Hey guys, it's Garretz here, and I'd like to discuss today a topic I love - game theory, specifically around one of the game’s main concepts, Snapping. More specifically, the relationship with snapping. By that, I mean snapping to double the stakes, not Marvel Snap in general. Before we get into it I’d like to point out that currently there are many bots on the ladder. In bot matches, all we're going to discuss today isn't applicable, since their behavior differs from players.
The Odds
So what I came here today for is to tell you that you might be losing cubes by snapping. It's clear that when you snap and lose a game, you lose cubes from snapping - you doubled the stakes, and lost. But that's not the only scenario. After the last turn of the game, the stakes are doubled, which means that when you win, you gain more when your opponent doesn't retreat. If by snapping, you're forcing a retreat, that means you are earning 1 less cube, and essentially that snap just costs you 1 cube. In the last turn, this is a very likely scenario. No one snapped, no one is retreating, you're ahead, you feel confident, you snap. If this was going from 1 to 2 cubes, your opponent could be willing to keep playing. Going from 1 to 4, it's way less likely for your opponent to stick it out. Your expected value (EV) is always a sum of the EV of every different scenario. To expand on this, let’s go into some quick math. You need the odds of every scenario multiplied by the value you gain.
Without snapping:
Odds of retreat * 1 - odds of staying and losing * 2 + odds of staying and winning * 2
Snapping:
Odds of retreat * 1 - odds of staying and losing * 4 + odds of staying and winning * 4 - odds of getting snapped and losing * 8 + odds of getting snapped and winning * 8
It's very hard to determine the odds of them staying, snapping back, or even the exact odds of winning, so getting deep into the math is no use for us. What we can intuitively understand is how the general values get changed by snapping: The odds of them retreating get higher. The odds of you losing provided that they called the snap also get higher: since they accepted the snap, they must have a strong hand. The odds of you losing if you get snapped back get significantly higher. A player who is losing gets snapped and snaps back is very likely to have an extremely strong hand.
If this was too theoretical, let me provide an example. A few days ago I was testing a card in a deck of mine. We had Sinister London on the left, a middle location without relevant effect, and a locked-up losing The Space Throne for me on the right. The image below provides the whole situation:
My opponent felt confident. Their last turn play was a Death, 12 to both locations they were already winning. That would bring them to an advantage of 25 points on Sinister London (22 considering ant-man active) and 21 (18 considering ant-man) in the middle. They snapped. What could go wrong, right? Three slots total, I had to swing two lanes because the third one was lost, they should snap!
Well, they didn't need my deck list to know not to snap there. You know, I'm losing hard, very unlikely to win. They're snapping, they probably have something to play. I will retreat here every time unless I have some ace up my sleeve to turn this game around. They reveal first, so anything big will also get killed in Shang Chi's reveal. Cosmo and Armor are the 2 ways out they have from this play of mine. So I snap back, clear their two lanes with Shang Chi, and triumphantly walk out with my EIGHT cubes.
There are some players out there that would keep playing even in a sure loss, this is why we always talk about the odds, never about deterministic behaviors. Their snap didn't change much concerning the odds of me retreating, what it did change was the odds of me winning the game provided that I accepted the snap. Also how much they would lose, given the fact that if I accepted, it was far more likely for them to lose. After me snapping back, that was almost certain.
This is another risk of snapping in the late game when ahead. What will you do if you get snapped back? Because they are saying they have that golden ace up their sleeve. Will you be bullied out of the game? Losing 2 cubes is better than losing 8, but it's uncomfortable nonetheless. After all, it was YOUR move of snapping that put you into that situation.
With that we will jump to a second topic, which is just as important as snapping, to maximize the cubes you gain as you try to climb the ranks - that is retreating.
Retreating
First, we must establish the two forms - immediately or later. I was deceived the first few times I saw the button “retreat later” with that asterisk saying it ties if the opponent also retreats, thinking to myself where was the catch. There is no catch. Retreating later is just the same, but the game ends in a tie if your opponent also retreats, which is better than getting a loss. What the game does is that it waits until your opponent also submits their play to submit your surrender, allowing for a “simultaneous surrender”, which would be a tie. So, if we’re talking about maximizing your cubes, you should ALWAYS retreat later. The only downside is having to wait a few seconds until the opposing player confirms his move.
Now that we have decided which button to press, the second question is when do we press it? It impresses me how people simply won’t retreat from a normal game. I just went over not snapping so that you won’t scare your opponent away from the doubling cubes in the last turn. Now I tell you not to be that opponent who won’t run. It's clear that if you can’t even beat what’s on the board, you should cut your losses by retreating, but that’s far from the only scenario when retreating is correct.
