for Skyglow, but it's a long shot.

3. Boneknapper
A morbid standalone species that specializes on blocking enemy healing. The 3star Bog-Blight is a rather hit and miss weak AoE DoT dragon with AoE anti-heal. Its 4star sibling, Skullcrown, changes the mediocre AoE for a solid single target burst and a hefty 40% shield for the weakest dragon, becoming much more tactically viable. The 5star Cryptic Collector is a nightmare to fight against and arguably the best 5star green dragon in the game. It has fast charing, average AoE damagebuff and a whooping 28% spirit recharge for your team, on top of still having AoE anti-heal. It's not a dragon to take against Alphas, but it's the dragon to take to the deep endgame to make sure enemies don't heal and your dragons have spirit. Verdict: You can breed for them, but I wouldn't make it a priority.

4. Thunderdrum
The last original standalone species revolves around impactful AoE effects. The 2star Tectonic Thunderdrum settles on ricochet damage, but has a very rare ability lock effect, making it an absolute early gamechanger, especially with its fast charging. 3star Warcry can't quite match Brute-Wurst's healing value, but the ability to drain enemy spirit servers as a soft AoE ability block, and a large 25% spirit recharge helps to keep your team active. 4star Waveshaker trades ability lock for incredible offensive capabilities. Solid AoE damage with AoE damage taken debuff and even a huge damage taken reduction for your weakest dragon - this dragon can both pack a punch and keep your team a step away from the grave.
5star Shifty Murklurker is probably the only god-tier dragon so far in the game, because it completely locks enemies away for 5 turns. If you can sustain his Slow charging spirit over these 5 turns - enemies do not even get to cast anything, and that's what you want in the deep endgame. It also has a solid single target burst with lifesteal, and it further slows down enemy spirit gain, even if they aren't about to cast an ability. If you fight against this dragon and let it cast its ability while it's not alone - most likely you are done for.

Verdict: You can breed these species freely, they are excellent at every tier.

5. Hotburple
This latest introduced standalone species focuses heavily on slacking and shield healing. The 2star Hearty Hotburple is a good self-sufficient combatant in the early game, 3star Bush-Wacker doesn't have any change that would justify its 3star status. The 4star Beachcomber is one of the few dragons to have AoE shield effect, and this makes it uniquely valuable. 5star Blistering Belcher isn't a particularly great 5star, because AoE damage reduction buff isn't as rare as AoE shield, but it is probably one of the best if not the best RED 5star, because the other red 5-star dragons are just that bad. Verdict: If you have nothing better to breed for - you can try to breed for 4star Beachcomber, it would be an above-average option against Alphas and for early midgame. Breeding for 5s Belcher is a long shot, and not as valuable as other breeding options.

6. Timberjack-Hideous Zippleback-Ghastly Zapplejack
I'll say this straight out - Timberjacks are the worst, most undertuned and worthless dragons in the entire game(and that's saying something). It goes for the ENTIRE species, from 1 to 5 stars. Do NOT breed for them, do NOT level them, unless you have absolutely no other options. The Zapplejack hybrids, in turn, have only ONE valuable dragon - 3star Anveil that you get from the tutorial. You can try to breed for its copy, but there are better breeding goals.
Zipplebacks are not too different. Only 4star Tricky Two-heads is one of the best dragons in the entire game, the rest of the species are worthless.
Verdict: You can try to breed for Tricky, but there is no point to breed for anything else in this family of species.

7. Stormcutter-Deadly Nadder-Deathly Galeslash
Stormcutters are a great supporting species, who mix damage and healing to offer the best of both worlds. 3star Tripfire is not an outstanding dragon, but it's a Fast-charging straight shooter that will keep your weakest dragons alive, and it's a decent inclusion for Alphas next to Torch. 4star Gloomleer is incredibly powerful, one of the best 4star dragons with high single target damage with self-heal, ability lock and an AoE damage taken buff. 5star Tempestuous Scalestorm combines some of the power of the lesser dragons, getting AoE damage, and retaining weakest dragon healing and AoE damage taken reduction.
Deadly Nadders are sort of mixed up species, their 2star Nimble Nadders are one of the best 2star dragons, an incredibly powerful Fast-charging, self-healing, large DoT-ing dragon that can be incredibly useful against Alphas. The 3star Pincher has a very powerful single target attack and a rare AoE shield effect, but its slow charge and low shield value make it just a losing case vs Brute-Wurst. Fireshrike is the only non-Snafflefang AoE healer, that's all there is to say about it - if you really need to drop your Brute-Wurst, it's either Fireshrike or nothing. 5s Lethal Lancebeak is just bad, a sad sniper that doesn't even get a decent charging speed.
Their hybrid progeny, Galeslashes, are similarly mixed up. 3star Fog-stalker retains Pincher's abilities(losing only a bit in damage), but also offersa damage done debuff for the target. It's not a big step up from Pincher, but there are no better 3star purples for standard battles. 4star Corpsekeep is basically an upgraded Torch with added self-healing and personal cleanse. The 5star Celestial Royalwing is an improved Lancebeak with a better third effect.

