I do love how Flak's day-to-day bonus means that he actually has _worse_ average luck damage than someone with no day-to-day luck modifiers whatsoever.
The devs noticed and addressed this in Dual Strike, DtD increases from +15% > +25% and the -10% is unchanged so the average roll is 7.5% which at least makes it better than the average. But of course they weren't going to implement Dual Strike changes in Re-boot Camp.
I'm pretty sure the AI can see it's luck roll, so it'll choose engagements that roll high. That's probably why people are seeing Flak doing so well in the campaign.
I haven't played many of the campaign levels against Flak yet, but so far from what I've seen he hasn't really gotten many big rolls against me. If anything he seems to have poor rolls. Then again, the AI might be different on lower difficulty levels. Maybe on Casual it knows its luck roll and occasionally sandbags, taking bad engagements on purpose. He's RNG, so it's difficult to pin down exactly how he works. People are gonna have varied results... It could also be that certain rolls stick out more in people's memory based on how it changed the battle. Maybe that big roll he got that destroyed one of your key units stands out more to you than that low roll he got that didn't do anything.
That was some amazing timing, such a pile of high rolls specifically during the bathroom break. Has he gone back and watched that? He should if he hasn't. In any case, I don't think Mangs will ever pee during Flak's turn again.
The problem with flak appears if you deal 100% or more damage. For every co except flak (and sonja) you have a guaranteed kill. That means a lander or transport copter may survive a hit it shouldn't.
i think the only reason this happens in the campaign more often is because the AI cheats like it does in the original, and knows how much damage its going to do with the luck stat, so it will "take the gamble"
possible, though for what it's worth the AI notably cheats less than it did in the original since it can no longer tell where your units are in fog of war. It won't *forget* where a submarine dove or if it sees you run into an easily accessable forest/reef but it does in fact get trapped fairly often if it loses sight of your units in dense fog and forest.
@@edfreak9001 That's cheating by unseeable knowledge (aka something visual). I think the greater concern is being able to peer into the crystal ball and know what the dice rolls are. It might not help when it cannot determine where to use it at, but when it does...
@@FlameSoulis honestly, I think cheating by knowing where unseeable things are is more overt and more annoying to the player than having AI cheat by letting the AI know if it will get good rolls or not: it still can’t CHOOSE the rolls, it just gets a peak
I think the main reason flak feels better is that the AI is willing to gamble when it never did before, so you never saw ai flak deal 5 damage to a md. Tank with an infantry, but now you can. The AI is also willing to take losing engagements sometimes if its got nothing better to do, which is also new. Just a lot more chances for good luck rolls.
I think you're onto something. You know how in Re:Boot, the COs have phrases that they sometimes say in the start of their turn? I think this is an indicator that the AI is changing their auto-strategy. I haven't unlocked AW2 yet, but I'm guessing that Flak may always be set to strike in this game, allowing more hijinx to happen through.. dumb luck.
Came to say this! You need something that maybe deals 20 base damage to get a better idea. Given his luck range is more than 100 there’s no real way to use real life scenarios.
@@youvegotmailed It's not just possible but fact, really. However it wasn't what I was disputing; one of Flak's biggest characteristics is his bad luck and is being ignored here.
Anybody still find it weird that Flak needs a -10 luck modifier for is +15 luck modifier, but Nell has a -0 luck modifier for +20 luck? Or that if you swap their positive luck values, powers and all, Flak’s negative luck in all his powers feel more justified?
Flak isn’t meant to be good, he is basically meant as a tutorial enemy, no special unit bonuses, no strange powers etc. But I agree that swapping the good luck values of flak and Nell would be a good idea
The COs were designed with the campaign in mind, including Nell and Flak. Nell is a bonus CO who wasn't meant to be playable outside of Versus and War Room. She also has harder unlock requirements than everyone else. She's a reward for beating the game with a good score. Flak is meant to be an introduction to Advance Wars and the new mechanics of AW2. He's meant to be weak to ease the player into the difficulty. If he was actually good, new players would have a hard time beating the campaign. Each Black Hole CO gets progressively stronger until you reach Sturm, the ultimate Black Hole CO. This even shows in the AWBW tiers. Flak is tier 4 and trash. Adder is tier 4 but one of the best tier 4s. Lash is tier 3. Hawke is tier 1. Strum is tier 0. AW1 actually has good COs as your enemy very early, making it far more challenging. Grit appears by Mission 2 and that extra +1 range and +3 total with COP is overwhelming to a new player who isn't even used to the mechanics yet. Eagle appears by Mission 3 and he's capable of getting an entire extra turn already. Good lord, that's an insane challenge to introduce to a new player that early. Kanbei arrives Mission 9. He has +20% ATK and DEF and it's considered a Sami mission. Bruh, what were they smoking when they thought "Yeah, this is a good build-up to the mid-game difficulty. The CO strength in AW2 is a far more gradual in increasing difficulty. AW1 throws you into the deep end and says "Get good, scrub!"
The games were balanced around the PvE content, with weak enemy COs to fight at first, and strong unlockable COs for War Room. A PvP balanced Advance Wars probably wouldn't even have nonstandard luck values.
You could test faster if you had this setup instead of your current: tank|inf|inf|tank|tank|inf|inf|tank This way you wouldn't need to select targets for the rows that can attack on both sides in your current setup (because with the new setup that'd stop being the case).
There is a rumor about the AI can see the luck roll BEFORE the engagement is done. So Flak is bad but not so in the campaign or Mission 1-5 when Mangs is taking a bathroom break (at least in GBA because the AI is very stupid but cheats, especially at fog of war's vision limit doesn't exist unless unit is hidden at forest of reef).
Maybe the luck formula has changed in that it only uses one random number instead of 2 RNs. Because of this, it would be more likely for Flak to roll high good luck without it getting screwed over by a bad luck afterwords. Just a theory.
@@FireBurn256 that’s true, but wouldn’t rolling only one number mean that every outcome is equally likely to occur, as opposed to one good luck roll and one bad luck roll resulting in the result being more likely to be near the middle?
Considering the number of jackpots that was gotten I think this is the case. This means that, at the very least, he's got some very high highs and some very low lows. He's not just... constantly below average.
I think its just the times he has streaks of good luck make him look stronger. I've had him wipe out 9hp anti-air with B copters in the normal campaign, and that was certainly the most memorable thing that happened.
Hey Mangs I just noticed something you would like. Look at the physical box art of your advance wars copy, it is fully reversible! You get the art from black hole rising on the inside and can be flipped. Thought it was pretty cool.
I would have been interested to see how many of those luck rolls were bad luck rolls. I think what people are curious about is how the game handles good luck+bad luck interactions, so only looking at often Flak does higher than average damage only tells part of the story. I was personally surprised by his damage rolls when I played the campaign, so I have been waiting for people who are good with statistics to maybe take a look at it, but thanks for taking things into your own hand.
I think the only thing that may have been going on in the campaign is that when we see Flak roll good and then seeing the results. Luck values is... really hard to measure it out at times. The numbert is random by the end of the day and you won't know if you get a good or bad luck roll until you see it.
My thought is that it has to do with human nature. Humans by default remember their losses more than their wins. So the rare times when a big hit happened against them are things they're likely to remember in the match. Therefore just having a natural bias towards they must have changed him when its more likely that they didn't.
Imo it would have been cool if flak didn't have the low hp anti luck. It would make him be able to zerg with low hp units and it would be a neat strategy
Maybe even remove the -luck of his day-to-day altogether for units on very low health. After all, it’s hard for things to get much worse from there, right?
An interesting experiment. Having observed AI behavior in the campaign myself, I think it's more that just like in the original, the AI reads the RNG seed to see what will happen, and often doesn't bother to take engagements at all if it sees a bad luck roll. But when it sees those jackpots, it can't snap them up fast enough XD
Unfortunate that Flak gets no flat bonuses to his Attack, that would have helped offset his Bad Luck somewhat and enhanced his overall damage, while making him at least a _little_ more distinct to Nell. Doesn't even have to be massive increases, it could just be like +15% and +30% for each Power tier or something.
I've always thought that Jugger should have been this. Basically make him the anti-Flak and anti-Nell. Have his daily be that "baseline" luck is completely negated for both himself and the opponent, and both his powers give the opponent a turn of super bad luck.
