Optimum Axis Strategy

Hi!

It seems to me that most people (who still play Axis and Allies 2nd ed) have resorted to balancing the game out by using RR and bids.

The reason for this is by using the “Optimum strategy” by both parties, the Allies are nearly guaranteed to win. I think the consensus here is that it’s supposed to be 90%-95% chance of Allied victory (without RR and no bids).

The allied strategy essentially involves UK building AC + transport, USA flying its fighters over, then UK and USA builds masses of infantry to kill Germany first. Of course, there are different tactical points like (should Russia attack Manchuria first or not?, should UK build IC in India ? (and only AC 2nd turn etc). However the overall plan is pretty much the same.

The so-called “optimum Axis strategy” on the other hand is pretty much Germany build infantry and hold out Europe, hoping Japan kills Russia before Germany falls.

Unfortunately, after deciding that Axis has a low probabilty on winning with this strategy, much future research/discussion on the game has gone into strategies in RR and bids.

However, imagine if there was a match between two top-notch Axis and Allied players (no RR, no bids). The scoring system is to be similar to tennis without the tie-break. Ie: One person takes Allies, then the other. Winner is first to 7 victories, unless it’s 6-6. In which case it has to be 8-6, 9-7 etc.

If so, wouldn’t the Axis player try sometime more creative than the so called “Axis Optimum strategy”? Basically to try different risky strategies and hope for lucky rolls, since the probability of winning a long game is so low anyway. Of course, 50% of the time the Axis may get crushed in a few turns, but so what?

In my opinion, the “optimum” Axis strategy would be to mix “risky” strategies over several different games, such as:

1. Germany buying 1 or 2 Tech rolls + Infantry each turn
2. Germany buying bomber + Infantry in first turn, infantry subsequent turns (trying a “lucky” G2 attack on Navy).
3. Japan bringing an early transport and try take Brazil, followed by IC (if USA ignores it).
4. Japan moving fleet below South America and go for a lucky shot at the allied navy.
5. If England builds an IC, bring entire Japan navy to India.
6. Japan builds transports + infantry, followed by early tech rolls and infantry.

Surely creative thinking and playing riskily should yield a better result than the supposed 95% allied victory (in no RR, no bid). For example, say Germany goes 2 Tech roll + Infantry in the the first turn. The probability of getting at least one Weapons Development is 30.6%.

Then the probability of getting Industrial relief is 5.1% (and 5.1% also for heavy bombers). Of course, even if Germany gets either it’s by no means a victory. But I believe it demonstrates that Axis can surely do better than the supposed 90-95% probability of allied victory.

I would say that Axis could be able to achieve at least 80% (even without RR & bids), through risky playing and inducing mistakes. But I was wondering what people think.

Notes:
By inducing mistakes I mean that through changing strategies, even a top-notch player would lose track of all the required probability calculations that maintain the 90-95% chance of allied victory. If the allies allow Germany a risky move that has a 80% chance of quick Axis loss in three turns, but a 20% chance of victory - then Germany should take this chance.

@Bashir:

I think giving the Axis 90-95% is really optimistic if the allies are skilled enough. I would say only if the Allies get dicefucked during the whole game the Axis has a chance of winning.

The systems you describe are either based on luck or incompetence of the allies player. In the second scenario it would be much wiser to play the optimal Axis strategy than playing a risky strat.

1. Based on luck, in this scenario only heavy bombers, industrial advance and maybe long range aircrafts are useful, but wasting 3 inf every round really hurts germany…
2. Again based on luck, what is the fun in playing AA if you rely on luck? Go play risk…
3. Wasting a transport will delay the Japanese on the Russian front, spending 15 IPC’s more on a useless IC on Brazil is even more worthless. The fact that the allies see the transport comming for 2 rounds does not help a lot either
4. Wasting your Japanse fleet for 4 turns to move them to the Atlantic… I rest my case…
5. England buying an IC for India is not that bad for the axis, it delays the allies in retaking Afrika while they need to spend a lot of rescources to defend the IC in India (which will fall inevitably)
6. I hate to brake it to you, but Japan has only 25 IPC’s to start with, you simply do not have money for tech rolls in early game!

5 of the 6 scenarios which you describe will only shorten te game significantly. Ow if there is no RR or no bid, Rus attacks Manchuria and UK will attack Kwantung, leaving the Japanese in an even bigger hole…

Playing without bids or rr (+ bid) is just no fun… To win as the axis you have to get really lucky and that is just the reason why I stopped playing risk.

Greetings. Thanks for the comments.