Verdict: You can safely breed any dragons from this family, the worst that you can get is Fireshrike and Lancebeak, and even that's not too bad.

8. Gronkle-Monstrous Nightmare-Abomibumble
Like Nadders, Gronkles are very mixed up. 3star Tuffnut Jr. is a nasty thing to fight against, but it's hardly a dragon I would recommend, because you wouldn't get it to charge up. 4star Obsurdian has a capability of denying enemies their buffs and is a decent alternative to Gloomleer for standard battles. 5star Molten Magmanette is a very weird dragon. Like Skywarden, in a very specific situation - it can both be an offensive dragon with good damage and a support with AoE damage buff and cleanse. In "some" deep endgame scenarios - it works. Everywhere else - it's a Very Slow charging slab of DoT-dependent damage and effects that could either be interchanged or ignored.
Monstrous Nightmares are not exactly as bad as Timberjacks, but they are in the same ballpark. The only remotely useful dragon of this species is a 3star Fanghook, due to its Very Fast charging, ricochet-capable attack and an average 35% damage taken debuff. It can be a sniper in standard fights, or it can be used to boost your team's damage in Alpha fights.
4star Charsoul is a mild annoyance to fight against and a useless damage bot in your team. 5star Blazing Phoenixfire is a somewhat reverse Magmanette. Fast charging with still AoE damage, a huge 50% spirit boost to a weakest ally and a hostile AoE cleanse. The ironic part is that it can restore its own spirit and end up refreshing its DoT on enemies without dealing its full value. The cleansing effect is very nice, but there are much better options among 5star green dragons.
The hybrid abomibumbles are "jack of all trades, master of none" kind of species. 3star Scally-Slander has an above average AoE 40% damage taken debuff, but the slow charge makes this dragon a bit unruly. 4star Gladgut is ruined by its Very Slow charge and only 18% AoE accuracy debuff, despite having a decent AoE damage buff. It's a safer bet than Skywarden for a 4star yellow, but it's not something to breed for. 5star Smoldering Slaggert is just a give-or-take buffed Gladgut with the same Very Slow charge. These dragons are semi-annoying to fight against, but they won't do too much for your own team.

Verdict: I wouldn't breed these species at all, unless you get Valka's duty for it.

9. Skrill-Rumblehorn-Brooding Boltstamper
The Skrill are yet another mixed up species. 3star Hide-Fryer is better than Fanghook against Alphas, and worse in standard battles. 4star Revenger is a more reliable Skywarden with 4 turns of spirit buff instead of 3 and solid single target damage instead of Strong Hit. And 5star Shrill Boltbeak is a semi-staple 5star red dragon with AoE damage, damage buff and CA, set back only by Slow charge.
Rumblehorns have one unique feature - they are...bulky. Their power values are engorged by a large percentage of HP, however, their ATK\DEF values aren't anything special. In terms of performance, these dragons have almost nothing in common. 3star Flank-Tanker is a very slow turtle that still manages to be useful for AoE healing. 4star Son of Skullcrusher is just a very sad sniper in your team, and a very annoying slab of health to fight against. 5star Valiant Scarbearer is probably the only reasonable replacement of Brute-Wurst, if you keep yellow spot for support. It has a large 45% healing for the weakest dragon, an AoE damage taken debuff and AoE hostile cleanse. Unlike its 4star sibling, the Scarbearer is very easy to fight against, but it's a much more capable dragon for your own team.
Boltstampers are a unique hybrid species that all boast special abilities. Both 3star Knock-Rocket and 5star Chestnut Knight have the ability to revive your dragons, and both have additional effects that further secure their value. 4star Skrillcrusher is a bit set back by slow charging, but it's a direct upgrade of an Obsurdian, this dragon deals good AoE damage and wipes the board off anything harmful to you or beneficial to the enemy.

Verdict: Both parent species have dragons that are good if they are drafted, but none of them should be specifically bred for. The hybrids, despite their uniquity, are simply unreliable. Knock-Rocket takes too much to max out, when there are pefectly accessible Torch\Tripfire for the red slot. Chestnut Knight could overlap its 2 AoE effects with other dragons and its revive ability is only good when your dragons die, which isn't a good thing to begin with, and if it fails to revive when it needs to.. Well, you get the idea. These species aren't as bad as the Timberjack or Monstrous Nightmare families, but... I still wouldn't recommend breeding for them.