So fun fact with this experiment, Flak had more 10 HP Md Tanks during his SCOP than his COP. Overall: 10 HP - 18 times (45%) 9 HP - 3 times (7.5%) 8 HP - 4 times (10%) 7 HP - 3 times (7.5%) 6 HP - 3 times (7.5%) 5 HP - 5 times (12.5%) 4 HP - 2 times (5%) 3 HP - 0 times (0%) 2 HP - 2 times (5%) Im sure if you ran this experiment 1000 times over, you would find a skew left curve, where there's about a 50% chance it does 0-1 extra luck damage and a 50% chance of the rest of the numbers. Also, another comment gave the COP numbers, and it appears Brute Force is more consistent/even with dealing higher numbers, whereas the SCOP high rolls trend toward the 7-5 range. Pretty interesting IMO and would look even cooler on a statistical graph.
I remember playing T Minus 15 where in one case, Flak used Barbaric Blow and got bad luck when attacking my Medium Tank with one of his. Both at less than 10 HP and his unit was lost from his own attack. So yes, he could be sending his units to die, if he gets only bad luck
They don’t really need a buff, they do their job just fine, if you want too buff the AA capabilities of wars world, they need too buff cruisers and missiles
My assumption is that the game's PRNG is determined differently. Like maybe he can have lucky or unlucky streaks depending on the calculation, so if he starts lucky then he's going to continue to be lucky and vice-versa.
The campaign can be very quirky, as it uses fixed luck values when it can (AW1 does this, too.) I have no idea how that relates to the luck CO's. One thing worth a test might be counter attacks of flak. They are using the AW1 formula, so no Luck on counters, but he might be a special case.
Power Test: 53 displayed damage/40 tanks = 13.25% average Super Power Test: 85/40 = 21.25% average Weak Infantry Test: 6/40 = 1.5% average Weak Mech Test: 23/40 = 5.75% average These numbers include damage they would normally deal.
For awnsering Mangs request * Infantry full hp Power : - avg dmg = 1,325 - possible max roll = 12,5% Infantry full hp Super Power : - avg dmg = 2,125 - possible max roll = 2,5% Infantry low hp Super Power : - avg dmg = 0,15 - possible max roll = 15% Mec low hp Super Power : - avg dmg = 0,475 - possible max roll = 12,5%
4:52 Would have been the perfect moment to use your own "Barbaric Blow" meme video that was made XD. (I think Davis was the name of the person who made it?) Would have been funny XD
If anyone is curious, here are the statistical results of the first two tests: COP: Sample mean damage of 1.358974, p-value of 0.1329, which is non-significant so it seems the average damage is 1 as expected SCOP: Sample mean damage of 2.125, p-value of 0.7508, which is REALLY non-significant, so it seems the average damage is 2, as expected Technically Mangs (gasp) did not follow statistical procedure in tests with damaged units because the attacking units are not all the same HP, so you can't really run standard statistical tests because the tests don't have the same starting values. If anyone wants to double check my R code: COP_dmg_vals
I'd still assume they made Luck into one roll. It has on average the same result, but the more extreme outcomes are much more probable. Max rolls should not be as likely as they are in this showcase here. If we use the "Good Luck + Bad Luck" mechanic the odds for a max roll for Brute Force would be 1/40 x 1/20 = 1/800 or .125% while Barbaric Blow would be 1/80 x 1/30 = 1/2400 or .042%. If we only use "one Luck roll", the odds would be 1/59 or 1.695% and 1/109 or .917%. Max rolls would be about 13,5 times as likely for Brute Force and about 22 times as likely for Barbaric Blow. I've calculated the odds in this video and hope I'm not doing Mangs Math right now: Mangs got five max rolls for Brute Force. For "one Luck roll" the odds for at least five max rolls in 40 engagements would be about .093%, for "Good Luck + Bad Luck" it would be an abysmal .0000002% - 1 in 500 million! Mangs got two max rolls for Barbaric Blow. For "one Luck roll" the odds for at least two max rolls in 40 engagements would be about 6.56%, for "Good Luck + Bad Luck" it would be merely .0135%. If my Math is somehow correct, theis showcases would already be pretty lucky if we go with the "one Luck roll" theory, while "Good Luck + Bad Luck" would be pretty much impossible. The numbers displayed on the Medium Tanks also seem to be more evenly distributed, so it would support the "one Luck roll" theory.
I Don't think Flak is the worst co in the game.Flak is better than you think. I think he is pretty strong for lower tier, atleast stronger than Grimm,Sonja & Koal .It's not only me (Go7) one of the best awbw player says that Flak is tier 3.5.
@@WavemasterAshi Flak is banned for use in tournaments, as are all luck COs (so Jugger and Nell too) because the AWBW admin does not like luck-based COs. He's stated he doesn't want games determined by luck (outside of the uncontrollable default luck.)
I played that campaign mission where Olof is attacking a giant missile. I had 2 MDs blocking off the access to my other units the popped his supper and did 0 damage on both fronts. It was pretty bad.
Barbaric Blow - and two max rolls on 40 tanks seems ridiculous. But I did a few 100 attempts myself and shockingly enough: 1 out of 16 attempts reduced an Md Tank to 2 HP - which should only be possible with the highest luck value. In the GBA original that would be a rarity: If you need to achieve both max good luck and min bad luck, the odds would be 1/2400. Chances are you'd be getting bored long before it happens even once! I suggest not only repeating the test, but also performing it on the original game. 3 HP should be significantly less common and you really need good luck (heh) to get a Md Tank down to 2 HP.
I think people didn't use to pay attention to these mechanics when they were kids playing the original on GBA and now that they watch CZcams content over-explaining these mechanics they are constantly on the lookout for good rolls.
So I ran some numbers... tl;dr/abstract: It seems most likely from these results that they have changed the luck values to use 1 random number. They may also have adjusted the luck values to cover all the values they refer to rather than being one short (so a +40 luck bonus would mean 0-40 rather than 0-39), however, this is much less clear from the data. Either that, or Mangs is just incredibly lucky. More research is needed. Now, if we run the numbers exactly as they were in the original game as Mangs described at the start of the video, for the Brute Force Case we determine that the chances of getting a max roll of 40% =4HP of damage are 1/40 * 1/20 = 1/800 (Good luck has to be maximum value of 39, bad luck has to be 0), bearing in mind the 1 base damage so we need 39 luck damage. Already, we see the issue - this means, on average, we should expect to need 800 attacks to get maximum damage, which... we don't have anywhere near. But to be exact, the chances of getting at least 5 maximum rolls in 40 tests is 1.94E-09, or 0.00000000194 For the Barbaric Blow case, the chance of a maximum roll is 1/2400 and the chance of getting that at least twice in 40 tests is 1.34E-04 or 0.000134 - this is definitely a lot more likely than our , but the chances of both these sets of events occurring is 2.5996e-13, or less than 1 in a trillion. So something definitely appears to be different here. There are three possibilities: A. The luck values are now 'corrected' so they vary up to the maximum value indicated, e.g. a luck value of 40 goes from 0-40 (rather than 0-39) B. The luck values for bad luck and good luck are combined as 1RN (this has been suggested in other comments) C. Both of the above are the case Case A would mean that there are 3 chances for success in each of our examples, rather than 1 over a slightly larger (but not 3 times larger) range of outcomes. E.g. for Brute force, a maximum value of 40% would occur in Mangs' example when the Good Luck and Bad Luck values were any of the pairs - (39, 0), (40, 0), (40, 1), while there are now 41*21 = 861 outcomes. Case B simply flattens the curve, so for Brute force the chance of getting a total of 40 damage become 1/59 [as you pick a random number from -19 to 39, including 0] Case C both flattens the curve and increased the number of successful max damage rolls (this time only to 2) and slightly increasing the number of possible outcomes - in the case of Brute Force, you would need to roll a 39 or 40 and the possible numbers you could roll are -20 to 40, so there are 2 good outcomes out of 61 [so chance is 2/61] Using each of these cases, the chances of getting each of the outcomes shown in the video (5 Max rolls on Brute force, 2 on Barbaric Blow, each out of 40 attempts) are: A: Brute Force - 3.05E-07, Barbaric Blow - 0.00108 B: Brute Force - 5.55E-05, Barbaric Blow - 0.0522 C: Brute Force - 9.59E-03, Barbaric Blow - 0.162 It therefore seems very unlikely that scenario A is the case, and much more likely that B or C is the case. In conclusion, I therefore posit that the luck calculation method has changed to 1RN. Whether or not they have also extended the range of luck values to include the endpoints, I do not feel we can conclude, but it seems somewhat likely. I would also caveat this with three points: First, even if I've done all the numbers correctly, this does not 'prove' the system of calculation has changed, just strongly hints at it. Even in case C, the chances Mangs would get results at least as good as he got were about 0.00155358, or a little over 1.5 in a thousand - he got pretty good luck (haha) regardless of what the system used was. In order to make a better conclusion, more data would be required. Second, if there was anything else affecting the damage calculation, the above numbers would get thrown off wildly. As we saw, just allowing the damage from luck to be 1% lower made the event 3 times as likely - from what I understand the mechanics of the game (I am by no means an expert), there should have been no other effects changing the luck value (as Andy has no defence bonus and the tanks were on roads), but if we even had a hidden +5% bonus to damage suddenly we have 21 'success' cases occurring. Third - technically in the GBA games, as I understand it, you were very slightly more likely to get lower numbers for your luck values than the higher values. This is because the luck value was determined by taking an unsigned 32-bit integer and finding its remainder when divided by the luck value (so for most COs, this is just the last digit of the number). As a 32-bit number takes values from 0 to 4,294,967,295, 0-5 would be very slightly more likely to appear than other numbers. This is so tiny a difference and makes the maths so much more complex that I omitted it in the calculations. Most likely, with a modern console, they would just call a standard rng function between limits (which is likely also biased, but may be in a different way from the original). For reference (and in case anyone can spot an issue with my methodology), the formula used to calculate the chances of getting *exactly* X max rolls is: P(Maximum Damage Rolls = X) = (S/Q)^X * ((Q-S)/Q)^(T-X) * T!/(X! * (T-X)!) Where Q = the number of Possible rolls of luck S = the number of those rolls resulting in a maximum damage value T = the number of tests (40 in every case) ...thank you for coming to my TED talk.