I think you totally misread my post. I said that the general consensus is a 90-95% probability of ALLIES winning - ie: Axis has almost no chance when Allies are playing with an optimum strategy.

Perhaps that’s why you didn’t understand the idea I was proposing. I’m assuming two top notch axis and allied players playing No RR & No bids. They will be playing a format where one player takes a turn each of allies, then axis. Assume player A and player B. First to 7 games, where you need at least a two win difference (8-6, 9-7, 10-8) etc.

Both players A and B would obviously play the optimum Allied strategy (as discussed). However, for Axis:

Player A plays the textbook axis strategy of German infantry and hold Europe, Japan attacks Russia EVERY Axis game. Player A expects 5-10% chance of victory only every axis game.

Player B plays a different risky strategy each axis game. Relying on luck and trying to confuse the opponent (into losing track of probability calculations that continue the 90-95% allied victory).

I believe Player B will win more “sets” overall.

If you understand this scenario, then you’ll see what I’m saying. Why isn’t the optimum Axis strategy to go for all-out risky attacks or tech roles? A loss is a loss. It doesn’t matter if it’s a quick loss after Germany loses it’s entire airforce, a loss after Germany doesn’t buy enough infantry or a 15 turn grind, played with textbook strategies.

The strategies I proposed ARE based on luck - that’s the intention! And no, they are not based upon the incompetence of the allies player. It’s based upon the fact that agains a top-notch allied player, the axis player using the textbook strategy has almost no chance anyway. Ie: Axis hopes to be REALLY Lucky for a few turns, rather than hoping to be Lucky for many turns.

Lastly, I think it’s unfortunate people gave up on creative playing. I mean in Tennis, it’s not like it’s 50%/50% chance for the server - which is WHY it’s such a great game.

Comments on the Strategy notes:

1. Strategy 1 IS meant to be based on luck (tech roles)
2. Strategy 2 IS meant to be based on luck (Germany destroying UK navy without many loss of planes).
   3 & 4) These two strategies are basically to bring the Japanese Navy to a risky attack in the Atlantic. USA can easily prepare for this, but again, Japan can get lucky. Besides, Japan can’t do much with its fleet anyway.
3. I think IC in India both has good/bad points (I’m not going to argue for/against it). But I’m saying that if Allies do choose to build it, Japan should try a RISKY early attack against it.
4. What’s wrong with this? J1 buys 2 transports 3 infantry. J2 is one transport + infantry. Japan only buys tech around turn 3, when he has enough transports. Again, Tech roles rely on luck - but that’s the point!

@Bashir:

I understood perfectly what you trying to point out, but the point I was making, why do you want to play with these rules? If your are relying on luck you should stick to Risk imho. Axis and Allies was designed as a tactical game and not yatzee… And top notch players know how to handle those kind of strats, if Germany goes for tech, the US will do some tech roles as well, because in no RR no bid games Germany is under a lot of pressure from Russia alone and if they spend money on tech the US can easily spend IPC’s on tech as well (and more than Germany)

The risky Japan moves are quite useless, because you need your navy to assist your ground push in the west. Ok say you want to move your fleet to the atlantic, and Russia attacks Manchuria on R1 and the UK attacks kwantung on U1, you really need your navy to assist taking it back and establish a front in the west.

Your Japanese tech rolls on J3 will make Germany crumble! They will be under a lot of pressure and you are nowhere near Moscow, so again the Allies can easily counter your tech rolls with theirs…

I think with top notch players your winning% will drop below the 5% winning rate you might archieve with the Axis.

LOL… I still don’t think you understand my concept:

My post is about discussing the optimum axis strategy in no bids, no rr.

1. I’m not arguing whether Axis & Allies should be played with these rules or not. It is clear that many people prefer playing with bids -  Personally, I think bids is great as it adds a alot of variety and depth to the game

2. I’m also not arguing whether Axis & Allies should have a strong element of luck or not. I myself play chess at a high level.

I’ll give you another analogy of what I’m proposing - I shall refer a lot to chess as this game doesn’t involve rolls.

Imagine if you had a supercomputer which could somehow calculate the exact probability (at any stage of the game) of whether the Axis or Allies would win. In addition, the supercomputer would suggest moves that maximes the probability of winning.

It is my impression that this supercomputer would play allies almost identical to the standard textbook allies strategy. KGF with US doing shuck shuck. As I mentioned before, there are different tactical points but I believe the overall strategy would be the same.

However, I would think that the supercomputer would play Axis VERY differently to just always following the normal textbook Axis strategy. Let’s look at KwangBang as an example.