10. Snow Wraith-Whispering Death-Hushbogle
The Snow Wraiths are peculiar support species. Their 2star Wild Wraith can be included against 1-4star Alphas for its fast charing and AoE Strong Hit, 3star Frostfang looks great on paper with a huge AoE damage taken buff and retained AoE Strong Hit, but in reality it's not as useful. 4star Coldreign is a sort of a stronger Fanghook with an anti-heal debuff instead. Unfortunately, it makes Coldreign very niche. 5star Furtive Fleetsleet is a more reliable version of Molten Magmanette, having similar effects with less drawbacks, but the tactical value isn't much greater either.
Whispering Deaths are rapidly charging offensive dragons that can either be your dragons of choice or absolutely not fit your team. 3star Sky-Pirate is only marginally useful in standard battles. 4star Bombwelter is a very flexible hostile dragon, having lifestealing single target attack and a weak AoE damage buff and Strong Hit. Between Corpsekeep and Bombwelter - Bombwelter is a more reliable upgrade from Torch. It's also very useful against Alphas due to Strong Hit. 5star Gritty Sawmaw isn't a tactical powerhouse, but it's a Very Fast charging 180% per target AoE lifestealing self-sufficient annihilator. It also slows down enemy spirit gain, so it might not be a specific-use dragon, but it's a dragon you can't fail with.
Hushbogles are rare hybrids that actually retain the better features of their parent dragons. 3star Mob-clobber retains Frostfang's AoE Strong Hit and adds some lifestealing burst that could be useful against Alphas. 4star Foehammer trades Mob-clobber's useless personal CA for Coldreign's single target anti-heal, but otherwise remains a directly upgraded version of Mob-Clobber. 5star Mythic Murmurquill is, at the time of writing this post, a broken\bugged dragon. In text - it's an upgraded Foehammer that trades Strong hit for AoE damage buff and makes anti-heal also AoE, already a very dangerous dragon. In reality - its damage works as AoE with single target values, which makes it an almost unkillable nightmare to fight against. It also charges fast, so it's a natural upgrade from Gritty Sawmaw... if you can get it.

Verdict: Do not breed Snow Wraiths, breed Whispering Deaths instead, starting with 4star Bombwelter. And, once you get it, you can breed it and Frostfang for a wide range of dragons that could in a distant future be bred for Sawmaw\Murmurquill\Fleetsleet.

11. Event\Unique dragons
-Toothless is the most powerful event dragon so far. If you don't have Shifty - it's your go-to purple dragon with fast charging, spirit drain and spirit buff and a powerful ricochet attack.
-Stormfly is a close second, having also fast charge, but focusing more on the offensive with a huge 170% AoE DoT and 40% damage taken debuff for 5 turns, she can probably sustain these effects with 75%+ uptime. As a cherry on top she gives the weakest ally a 30% shield. It may not be the most tactical blue dragon either, but it's clearly the go-to option.
-Barf&Belch may not be an ideal for standard battles, but this fast-charging, 250% bursting, 40% damage taken debuffing powerhouse is THE best green dragon against Alphas.
-Meatlug is not as straightforward as the previous 3 dragons, being set back by a slow charge, but it has a powerful AoE attack, a rare AoE damage done DEbuff and a huge 60% spirit gain bonus for 4 turns for the weakest dragon. If used tactically, this spirit-buffed dragon can become a nightmare.
-Skullcrusher is a bigger Flank-Tanker, upgrading the charging from Very Slow to merely Slow, getting a significantly more powerful ricochet and a single target hostile cleanse, while retaining 25% AoE healing.
-Cloudjumper is just a worse version of Meatlug. Instead of 25% AoE damage done DEbuff, it has 25% accuracy debuff, which is better against Strong Hit\CA, but worse against ability-based dragons. And instead of a huge sharable60% spirit bonus, this dragon only has a "personal" 35% shield.
-Lightfury doesn't go far on the list, because against Alphas - it has a worse uptime and debuff than Barf&Belch, in standard battles it's much less valuable than even Cloudjumper, and... at the time of writing this post, her unique evasion is broken. Sure, it allows her to avoid enemy ABILITIES... but it also allows her to avoid YOUR abilities. Including healing, spirit and all kinds of buffs. So... she can pull her weight, but that's about it. Her AoE damage taken debuff is just too weak, 20%, even if it lasts for 5 turns.
-Hookfang takes the last spot, because it's just a worse version of Lightfury in every way imaginable. It has a similar random-blast mechanic, but it charges Very Slow, and even if it manages to charge once, its own 30% spirit buff doesn't quite make it match a Moderate charging Lightfury. And its random-blast is not only a bit weaker, but its personal damage buff is also quite weak, even if lasts for 6 turns.