A good way to power up his super is to make it a 50/50 roll between the lowest and the highest roll. That way theres a chance of floundering or getting that heavy blow you are searching for.
I actually have a suggestion for how they can buff Flak in a way that makes sense without changing what they do. Make the good luck rolls and bad luck rolls take precedence over another. What this means is that Flak will roll for his good luck, and then roll for his bad luck. Whichever number value for either positive or negative is higher, it will then only add/subtract from that attack. If he rolls a +7 for good luck, but then a minus 5 for bad luck, instead of adding both values together, the highest number overrides the tally: now his attack will deal +7% extra luck damage. On the flipside, if he rolls a minus 9 on another attack for bad luck, but rolls a +3 on good luck, the attack will now deal minus 9% damage because the bad luck roll was higher than the good luck. If both rolls are in equal value, then the attack will be an unmodified attack with an overall 0% increase/decrease in power. All of this would carry over his CO and Super CO power, of course, and implementing this change should be incredibly easy. This will make Flak's power feel a lot more impactful and won't result in an awkward spot where his good luck roll and bad roll cancel each other out, making his power effectively non-existent for the most part. Flak's power will now be meaningful in some way. He'll still be one of the worse COs out there and still ultimately be a worse Nell on average, but now, if he's lucky, he can now hit harder than Nell ever could- albeit at a risk.
So if there are changes to the Reboot camp will that be reflected on the website versions? Doesn't look like there are any changes to the multiplayer side yet, but do you have a plan or know of one if they do end up existing?
The way that Flak's Luck values are shown is a big case of "Programmer Logic". In most programming languages, if you need a random integer from 0 to X, the way you usually calculate it is to write something like Random.nextInt(X+1). So if you need a number between 0 and 14, you write Random.nextInt(15). On the GBA, where CPU cycles were at a premium, it's even possible the programmers just used something like rand() % 15, which would also give a number between 0 and 14 (though slightly unbalanced, because modulo only gives uniformly distributed numbers when the divisor is a whole power of 2). Either way, it's easy to see why they're writing it like "-10 to +15": it conforms to the underlying logic they'd have to had implemented when they were writing the game's code.
I think some people simply have a condition that I’ve called “0 in 1 rng” where they get either the best or worst rng possible with no in-between for several games in a row, I happen to have this and the odds can very quickly go from above average or normal, to below a 0.0X% chance [x being numbers in the 0.09- range] very consistently. It can be very frustrating seeing this multiple games in a row...or it was legitimately bugged lol
How about Drake? Has he actually gotten a buff? Or is it just a mistake in the CO description? I still haven't been able to find the time to beat the advance wars 2 campaign, so I can't test it out...
The description is wrong. If you go to gallery, to COs, and go to Drake's third page, and you swap between style 1 and style 2, you can clearly see his air units have worse attack in style 2 compared to style 1.
Maybe is just Percive extra damage? As most time will be a half health tank, so it getting lucky and causing 20-30% more damage can be a surprise. As expose here, flak power are very underwelming, even super has like 40% chance of doing nothing, so it kinda surprise you welhen it does.
this is a great video about how worthless flak is,hes basically like playing with co powers off only you sometimes get a lucky hit like 10% of the time.
brute force netted 53 hp worth of damage on those medium tanks. 84800 worth of funds gone from brute force. done with less than half that in value of units, in one turn. While it's definitely not the most powerful, Flak's the only one besides Nell who can get such a value return like that. barbaric blow got 85 hp worth of damage. Only an extra 32 over brute force. That's still 136000 worth of funds. seems like brute forcing multiple times and ignoring barbaric blow altogether might be a better option
Total damage of 51 Total health of md tanks 400 Aaverage of 1.275 hp. However, we know 1hp is actually 10% and it is possible a lot of those numbers are up to 5% more (assuming reboot camp rounds up 9.5hp to 10hp etc). 40 x 0.5 (for the potentially hidden extra 5%) = 20 Flak potentially had 71 damage which is 1.775 hp on average. I don't know if this is good compared to other co's in the same scenario but 17.75% of a medium tank that costs £16,000 is £2,840. We could take away the damage cost to the infantry on average (but i am not doing that maths) to see how cost effective it actually is, but assuming 5hp counter attack (£500) for example, means a potential net profit of £2,340. 12.75% is £2.040 btw, and using the same 5hp damage that would be £1,540. Infantry attacking md tank during super co power is somewhere between £1,540 and £2,340. The average between the best and worse case scenario for the average is £1,940. The absolute best in the video was 45% damage, meaning that engagement is worth £7200 (-£500 for counter) = £6,700. Of course, the worse case scenario was 0 damage and 9hp lost = -£900. TLDR; Flak infantry attacking a md tank during a super co power will range between 0hp and 4.5hp meaning a net cost of -£900 and £6,700 (assuming 5hp of damage to infantry). Edit; this was for co power NOT super co power!
Super co power did 85 damage. Maximum (including 0.5 hp damage (20hp) of 105hp damage. 105hp/40 md tanks= 2.625hp of damage on average. £4200-£500=£3700 profit on average. Max would be £12,800 and that results in 2hp return damage iirc, meaning a profit on £12,600. Definitely cost effective on average but this experiment does not really show the bad luck side where flak will do less damage than usual as an infantry is expected to do 0hp to a md tank, meaning we only see the "positive results" so keep that in mind. Flak is actually worse than what the video shows because the bad luck blends in with the other results.
@@EpicEverz 0.1% rounds up to 1hp so i think you are right! From playing the game myself, a unit with just a sliver over 5hp shows as 6hp and 8hp shows as 9hp. Not changing the maths, but well spotted! I think this experiment would have more readable results with infantry versus infantry or tanks. Same conclusion though; flak is bad!
Honestly probably just confirmation bias, Flak gets a couple of lucky rolls in a playthrough and those stand out in people's minds more than all of the mediocre/bad rolls so it feels like he's doing more damage than he should
does luck reroll on restarting the turn or will that always stay the same? i never knew if you could keep restarting till most of your units land a decent roll
Mangs can you perhaps look into the weird "Indirect Bomber" glitch that seems to be happening, sometimes the AI (mostly Sonja it seems) can strike indirectly with direct units. I've had a Sonja tank attack me from 4 tiles away when she used her power in the mission Sonja's Goal
Current theory is that when she uses her power and extends her field of view during fog of war it causes the issue. It is most noticable with bombers due to the extreme move range.