Let’s talk first about the later turns and getting snapped. If you’re getting snapped on the last turn, you can expect some heavy bomb coming your way. Especially if they played Moon Girl earlier, a huge combo is incoming - so you should just bail unless you can match it. Whatever their archetype is when they snap, it’s safe to assume they’ll have their finisher, whether that is Iron Man, Devil Dinosaur, Nova + Carnage + Bucky Barnes, Shang Chi, or Odin.
Of course, it’s not so easy and safe to assume archetypes to pinpoint finishers, but you can have some sort of direction based on what they already played. If they played all their hand and just top-decked, it’s pretty safe to assume they built the deck to do that and play America Chavez in their last turn. In this game, an average good move is 2 points per energy spent, so when getting snapped, you should be prepared for something better than that. If you can recognize the incoming threat, sometimes you can play around with it. One of the ways we can combat this is giving up the lane or matching the pressure on the lane where it will probably be played.
When you’re not getting snapped, you have less information to work with, but you still have some evaluation to do. Just beating the board is very likely not enough. You can just beat a lane where they have 4 units since it’s harder for them to put points there, but I wouldn’t go into the last turn where, looking at the board plus your move, you end up winning by 6 points. Many bad hands can easily put out 6 points on the board on turn 6, so I wouldn’t consider that a safe margin.
Of course how many points above your opponent you need to go depends heavily on what they played, if they improved their hand (Moon Girl, Nakia, Okoye), how many slots are left on the board, what locations are in dispute, what locations are locked. But saying no special scenarios, I’d say you need at least an America Chavez worth of advantage considering your play and not theirs to be willing to go on the last turn. Otherwise, you’re setting yourself up for some bigger losses. EV calculations say you need at least 25% odds of winning the last turn to be worth going for it. That may seem like a low bar, but remind yourself that the decks are very consistent in this game, so you’re very unlikely to be facing a dead-ish hand.
Now the harder choice - when you’re snapped in the middle of the game. It’s so tempting to just stick to the game because you might just draw into your cards… Well, not quite. You see, by turn 3, you have drawn 6 cards, and you’re going to draw 3 more of the remaining 3, giving you an exact 50% odds of finding any single needed card. The problem is, by snapping on 3, your opponent just told you he found what he needed. Whether that’s a desirable location, a full combo-oriented Moon Girl game plan, or any dream hand.
They are confident enough to say on turn 3 because they think they’re going to win. By calling, you’re immediately raising the stakes to two cubes, and you can’t win unless you call once again on the last turn. The math gets more complicated here because you could be expecting to hit another card in the next 3 turns and just retreat if you don’t. But to simplify the math, let’s consider you’re going straight for the 4 cubes because they are very likely to stay.
In this scenario, EV calculations say you must have 37.5% odds of winning to call. Remember, though, that this isn’t just 37.5% odds of winning a game. It’s having this odd of winning a game when your opponent snapped, which means you must almost match a snapping hand. In more concrete terms, that means that if you have Carnage in hand, you can only accept the snap in this position if Nova is probably going to win you the game if he’s drawn. Saying you’re in pool 2, playing Bucky Barnes + Nova + Carnage, you need at least 2 of those 3 to even think about accepting. The decks are ultra-consistent in this game. If you don’t have your game plan mapped out and you get snapped, go next. Your odds will be better there.
Conclusion
So then, what is the conclusion in all of this? What is the easy, applicable thinking we can use in our games for our benefit, considering this complexity in snapping? Well, it's quite simple. To get your opponent on board with you, you should snap earlier, and you should snap based on your hand more than your onboard advantage. Your opponent can gauge their ability to beat your board advantage precisely, while with your hand they can't do that as much. Also, when you snap, you should think not that you beat their average hand, but you should beat the hand that calls or raises your snap.
Alternatively, when getting snapped, or when going to play the last turn, remember that you must beat the opposing hands that would do that move, otherwise, you should cut your losses by taking a retreat. Also, you don’t have to keep insisting on a mistake. If you accepted a snap, but things didn’t turn your way, feel no shame in retreating before the last turn. It’s always wise to diminish your losses when possible.
All of this also brings another move to the table for you. If your last play is completely awful, and you think you'll lose, but you're ahead, you could try to snap to press your opponent out of the game. Be very careful, though: That is trying to get 1 with the risk of losing 4. It could be better than taking your chances on 2 at a normal game, or your certain-but-smaller loss if you retreat immediately. In the end, it comes down to considering your hand and board state to find the move with the highest EV.