First I'd like to try to explain ATK\DEF.
ATK is the base value of your damage. Ability damage % is calcuated based on this number, and this number approximately accounts for 3 tiles worth of damage.
DEF works by reducing the damage of both tiles\blasts and abilities proportionally to the difference between ATK and DEF. Unfortunately I did not bother to gather concrete data, it may be updated later, but the approxiate observations from gameplay show something like this:
ATK-DEF proportion of 100%-100% would reduce the damage by ~40%. 100%-125% proprotion would reduce it by 60%+. 100%-150%< will reduce it by 75% and more. 125-100% would reduce the damage by ~25-30%. 150%-100% would reduce the damage only by 10% or less.
In other words, all this means that if you used a ~400 power dragon against a ~800 power dragon, it would deal about 20 damage per tile instead of 40. And, in turn, if you fought a 1300+ dragon with your 900 dragon, instead of ~90 damage per tile, it would deal only around 50.

ATK\DEF are also two of the four parameters defining the total power value of your dragon, the other being ability level and HP. Obviously ability level caps at 10, and ATK\DEF caps out when your dragon is max trained\max leveled. However, HP can further be scaled up by enemy dragons. Normally you can't see enemy stats beyond HP, total power, ability level and star training, but, for most of the game, these should be enough to distinguish what you are dealing with.
However, to put it mildly, if you don't know the dragons well yet, or do not yet understand all this ATK\DEF, the most concise way to put it is that "you shouldn't challenge dragons with more than 200 power over your dragons, at least until ~1000 power per dragon."
Never pay attention to the total team power, it doesn't even always reflect the dragons per wave(something devs should definitely illustrate), always focus on the power rating of individual dragons, especially if your team has non-linear powered dragons(like 450 500 550 600 650). If an enemy is of the strong color vs your weaker dragon - it might never survive past the first wave. So with an example of these 5 numbers, your team's total power is ~2750. Anything above 700 power per enemy dragon can be lethal to at least two of your dragons, so the cap of what you should try is 3500, which is 5x700 dragons. Of course, you can get lucky with stronger dragons, unlucky with weaker, and most all the type of dragon is important, but I'm trying to generalize here.
If you had a theoretical team of 600 power dragons, around 3000 total, then, in turn, it would be unwise to risk things that have more than 800 power per dragon, so anything above 4k. Of course, exceptions can be made for limited time events, like trust dragons, or if you truly want to get that Premium Draft card, but in general, as I said, energy is your most valuable resource, there is no point in wasting it to get standard resources.

Now, there are two primary modes of combat currently in the game: standard battles and clan alpha fights. At some point arenas will be added, and they might have their own set of rules, but for now there are only distinct two.

Alphas deserve their own section, and so does team optimization, so I'll conclude this section with a deeper overview of standard battles.

The standard battles can further be separated into: early game, midgame, endgame and deep endgame:

-The Early Game is the period when you have 1-3star dragons with "whatever you have" kinds of abilities. Most of this period can as well be auto-matched to save time. Usually it goes to about 3000 team power.

-The Midgame picks up after 3000, this is where you start facing 3-4star dragons, where enemies can start blocking your abilities, preventing healing and more. Obviously, to counteract that, you need to optimize your team, and, once you are done - you can fight enemies even above your power level. This period goes to about 5000 power, because that's where 3star dragons max out.

-The Endgame begins when you are physically required to have 4-5star dragon ATK\DEF stats, that's usually happens beyond 5000. 4-5star dragons also offer new tactical possibilities, more ability effects and combinations.

-The Deep Endgame is, for now, present only in Story Mode. The Deep Endgame can stretch a dragon's HP to any amount, and therefore adjust its power. The idea of Deep Endgame is that you should have enough defensive measures to survive the stretched-out waves, to single out enemy dragons and defeat them one by one. Normally you would leave the last dragon of a wave, heal up, then be ready for the next wave. But the Deep Endgame requires you to put the effort of multiple waves within the same wave. And, obviously, if you don't prepare for the next wave - the first wave will be your last.
To do this - you need an absolute optimal setup of endgame dragons, sometimes even tailored to a specific level. This stage is not necessary for normal gameplay, it's only a chance for the most paying players to push themselves even further.

Team Optimization
This section is currently in development
Community content is available under CC-BY-SA unless otherwise noted.