The thing I find so odd about luck COs, is that everyone looks at what an infantry can do to the strongest tank in the game... When in reality, there are far better units that take advantage of luck damage, and would be far and away better with a CO like Flak. Indirects, Bombers, Battle Copters, and any of the big guns like; Neo Tanks, Medium Tanks and even just regular Tanks honestly. Yes, you can score devastating hits with infantry. However, even with Nell this is a risky and quite noobish gamble. Losing your infantry means losing your capture game, even if you take a few crazy units out. I understand why Mangs is using infantry here to demonstrate this concept, but I'm not talking about this demonstration. I'm talking about the folks who do this legitimately.
Flak pretty much exist just to be a "bad cpu" CO and the devs just let us play him for fun. Like you could handicap yourself picking flak when you play vs your siblings or something.
Do you think you could do another little test video on using CO powers with extra charge stored in the bank? Like having your Super ready but only using the Little Power, and seeing if that actually costs extra like you say it does in your recent Secret Tips video (Tip #4). I strongly believe this to be in error, and would like to see it given official testing in the Reboot Camp. To clarify a bit: Tip #3 in that same video suggests (I believe accurately) that additional charge is required after every use of the CO Power (normal or super). As such, if you had six stars charged at 100 points apiece, and used a 3-star power, you would normally expect to see three full stars once your power ends, but each star now costs 120 points to fill, so your 300 points fill two stars with only 60/100 on the third star. This makes it look like you lost 3.4 stars, and thus 340 points, but in reality you only lost the 300 you expected to spend. If any CO in the game has exactly 6 more Super stars than Little stars, the math works out perfectly to go from precisely 6 stars remaining to precisely 5 stars after the very first use of the power in that game. I can't remember if any COs have this, but the first one I'd look at is Eagle.
Assuming my understanding of the luck rolls is correct, in order for Brute Force to get the Infantry to deal 4HP of damage to a MD tank, you'd need to roll a perfect 40 on the Good Luck and no Bad Luck. If every number has an equal chance of being rolled, that would be a 1/40 (chance of rolling max Good Luck) * 1/20 (chance of rolling 0 Bad Luck) = 0.00125, or something you'd expect to see about 0.1% of the time. Any other result would do less than 40 HP of damage. Advance Wars does not round HP values down, not to the best of my knowledge. If the display does round down starting at 65 HP, it makes it more complicated - if you get a 35 Good Luck and a 0 Bad Luck, it will take the MD Tank to 6, then 36 and 0 or 1 will work as well, 37 and 0-2, 38 and 0-3, 39 and 0-4 and 40 with 0-5. This does dramatically increase the odds of doing 4HP of damage on-screen, but it's still only 17 possible rolls out of a total 800, about 2.125% Hell, let's get ridiculous - if the HP rounds down to 6 starting at 69 HP (meaning 31 damage) then Flak has 45 possible rolls to get that threshold, or a 6.875% chance of displaying 6HP. This last number is, of course, absurd, but when I compare it to what is seen in the video then you see why I chose an absurd scenario like that. But assuming the first formula where the chance of 40 HP of damage is 0.00125, out of a random sample of 40 MD Tanks you would only expect one of them to be taken down to 6HP. In the test we see 5 Brute Force Infantry do the full 4HP of Luck damage on 5 of the 40 MD Tanks, or a rate of 12.5% which is NOTICEABLY higher than even the one using an absurd scenario where 69 rounds down to 6. To be precise, using the first estimate (no rounding), the statistical probability of Flak doing the full 4 damage 5 or more times out of 40 trials is 0.00000000193619 - or, as you might say, "practically impossible". With rounding his 35 or higher rolls to 40 it's 0.15% (meaning not something you'd ever expect out of a single random sample but far from impossible). So the luck really doesn't make sense to me but I think the test here shows a sufficiently STRANGE result to believe it's not just "he rolls both luck values independently, with all numbers equally likely to come up, and then it just applies as he rolled them."
@Mangs From what I've seen, people are talking about Flak's damage fighting him in the campaign, not multiplayer: for whatever reason, Flak's rolls didn't actually seem like true random in the campaign (in fact, I'm pretty sure they were outright fixed,) so people didn't see much, if any, extreme damage. Now, Flak's RNG is consistent, so people are actually seeing Flak get the same nutty damage rolls they would with multiplayer.
I do love how Flak's day-to-day bonus means that he actually has _worse_ average luck damage than someone with no day-to-day luck modifiers whatsoever.
Isn't his average luck bonus 2%?
@@FireBurn256 he have 2.5% average luck
And he is just straight up worse than Nell.
kid named Sonja:
The devs noticed and addressed this in Dual Strike, DtD increases from +15% > +25% and the -10% is unchanged so the average roll is 7.5% which at least makes it better than the average. But of course they weren't going to implement Dual Strike changes in Re-boot Camp.
Flak is flak. No matter what happens he can never escape being Flak. F is for Flak, but it’s also for F tier too
FLAK TIER!!!!
no wonder mangs fans are just as retarded as mangs is.
Flak is the top of the bottom baby, the cream of the crap, the king of the garbage heap. And no one can take that from our precious boy
@@braedenmclean5304 **Jugger rebooted**
I love Flak, unfortunately :(
I'm pretty sure the AI can see it's luck roll, so it'll choose engagements that roll high. That's probably why people are seeing Flak doing so well in the campaign.
I haven't played many of the campaign levels against Flak yet, but so far from what I've seen he hasn't really gotten many big rolls against me. If anything he seems to have poor rolls. Then again, the AI might be different on lower difficulty levels. Maybe on Casual it knows its luck roll and occasionally sandbags, taking bad engagements on purpose.
He's RNG, so it's difficult to pin down exactly how he works. People are gonna have varied results...
It could also be that certain rolls stick out more in people's memory based on how it changed the battle. Maybe that big roll he got that destroyed one of your key units stands out more to you than that low roll he got that didn't do anything.
I wouldn't be surprised if this was the case, too. Edit: Removed some insane rambling about Flak being in AW1.
@@TJ-vo3rv Flak was in AW1? o_O
@@paragonyoshi4237 Ignore me, I've had a day xD
Yes but it re-roll'd on engagement so it didn't actually help him in the previous games.
I love how sometimes, when you go into the shop, Hachi pretty much straight-up calls Flak a sucky CO. Even WayForward knows how notoriously bad he is.
He straight out tells you that you shouldn't pick him unless you're doing a challenge or something.
He also acknowledges how broken Eagle’s AW1 power is
I love Flak :(
I really feel like Flak should be incredible offense with little defense, but dual strike has a CO that already does that in Grimm.
Nothing will change the fact this is mangs studying the game code to figure out how Normal AI Flak outsmarted him in T-15.
That was some amazing timing, such a pile of high rolls specifically during the bathroom break. Has he gone back and watched that? He should if he hasn't.
In any case, I don't think Mangs will ever pee during Flak's turn again.
What video was that? I need to see this.
@@gabrielnoblat8017 his AW 2 reboot stream.
@@CrowMercury Thanks man!
The problem with flak appears if you deal 100% or more damage. For every co except flak (and sonja) you have a guaranteed kill. That means a lander or transport copter may survive a hit it shouldn't.
i think the only reason this happens in the campaign more often is because the AI cheats like it does in the original, and knows how much damage its going to do with the luck stat, so it will "take the gamble"
possible, though for what it's worth the AI notably cheats less than it did in the original since it can no longer tell where your units are in fog of war. It won't *forget* where a submarine dove or if it sees you run into an easily accessable forest/reef but it does in fact get trapped fairly often if it loses sight of your units in dense fog and forest.
@@edfreak9001 That's cheating by unseeable knowledge (aka something visual). I think the greater concern is being able to peer into the crystal ball and know what the dice rolls are. It might not help when it cannot determine where to use it at, but when it does...
@@FlameSoulis honestly, I think cheating by knowing where unseeable things are is more overt and more annoying to the player than having AI cheat by letting the AI know if it will get good rolls or not: it still can’t CHOOSE the rolls, it just gets a peak
He was nerfed hard. He's now treated as a dog, but he was supposed to be a gorilla.
If he's ended up a dog instead of 🦍, then I believe Jugger's gonna be scraps instead of a robot...
Wait I think that's already a thing.
I wonder why that line was changed.