Imagine if Japan gets terrible rolls and Russia takes Manchuria and UK takes Kwantung with little or no loss. Also assume USSR has killed the Baltic fleet and strafed Ukraine (with standard results). Also assume Germany destroyed the UK fleet for the loss of 1-2 planes and didn’t do a suicidal attack on Karelia.

What do you think the supercomputer would suggest for the axis? I very much doubt it would follow the textbook Axis strategy. Why? Because even if Axis were to play “properly”, Axis is dead anyway. Set up the board at J1, with Russia already occupying Manchuria and UK already occuping Kwantung (with little loss), plus typical results of other USSR, Germany and UK moves. Try playing Japan using “textbook strategy” against top Allied players (or yourself) and see how many games you win. I would say close to 1%. Ironically, Axis should play even riskier if KwanBang really hurts Japan. My comment about Japan in the Atlantic is if KwangBang fails (or allies chooses not to do KwangBang). If KwangBang succeeds spectacularly, Axis might as well go for tech rolls or hope for a miracle with a G2 attack of Navy (and resign if it fails).

Conversely, if KwangBang/Baltic/Ukraine/Navy goes very badly for the allies, I believe the computer would suggest for Axis to play completely “textbook style” as the optimum play.

In chess, people have managed to program a 6 piece Table base (takes 1.2 terrabytes). This means that a computer implementing a 6 piece Table Base can play such endgames PERFECTLY. For example, if a computer was playing White in a (won) 6 piece endgame like King, Rook and Knight vs King and Two Knights, it may announce eg: White wins in 246 moves. If the opponent plays “PERFECTLY” as well, this number counts down by 1 each turn and the computer wins. If the opponent plays a “poor” move, he/she just loses faster.

What people found was that many Table Base Endgames were completely incomprehensible and broke “strategic” principles of chess (that were developed over hundreds of years). But what was particulary interesting was chess endings that for hundreds of years were considered “dead draws” were actually winnable.

Similarly, I believe if a supercomputer could ever be programmed to play Axis such that it always maximises its probabilitiy of winning (of course not possible), it may play seemingly ridiculously. Buying bombers, tech rolls, suicidal attacks etc. It may even lose games very quickly (and play like a “beginner”). Because the computer wouldn’t care whether it would lose in 5 turn or 20 turns - it would only care about probability.

Of course, I cannot say for certain how the supercomputer would play, but personally, I envisage that the supercomputer would try increase the luck factor if things are going badly, and reduce the luck factor if things are going well. If the supercomputer plays allies, I believe it would try minimise “risky attacks” and simply prolong the game and grind the opponent down. If it is Axis, I believe it would choose different strategies based upon the current situation. If Axis are in a “dead” situation, it would probably go all-out luck, or if going well, it would buy more infantry etc.

In the same way that the Table Base showed how the human chess “textbook” ending strategies were flawed, I believe that the supercomputer would also show that the human axis textbook strategy is flawed. Tim Krabbé wrote about Table bases:

“A grandmaster wouldn’t be better at these endgames than someone who had learned chess yesterday. It’s a sort of chess that has nothing to do with chess, a chess that we could never have imagined without computers. The Stiller moves are awesome, almost scary, because you know they are the truth, God’s Algorithm – it’s like being revealed the Meaning of Life, but you don’t understand one word.”

Secondly, it is my impression that the computer would announce at the start of a no bid, no rr game that the probability of Axis winning would be closer to 10-15%. If a top-notch human player would play against it, perhaps the supercomputer might win up to 20% of its games (as the human won’t be able to determine exactly what maximises probability).

The whole point of my post is that I believe textbook axis strategy maximises the length of the game, and not necessarily the probability of winning. The problem is that people would rather lose in a “hard-fought” but close game than being crushed in a few turns. If you don’t care how you win or lose, then I believe there are more optimum ways to play Axis.

I believe the optimum Axis strategy isn’t just a single strategy, but multiple different strategies. At one end is the textbook Axis strategy, at the other end is buying multiple tech rolls. In between are a mixture of strategies like infantry/tech roll, infantry/bombers, G2 attack on Navy etc. Axis chooses which way to go depending on the current situation. Clearly, if the game is already lost, why bother buying infantry just to hold out another 5 turns? Alternatively, if something miraculous happens like Russia getting crushed in Ukraine, Baltic and Manchuria - Axis should probably play “textbook”.

I wrote this post to see what plausible strategies there are for Axis. Obviously, something like G1 1 battleship + 1 Armor / J1 battleship is rubbish, but G1 5 Infantry + bomber or 7 Infantry + tech /  J1 3 inf + 2 trans seems plausible to me.