@@FabledX_UK I dunno. They also changed the line "Two women CO, good country, must take it" and "indirect bombers".
Indeed, his name is Kong in japanese
@@thunderjellyfish3680 Wow!
I think the main reason flak feels better is that the AI is willing to gamble when it never did before, so you never saw ai flak deal 5 damage to a md. Tank with an infantry, but now you can. The AI is also willing to take losing engagements sometimes if its got nothing better to do, which is also new. Just a lot more chances for good luck rolls.
I think you're onto something. You know how in Re:Boot, the COs have phrases that they sometimes say in the start of their turn?
I think this is an indicator that the AI is changing their auto-strategy. I haven't unlocked AW2 yet, but I'm guessing that Flak may always be set to strike in this game, allowing more hijinx to happen through.. dumb luck.
The knowledge that Flak's luck is two rolls instead of one is significant. It means that he's more likely to get +0-+5 instead of -10-+15
The problem with testing this against MD Tanks is that you can't tell the difference between bad rolls and neutral rolls.
Came to say this! You need something that maybe deals 20 base damage to get a better idea. Given his luck range is more than 100 there’s no real way to use real life scenarios.
@@xaif4888 i think what you guys are saying is possible, but that would be such an minimal buff that Flak would still be bottom tier
@@youvegotmailed It's not just possible but fact, really. However it wasn't what I was disputing; one of Flak's biggest characteristics is his bad luck and is being ignored here.
On the other hand, Nell's Lady Luck can make her infantry decimate a Md Tank. Flak really got the short end of the stick.
Anybody still find it weird that Flak needs a -10 luck modifier for is +15 luck modifier, but Nell has a -0 luck modifier for +20 luck?
Or that if you swap their positive luck values, powers and all, Flak’s negative luck in all his powers feel more justified?
Since Flak is the first CO you face in Aw2, I suppose they wanted him to be really weak.
Flak isn’t meant to be good, he is basically meant as a tutorial enemy, no special unit bonuses, no strange powers etc. But I agree that swapping the good luck values of flak and Nell would be a good idea
The COs were designed with the campaign in mind, including Nell and Flak.
Nell is a bonus CO who wasn't meant to be playable outside of Versus and War Room. She also has harder unlock requirements than everyone else. She's a reward for beating the game with a good score.
Flak is meant to be an introduction to Advance Wars and the new mechanics of AW2. He's meant to be weak to ease the player into the difficulty. If he was actually good, new players would have a hard time beating the campaign. Each Black Hole CO gets progressively stronger until you reach Sturm, the ultimate Black Hole CO. This even shows in the AWBW tiers. Flak is tier 4 and trash. Adder is tier 4 but one of the best tier 4s. Lash is tier 3. Hawke is tier 1. Strum is tier 0.
AW1 actually has good COs as your enemy very early, making it far more challenging.
Grit appears by Mission 2 and that extra +1 range and +3 total with COP is overwhelming to a new player who isn't even used to the mechanics yet.
Eagle appears by Mission 3 and he's capable of getting an entire extra turn already. Good lord, that's an insane challenge to introduce to a new player that early.
Kanbei arrives Mission 9. He has +20% ATK and DEF and it's considered a Sami mission. Bruh, what were they smoking when they thought "Yeah, this is a good build-up to the mid-game difficulty.
The CO strength in AW2 is a far more gradual in increasing difficulty. AW1 throws you into the deep end and says "Get good, scrub!"
The games were balanced around the PvE content, with weak enemy COs to fight at first, and strong unlockable COs for War Room.
A PvP balanced Advance Wars probably wouldn't even have nonstandard luck values.
@Karsa hyena I'd forgotten Reboot Camp renumbers the missions and remembered the last video Mangs made. So yeah, AW1 just went in with no lube.
You could test faster if you had this setup instead of your current:
tank|inf|inf|tank|tank|inf|inf|tank
This way you wouldn't need to select targets for the rows that can attack on both sides in your current setup (because with the new setup that'd stop being the case).
flak try not to be bad challenge (impossible)
There is a rumor about the AI can see the luck roll BEFORE the engagement is done. So Flak is bad but not so in the campaign or Mission 1-5 when Mangs is taking a bathroom break (at least in GBA because the AI is very stupid but cheats, especially at fog of war's vision limit doesn't exist unless unit is hidden at forest of reef).
Maybe the luck formula has changed in that it only uses one random number instead of 2 RNs. Because of this, it would be more likely for Flak to roll high good luck without it getting screwed over by a bad luck afterwords. Just a theory.
He still would get just a plain +7% on average.
@@FireBurn256 that’s true, but wouldn’t rolling only one number mean that every outcome is equally likely to occur, as opposed to one good luck roll and one bad luck roll resulting in the result being more likely to be near the middle?
That's a very strange way to word it. It's not a buff, it just makes him even more inconsistent than before.
Flat random chance, like a die, rather than a bell curve.
Considering the number of jackpots that was gotten I think this is the case. This means that, at the very least, he's got some very high highs and some very low lows. He's not just... constantly below average.
He may still be the worse character in the game, he's still the best character in my heart.
I think its just the times he has streaks of good luck make him look stronger. I've had him wipe out 9hp anti-air with B copters in the normal campaign, and that was certainly the most memorable thing that happened.
Hey Mangs I just noticed something you would like.
Look at the physical box art of your advance wars copy, it is fully reversible! You get the art from black hole rising on the inside and can be flipped. Thought it was pretty cool.
Weird seeing Falk without his iconic helmet and googles talking about himself...
I would have been interested to see how many of those luck rolls were bad luck rolls. I think what people are curious about is how the game handles good luck+bad luck interactions, so only looking at often Flak does higher than average damage only tells part of the story.
I was personally surprised by his damage rolls when I played the campaign, so I have been waiting for people who are good with statistics to maybe take a look at it, but thanks for taking things into your own hand.
I think the only thing that may have been going on in the campaign is that when we see Flak roll good and then seeing the results. Luck values is... really hard to measure it out at times. The numbert is random by the end of the day and you won't know if you get a good or bad luck roll until you see it.
Andy: "I won by doing absolutely nothing!"
Luigi: "That's-a my boy!"
My thought is that it has to do with human nature. Humans by default remember their losses more than their wins. So the rare times when a big hit happened against them are things they're likely to remember in the match. Therefore just having a natural bias towards they must have changed him when its more likely that they didn't.
Imo it would have been cool if flak didn't have the low hp anti luck. It would make him be able to zerg with low hp units and it would be a neat strategy
where the 'strategy' is just throw shit at the wall and see what happens lol. that would've been cool tho + totally in line with flaks character.
Maybe even remove the -luck of his day-to-day altogether for units on very low health. After all, it’s hard for things to get much worse from there, right?
An interesting experiment. Having observed AI behavior in the campaign myself, I think it's more that just like in the original, the AI reads the RNG seed to see what will happen, and often doesn't bother to take engagements at all if it sees a bad luck roll. But when it sees those jackpots, it can't snap them up fast enough XD
His powers should be 1 star cheaper to at the very least get the extra defense more often. Or they could reduce terrain stars by one.
"sometimes you get lucky and get the full blow." - Mangs
They need to come up with Style 3, where all the CO's are balanced for multiplayer. Flak wouldn't suffer as much.
Unfortunate that Flak gets no flat bonuses to his Attack, that would have helped offset his Bad Luck somewhat and enhanced his overall damage, while making him at least a _little_ more distinct to Nell. Doesn't even have to be massive increases, it could just be like +15% and +30% for each Power tier or something.
What if there was a CO that could inflict bad luck onto other COs? That would be cool.
Call the Super Power 13th Hour
And their standard CO power Black Cat...
I've always thought that Jugger should have been this. Basically make him the anti-Flak and anti-Nell. Have his daily be that "baseline" luck is completely negated for both himself and the opponent, and both his powers give the opponent a turn of super bad luck.
Random tidbit but has anybody else noticed that when Flak's CO power theme kicks in, the drumline sounds like the intro to The Beast's theme?
You mean the beast from DoR/ dark conflict?
@@art-games6230 Yes.
@@racer927 then no, I didn’t notice
So fun fact with this experiment, Flak had more 10 HP Md Tanks during his SCOP than his COP. Overall:
10 HP - 18 times (45%)
9 HP - 3 times (7.5%)
8 HP - 4 times (10%)
7 HP - 3 times (7.5%)
6 HP - 3 times (7.5%)
5 HP - 5 times (12.5%)
4 HP - 2 times (5%)
3 HP - 0 times (0%)
2 HP - 2 times (5%)
Im sure if you ran this experiment 1000 times over, you would find a skew left curve, where there's about a 50% chance it does 0-1 extra luck damage and a 50% chance of the rest of the numbers. Also, another comment gave the COP numbers, and it appears Brute Force is more consistent/even with dealing higher numbers, whereas the SCOP high rolls trend toward the 7-5 range. Pretty interesting IMO and would look even cooler on a statistical graph.
Wow!
Hey guys, spirit of the law here
@@eksmad today were gonna take a look at the new civilization, the Flaks
I remember playing T Minus 15 where in one case, Flak used Barbaric Blow and got bad luck when attacking my Medium Tank with one of his. Both at less than 10 HP and his unit was lost from his own attack. So yes, he could be sending his units to die, if he gets only bad luck
Asked this on Reddit, didn’t get a straight answer, nice too see someone test
Oh the CO flak.
I thought this was about the anti-air guns.
Bro 💀
They don’t really need a buff, they do their job just fine, if you want too buff the AA capabilities of wars world, they need too buff cruisers and missiles
@@art-games6230 buffed AA would almost normalize them into tanks
@@germmanator basically, yes
Flaks are a ranged anti air unit likevan artilery but for antivair thats not in advance wars.
My assumption is that the game's PRNG is determined differently. Like maybe he can have lucky or unlucky streaks depending on the calculation, so if he starts lucky then he's going to continue to be lucky and vice-versa.
Kinda like the "Hot" slot machines in Pokémon right?
The campaign can be very quirky, as it uses fixed luck values when it can (AW1 does this, too.)
I have no idea how that relates to the luck CO's.
One thing worth a test might be counter attacks of flak. They are using the AW1 formula, so no Luck on counters, but he might be a special case.
Dammit Mangs! I thought I had a smear on my screen.
flak being the most powerful CO in the game by doing absolutely nothing (dream luck mode)
Flak installed cheats on accident
@tezereth yeah that sounds like Flak
Breaks preexisting laws of the game via sheer stupidity
@@OriginalGameteer Flak is Ed confirmed
mangs please keep up the good work you are the only person on youtube doing a hard mode guide and i need the help!
Play as Flak is for the most daredevil, hardcore AND lucky players apparently...
EDIT: Apparently IA don't give any Fu**s for Bad Luck too XD
Power Test: 53 displayed damage/40 tanks = 13.25% average
Super Power Test: 85/40 = 21.25% average
Weak Infantry Test: 6/40 = 1.5% average
Weak Mech Test: 23/40 = 5.75% average
These numbers include damage they would normally deal.
Also worth noting that the odds of doing no damage is roughly equivalent between Power and Super Power (42.5% and 45% respectively)
Flak still my favorite character
For awnsering Mangs request *
Infantry full hp Power :
- avg dmg = 1,325
- possible max roll = 12,5%
Infantry full hp Super Power :
- avg dmg = 2,125
- possible max roll = 2,5%
Infantry low hp Super Power :
- avg dmg = 0,15
- possible max roll = 15%
Mec low hp Super Power :
- avg dmg = 0,475
- possible max roll = 12,5%
4:52 Would have been the perfect moment to use your own "Barbaric Blow" meme video that was made XD. (I think Davis was the name of the person who made it?) Would have been funny XD
@4:27
Here is the damage breakdown.
Brute Force:
Average: 1.325
Stdev: 1.45
"Jackpot" 12.5%
Barbaric Blow:
Average: 2.125
Stdev: 2.47
"Jackpot" 5%
Low Blow:
Average: .36
Stdev: 0.15
"Jackpot" 15%
I wish Flak was more unique. Like I don’t mind him being super bad but he is literally just Nell but worse
If anyone is curious, here are the statistical results of the first two tests:
COP: Sample mean damage of 1.358974, p-value of 0.1329, which is non-significant so it seems the average damage is 1 as expected
SCOP: Sample mean damage of 2.125, p-value of 0.7508, which is REALLY non-significant, so it seems the average damage is 2, as expected
Technically Mangs (gasp) did not follow statistical procedure in tests with damaged units because the attacking units are not all the same HP, so you can't really run standard statistical tests because the tests don't have the same starting values.
If anyone wants to double check my R code:
COP_dmg_vals
Why isn’t this pinned ?! Thanks for doing the science !
I'd still assume they made Luck into one roll. It has on average the same result, but the more extreme outcomes are much more probable. Max rolls should not be as likely as they are in this showcase here. If we use the "Good Luck + Bad Luck" mechanic the odds for a max roll for Brute Force would be 1/40 x 1/20 = 1/800 or .125% while Barbaric Blow would be 1/80 x 1/30 = 1/2400 or .042%. If we only use "one Luck roll", the odds would be 1/59 or 1.695% and 1/109 or .917%. Max rolls would be about 13,5 times as likely for Brute Force and about 22 times as likely for Barbaric Blow.
I've calculated the odds in this video and hope I'm not doing Mangs Math right now:
Mangs got five max rolls for Brute Force. For "one Luck roll" the odds for at least five max rolls in 40 engagements would be about .093%, for "Good Luck + Bad Luck" it would be an abysmal .0000002% - 1 in 500 million!
Mangs got two max rolls for Barbaric Blow. For "one Luck roll" the odds for at least two max rolls in 40 engagements would be about 6.56%, for "Good Luck + Bad Luck" it would be merely .0135%.
If my Math is somehow correct, theis showcases would already be pretty lucky if we go with the "one Luck roll" theory, while "Good Luck + Bad Luck" would be pretty much impossible.
The numbers displayed on the Medium Tanks also seem to be more evenly distributed, so it would support the "one Luck roll" theory.
Maybe it only felt like it due to people experiencing Flak’s counterattacks when the player strikes first? A weird placebo effect perhaps.
Oh maybe the counter attacks should be tested too!
I Don't think Flak is the worst co in the game.Flak is better than you think.
I think he is pretty strong for lower tier, atleast stronger than Grimm,Sonja & Koal .It's not only me (Go7) one of the best awbw player says that Flak is tier 3.5.
Then why isn't Flak Guy in these big tourneys?
@@WavemasterAshi Flak is banned for use in tournaments, as are all luck COs (so Jugger and Nell too) because the AWBW admin does not like luck-based COs. He's stated he doesn't want games determined by luck (outside of the uncontrollable default luck.)
@@WavemasterAshi birth of Flak Guy
Tweaking the game so Sonya doesn't have bad luck and that you know what your luck damage will be before you attack would be consistent with that.
I played that campaign mission where Olof is attacking a giant missile. I had 2 MDs blocking off the access to my other units the popped his supper and did 0 damage on both fronts. It was pretty bad.
Barbaric Blow - and two max rolls on 40 tanks seems ridiculous. But I did a few 100 attempts myself and shockingly enough: 1 out of 16 attempts reduced an Md Tank to 2 HP - which should only be possible with the highest luck value. In the GBA original that would be a rarity: If you need to achieve both max good luck and min bad luck, the odds would be 1/2400. Chances are you'd be getting bored long before it happens even once!
I suggest not only repeating the test, but also performing it on the original game. 3 HP should be significantly less common and you really need good luck (heh) to get a Md Tank down to 2 HP.
I think people didn't use to pay attention to these mechanics when they were kids playing the original on GBA and now that they watch CZcams content over-explaining these mechanics they are constantly on the lookout for good rolls.
So I ran some numbers...
tl;dr/abstract: It seems most likely from these results that they have changed the luck values to use 1 random number.
They may also have adjusted the luck values to cover all the values they refer to rather than being one short (so a +40 luck bonus would mean 0-40 rather than 0-39), however, this is much less clear from the data.
Either that, or Mangs is just incredibly lucky.
More research is needed.
Now, if we run the numbers exactly as they were in the original game as Mangs described at the start of the video, for the Brute Force Case we determine that the chances of getting a max roll of 40% =4HP of damage are 1/40 * 1/20 = 1/800 (Good luck has to be maximum value of 39, bad luck has to be 0), bearing in mind the 1 base damage so we need 39 luck damage.
Already, we see the issue - this means, on average, we should expect to need 800 attacks to get maximum damage, which... we don't have anywhere near. But to be exact, the chances of getting at least 5 maximum rolls in 40 tests is 1.94E-09, or 0.00000000194
For the Barbaric Blow case, the chance of a maximum roll is 1/2400 and the chance of getting that at least twice in 40 tests is 1.34E-04 or 0.000134 - this is definitely a lot more likely than our , but the chances of both these sets of events occurring is 2.5996e-13, or less than 1 in a trillion.
So something definitely appears to be different here. There are three possibilities:
A. The luck values are now 'corrected' so they vary up to the maximum value indicated, e.g. a luck value of 40 goes from 0-40 (rather than 0-39)
B. The luck values for bad luck and good luck are combined as 1RN (this has been suggested in other comments)
C. Both of the above are the case
Case A would mean that there are 3 chances for success in each of our examples, rather than 1 over a slightly larger (but not 3 times larger) range of outcomes. E.g. for Brute force, a maximum value of 40% would occur in Mangs' example when the Good Luck and Bad Luck values were any of the pairs - (39, 0), (40, 0), (40, 1), while there are now 41*21 = 861 outcomes.
Case B simply flattens the curve, so for Brute force the chance of getting a total of 40 damage become 1/59 [as you pick a random number from -19 to 39, including 0]
Case C both flattens the curve and increased the number of successful max damage rolls (this time only to 2) and slightly increasing the number of possible outcomes - in the case of Brute Force, you would need to roll a 39 or 40 and the possible numbers you could roll are -20 to 40, so there are 2 good outcomes out of 61 [so chance is 2/61]
Using each of these cases, the chances of getting each of the outcomes shown in the video (5 Max rolls on Brute force, 2 on Barbaric Blow, each out of 40 attempts) are:
A: Brute Force - 3.05E-07, Barbaric Blow - 0.00108
B: Brute Force - 5.55E-05, Barbaric Blow - 0.0522
C: Brute Force - 9.59E-03, Barbaric Blow - 0.162
It therefore seems very unlikely that scenario A is the case, and much more likely that B or C is the case.
In conclusion, I therefore posit that the luck calculation method has changed to 1RN. Whether or not they have also extended the range of luck values to include the endpoints, I do not feel we can conclude, but it seems somewhat likely.
I would also caveat this with three points:
First, even if I've done all the numbers correctly, this does not 'prove' the system of calculation has changed, just strongly hints at it. Even in case C, the chances Mangs would get results at least as good as he got were about 0.00155358, or a little over 1.5 in a thousand - he got pretty good luck (haha) regardless of what the system used was. In order to make a better conclusion, more data would be required.
Second, if there was anything else affecting the damage calculation, the above numbers would get thrown off wildly. As we saw, just allowing the damage from luck to be 1% lower made the event 3 times as likely - from what I understand the mechanics of the game (I am by no means an expert), there should have been no other effects changing the luck value (as Andy has no defence bonus and the tanks were on roads), but if we even had a hidden +5% bonus to damage suddenly we have 21 'success' cases occurring.
Third - technically in the GBA games, as I understand it, you were very slightly more likely to get lower numbers for your luck values than the higher values. This is because the luck value was determined by taking an unsigned 32-bit integer and finding its remainder when divided by the luck value (so for most COs, this is just the last digit of the number). As a 32-bit number takes values from 0 to 4,294,967,295, 0-5 would be very slightly more likely to appear than other numbers. This is so tiny a difference and makes the maths so much more complex that I omitted it in the calculations. Most likely, with a modern console, they would just call a standard rng function between limits (which is likely also biased, but may be in a different way from the original).
For reference (and in case anyone can spot an issue with my methodology), the formula used to calculate the chances of getting *exactly* X max rolls is:
P(Maximum Damage Rolls = X) = (S/Q)^X * ((Q-S)/Q)^(T-X) * T!/(X! * (T-X)!)
Where
Q = the number of Possible rolls of luck
S = the number of those rolls resulting in a maximum damage value
T = the number of tests (40 in every case)
...thank you for coming to my TED talk.
That sure was some thorough math speak there, m'dude.
A good way to power up his super is to make it a 50/50 roll between the lowest and the highest roll. That way theres a chance of floundering or getting that heavy blow you are searching for.
I actually have a suggestion for how they can buff Flak in a way that makes sense without changing what they do. Make the good luck rolls and bad luck rolls take precedence over another. What this means is that Flak will roll for his good luck, and then roll for his bad luck. Whichever number value for either positive or negative is higher, it will then only add/subtract from that attack. If he rolls a +7 for good luck, but then a minus 5 for bad luck, instead of adding both values together, the highest number overrides the tally: now his attack will deal +7% extra luck damage. On the flipside, if he rolls a minus 9 on another attack for bad luck, but rolls a +3 on good luck, the attack will now deal minus 9% damage because the bad luck roll was higher than the good luck. If both rolls are in equal value, then the attack will be an unmodified attack with an overall 0% increase/decrease in power. All of this would carry over his CO and Super CO power, of course, and implementing this change should be incredibly easy. This will make Flak's power feel a lot more impactful and won't result in an awkward spot where his good luck roll and bad roll cancel each other out, making his power effectively non-existent for the most part. Flak's power will now be meaningful in some way. He'll still be one of the worse COs out there and still ultimately be a worse Nell on average, but now, if he's lucky, he can now hit harder than Nell ever could- albeit at a risk.
Brute Force? More like Brute FARCE! Flakalak's Super is called "Barbaric Blow" because he Barbaric BLOWS.
So if there are changes to the Reboot camp will that be reflected on the website versions?
Doesn't look like there are any changes to the multiplayer side yet, but do you have a plan or know of one if they do end up existing?
I might do a statistical analysis video on this since there are some very interesting things going on here
Man they nerfed him so much, they reworked his indirect bombers and gave them 1 range and they also took away his battleship movement on rivers
0:27 *best*
Flak Bros rise up!
The way that Flak's Luck values are shown is a big case of "Programmer Logic".
In most programming languages, if you need a random integer from 0 to X, the way you usually calculate it is to write something like Random.nextInt(X+1). So if you need a number between 0 and 14, you write Random.nextInt(15). On the GBA, where CPU cycles were at a premium, it's even possible the programmers just used something like rand() % 15, which would also give a number between 0 and 14 (though slightly unbalanced, because modulo only gives uniformly distributed numbers when the divisor is a whole power of 2).
Either way, it's easy to see why they're writing it like "-10 to +15": it conforms to the underlying logic they'd have to had implemented when they were writing the game's code.
I think some people simply have a condition that I’ve called “0 in 1 rng” where they get either the best or worst rng possible with no in-between for several games in a row, I happen to have this and the odds can very quickly go from above average or normal, to below a 0.0X% chance [x being numbers in the 0.09- range] very consistently. It can be very frustrating seeing this multiple games in a row...or it was legitimately bugged lol
How about Drake? Has he actually gotten a buff? Or is it just a mistake in the CO description? I still haven't been able to find the time to beat the advance wars 2 campaign, so I can't test it out...
Drake is 100% unchanged.
The description is wrong. If you go to gallery, to COs, and go to Drake's third page, and you swap between style 1 and style 2, you can clearly see his air units have worse attack in style 2 compared to style 1.
Fun Flak: In the German Version his Name is Helmut. :D
Maybe is just Percive extra damage?
As most time will be a half health tank, so it getting lucky and causing 20-30% more damage can be a surprise.
As expose here, flak power are very underwelming, even super has like 40% chance of doing nothing, so it kinda surprise you welhen it does.
this is the video that i was waiting for
It could just be like fire emblem where they didn't change luck THAT much but people think it changed a ton because of their experience (with full hp)
In the German version he is called HELMUT.
this is a great video about how worthless flak is,hes basically like playing with co powers off only you sometimes get a lucky hit like 10% of the time.
brute force netted 53 hp worth of damage on those medium tanks. 84800 worth of funds gone from brute force. done with less than half that in value of units, in one turn. While it's definitely not the most powerful, Flak's the only one besides Nell who can get such a value return like that.
barbaric blow got 85 hp worth of damage. Only an extra 32 over brute force. That's still 136000 worth of funds.
seems like brute forcing multiple times and ignoring barbaric blow altogether might be a better option
Mangs' head is disappearing in this video. Truly a wild night.
Total damage of 51
Total health of md tanks 400
Aaverage of 1.275 hp.
However, we know 1hp is actually 10% and it is possible a lot of those numbers are up to 5% more (assuming reboot camp rounds up 9.5hp to 10hp etc).
40 x 0.5 (for the potentially hidden extra 5%) = 20
Flak potentially had 71 damage which is 1.775 hp on average.
I don't know if this is good compared to other co's in the same scenario but 17.75% of a medium tank that costs £16,000 is £2,840.
We could take away the damage cost to the infantry on average (but i am not doing that maths) to see how cost effective it actually is, but assuming 5hp counter attack (£500) for example, means a potential net profit of £2,340.
12.75% is £2.040 btw, and using the same 5hp damage that would be £1,540.
Infantry attacking md tank during super co power is somewhere between £1,540 and £2,340.
The average between the best and worse case scenario for the average is £1,940.
The absolute best in the video was 45% damage, meaning that engagement is worth £7200 (-£500 for counter) = £6,700.
Of course, the worse case scenario was 0 damage and 9hp lost = -£900.
TLDR; Flak infantry attacking a md tank during a super co power will range between 0hp and 4.5hp meaning a net cost of -£900 and £6,700 (assuming 5hp of damage to infantry).
Edit; this was for co power NOT super co power!
Super co power did 85 damage.
Maximum (including 0.5 hp damage (20hp) of 105hp damage.
105hp/40 md tanks= 2.625hp of damage on average.
£4200-£500=£3700 profit on average.
Max would be £12,800 and that results in 2hp return damage iirc, meaning a profit on £12,600.
Definitely cost effective on average but this experiment does not really show the bad luck side where flak will do less damage than usual as an infantry is expected to do 0hp to a md tank, meaning we only see the "positive results" so keep that in mind. Flak is actually worse than what the video shows because the bad luck blends in with the other results.
Pretty sure even 9.1 HP gets rounded up to 10.
@@EpicEverz 0.1% rounds up to 1hp so i think you are right!
From playing the game myself, a unit with just a sliver over 5hp shows as 6hp and 8hp shows as 9hp.
Not changing the maths, but well spotted!
I think this experiment would have more readable results with infantry versus infantry or tanks. Same conclusion though; flak is bad!
Mangs doing math? NANI!?
Honestly probably just confirmation bias, Flak gets a couple of lucky rolls in a playthrough and those stand out in people's minds more than all of the mediocre/bad rolls so it feels like he's doing more damage than he should
Can you talk about the game crashing when you activate a co power?
Potemkiiiiin BUSTAH!
Oops wrong character.
Man this guy need a ton of re works, there is no different than having no ability and can nerf it self.
Are those players perhaps complaining about AI Flak (who seems to know when his luck roll will do good?)
Someone should put the values into an Excell spreadsheet and graph the results
Flak should become a form of gambling
does luck reroll on restarting the turn or will that always stay the same? i never knew if you could keep restarting till most of your units land a decent roll
Mangs can you perhaps look into the weird "Indirect Bomber" glitch that seems to be happening, sometimes the AI (mostly Sonja it seems) can strike indirectly with direct units.
I've had a Sonja tank attack me from 4 tiles away when she used her power in the mission Sonja's Goal
Current theory is that when she uses her power and extends her field of view during fog of war it causes the issue. It is most noticable with bombers due to the extreme move range.
Don't listen to his anti-Flak propaganda. Flak is love. Flak is life. Flak forever MVP in our hearts!
The thing I find so odd about luck COs, is that everyone looks at what an infantry can do to the strongest tank in the game... When in reality, there are far better units that take advantage of luck damage, and would be far and away better with a CO like Flak. Indirects, Bombers, Battle Copters, and any of the big guns like; Neo Tanks, Medium Tanks and even just regular Tanks honestly. Yes, you can score devastating hits with infantry. However, even with Nell this is a risky and quite noobish gamble. Losing your infantry means losing your capture game, even if you take a few crazy units out.
I understand why Mangs is using infantry here to demonstrate this concept, but I'm not talking about this demonstration. I'm talking about the folks who do this legitimately.
Does someone here know how to change your units color outside the story? Every CO I pick has red units.
explanation is easy. people just have a biased memory towards good rolls lol
Even if Flak's abilities were decent generally, wouldn't they still be bad in practice anyway? Can't really form proper strategies on RNG
i think we all wanted flak to be buffed, its sad to learn the truth.
Flak pretty much exist just to be a "bad cpu" CO and the devs just let us play him for fun. Like you could handicap yourself picking flak when you play vs your siblings or something.
i have never noticed that the screen shakes when you land a lucky hit.
Honestly I think the test would have been better if you did md tank vs md tank since infantry in general don't do alot of damage to md tanks.
I like how even hachi tells you how using flak in war room would be a really bad idea... Unless you're wanting for a challenge.
Do you think you could do another little test video on using CO powers with extra charge stored in the bank? Like having your Super ready but only using the Little Power, and seeing if that actually costs extra like you say it does in your recent Secret Tips video (Tip #4). I strongly believe this to be in error, and would like to see it given official testing in the Reboot Camp.
To clarify a bit: Tip #3 in that same video suggests (I believe accurately) that additional charge is required after every use of the CO Power (normal or super). As such, if you had six stars charged at 100 points apiece, and used a 3-star power, you would normally expect to see three full stars once your power ends, but each star now costs 120 points to fill, so your 300 points fill two stars with only 60/100 on the third star. This makes it look like you lost 3.4 stars, and thus 340 points, but in reality you only lost the 300 you expected to spend.
If any CO in the game has exactly 6 more Super stars than Little stars, the math works out perfectly to go from precisely 6 stars remaining to precisely 5 stars after the very first use of the power in that game. I can't remember if any COs have this, but the first one I'd look at is Eagle.
Assuming my understanding of the luck rolls is correct, in order for Brute Force to get the Infantry to deal 4HP of damage to a MD tank, you'd need to roll a perfect 40 on the Good Luck and no Bad Luck. If every number has an equal chance of being rolled, that would be a 1/40 (chance of rolling max Good Luck) * 1/20 (chance of rolling 0 Bad Luck) = 0.00125, or something you'd expect to see about 0.1% of the time. Any other result would do less than 40 HP of damage.
Advance Wars does not round HP values down, not to the best of my knowledge. If the display does round down starting at 65 HP, it makes it more complicated - if you get a 35 Good Luck and a 0 Bad Luck, it will take the MD Tank to 6, then 36 and 0 or 1 will work as well, 37 and 0-2, 38 and 0-3, 39 and 0-4 and 40 with 0-5. This does dramatically increase the odds of doing 4HP of damage on-screen, but it's still only 17 possible rolls out of a total 800, about 2.125% Hell, let's get ridiculous - if the HP rounds down to 6 starting at 69 HP (meaning 31 damage) then Flak has 45 possible rolls to get that threshold, or a 6.875% chance of displaying 6HP. This last number is, of course, absurd, but when I compare it to what is seen in the video then you see why I chose an absurd scenario like that.
But assuming the first formula where the chance of 40 HP of damage is 0.00125, out of a random sample of 40 MD Tanks you would only expect one of them to be taken down to 6HP.
In the test we see 5 Brute Force Infantry do the full 4HP of Luck damage on 5 of the 40 MD Tanks, or a rate of 12.5% which is NOTICEABLY higher than even the one using an absurd scenario where 69 rounds down to 6. To be precise, using the first estimate (no rounding), the statistical probability of Flak doing the full 4 damage 5 or more times out of 40 trials is 0.00000000193619 - or, as you might say, "practically impossible". With rounding his 35 or higher rolls to 40 it's 0.15% (meaning not something you'd ever expect out of a single random sample but far from impossible).
So the luck really doesn't make sense to me but I think the test here shows a sufficiently STRANGE result to believe it's not just "he rolls both luck values independently, with all numbers equally likely to come up, and then it just applies as he rolled them."
@Mangs From what I've seen, people are talking about Flak's damage fighting him in the campaign, not multiplayer: for whatever reason, Flak's rolls didn't actually seem like true random in the campaign (in fact, I'm pretty sure they were outright fixed,) so people didn't see much, if any, extreme damage. Now, Flak's RNG is consistent, so people are actually seeing Flak get the same nutty damage rolls they would with multiplayer.
Mangs hitting us with flashbangss for 12 minutes