One of the most interesting aspects of Magic, in my opinion, is knowing when you have to play to win and when you have to play not to lose. Playing to win the game means you make the decision that is most likely to, well, win the game. Playing not to lose means you make the decision that is less likely to cause an immediate loss. A lot of people do this intrinsically, and a lot of people have no clue they even have a choice on the matter - those are the people I'm writing this article for, because the choice between the two begins even before you show up for the tournament.
As soon as you are choosing the deck you are going to play, you are going to be faced with the choice between winning/not losing. Now this is somewhat of a personal opinion, but I generally prefer playing a deck that does not lose rather than a deck that wins, and from my experience so will most pro players. I've written many times about how I hate playing a deck that will lose to anyone who leaves home deciding they are going to beat you - instead, I prefer a deck that will give me more game against my bad matches, even if that means I have to work harder for my wins. Dredge is a deck that wins - it is very hit or miss - whereas Faeries is a deck that doesn't lose, as in it has the ability to beat everything, but not as easily.
Of course, this is not to say there is no merit to either strategy - take Pro Tour San Juan, for example. The Mono Green deck from Zvi's team was an example of a deck that wins - it just smashed you before you even knew what was going on. If you wanted to, however, you could stack your deck against it, and then it was suddenly not as good - Martin and Lucas, for example, played maindeck Chain Reaction in their RUG decks. For that tournament, though, most people were not gunning for it, which meant that its weakness was in fact not much of a weakness, and it was probably the best deck for that event (even though, contrary to popular belief, our deck had the best record *cough* *cough*). The deck we played (RUG), however, was not a deck that wins - it was a deck that doesn't lose. The real strength of that deck was that it didn't have an actual terrible match, and, even if people wanted to beat you, there was absolutely nothing they could do with their decks that would swing the match around. For that reason, it was a better choice for after the Pro Tour.
Basically, everyone who thinks they are better than the competition is going to try to go for a deck that doesn't lose as opposed to a deck that wins, unless they find something that is really much better than everything else. By contrast, people who think they are worse than the competition will tend to go for a deck that wins. One thing I always hear from people that go to the Pro Tour for the first time is that they think they play worse than the Pro players and they are not going to win if they "play fair", so they try to play a deck that wins, like Mono Red, as opposed to a grinding deck. That is actually a valid thought, but you have to make sure you are not handicapping yourself too much - if the choice is borderline you can use that as a tiebreaker but do not give up on a deck that you think is much better to go for a deck that wins because you think you'll be outplayed.
Your next decision of winning/not losing comes when you get your opening hand. There are some decks that are fine with hands that do not lose, but sometimes you just need a hand that wins, and you have to know those times. A hand of four lands, three Lightning Bolts is unkeepable in your mono-red aggro deck against RUG, because you need a hand that wins, but the same hand is very keepable from the RUG point of view, because all you need is not to lose.
Sometimes, there are matches where you won't be very favorable - maybe you have a bad matchup, maybe your limited deck is really bad. As such, you have to make decisions that, should they work out, will give you the best chance to win the game. To put it on percentages:
Lets say that you have a 30% chance to win a certain match with an average 6 card hand. If you keep a gambling hand, there is a 45% chance you will "get there", and then if you do there is a 75% chance you'll win. If you don't get there, you auto-lose - this gives you around 33% chance to win, which is more than you had before. On the other hand, if the chance to win is 50% with a 6 card hand, and then 90% if you hit with your gambler hand, that gives you around 40%, which is not a good deal. Of course those percentages are completely made up and you can never figure them out during a game, but this is just to illustrate the point.
Lets say you are playing against RDW with your Soul Sisters deck from old t2, with maindeck Kor Firewalkers, and you see an opening hand like this:
This is the kind of hand that wins - if you draw a Plains, you have to think you are a big favorite. However, you have a really good matchup here - if everything goes according to the plan, you are going to win. Therefore, there is no need to keep a hand like that and put yourself in the position where you don't draw a land and just die, because in this match all you are hoping for is a hand that doesn't auto-lose, so that you can have a real game (and then win like you're supposed to), and the one-lander has the biggest potential to be a losing hand. I know it is very tempting, but I would throw this one back.
Another thing I see often is that people play against better players and then keep gambling hands because they think they need those to win, but most of the time that is just not true. Yes, I know playing against Brad Nelson is scary and he wins a lot more than he loses and you have to think he is a favorite to beat you, but part of the reason he is more likely to beat you is because you just kept that one-lander. By doing so, you are actually fueling his favoritism instead of bringing it down. Keeping this kind of hand also allows for an easy excuse (kept a one lander, didn't get there!), but that is for weak people who need an excuse, and we all know you are not this person (besides, "mulliganed and lost!" makes for just as good as an excuse).
Most of your choices on this matter, though, will be during the game. Again, this varies a lot from player to player - I tend to lean more for the playing not to lose side, but you should always aim to get a balance between the two. As a rule of thumb, the more likely you are to win a game, the more you should play not to lose, because you are betting a lot (a game that is almost won) for very little (if you win, you win a game that was almost won anyway). As such, even a very small chance that things are going to go wrong should be enough to dissuade you from the "winning" play, and you should probably take the "not losing" play.
Taking an example from LSV's last report, imagine you have Geth and an otherwise even board. Your opponent has no creatures that can block and kill Geth, but he has WWW up. The "winning" play here is clearly to attack - after all, you are going to deal five damage, and five damage goes a long way toward winning. However, in this situation, you don't need to win the game - you simply need not to lose, because if the game goes on, you are going to win with your Geth, and there is no need to risk it for an upside that you absolutely don't need. This situation is pretty obvious, but there are way more intricate situations in which you should play not to lose.
A very common occurrence of this is when you don't leave enough blockers up - sometimes you know you are going to win the game, but then they draw and use their three cards to kill your guy, play a hasted dude and a Giant Growth and all of a sudden you are dead and wishing you had left that other guy to block. It is also common for you to play a creature you don't need, and then get it Wrathed or stolen - I know I have won games where my only out was my opponent playing an extra, unnecessary guy, so that I could steal it and survive. Another card to keep in mind for this is Mark of Mutiny - sometimes I see people who have complete control of the game play a gigantic unnecessary dude and then they get it Marked and die when nothing else could have killed them.
I remember in Toronto when Ben and Gabe were talking about a match on Magic Online; Gabe was playing and Ben told him maybe not to attack with everything, because the only way they could possibly lose the game was to Sunblast Angel. Gabe just shrugged and sent everybody in. Then next turn their opponent played Sunblast Angel and they lost. They didn't know the opponent had Sunblast Angel in his deck, but it was still right to play around it and not attack with everything, because the only way they could lose was to Sunblast Angel! When there is only one thing in the universe that kills you, you do everything in your power to prevent dying to that thing, no matter how unlikely it is that they have it, because you are basically freerolling it - if they don't have it, you win anyway! A lot of the time, people are lazy and just want to end the game and save themselves the trouble to play around something, and then they get punished for it and it really is all their fault. Of course this was only a MTGO match and I believe both those players would have played around it in a serious tournament, but the example is still illustrative.
Another example of a player who should have played not to lose was Gabriel Nassif in his match against Mihara in the Worlds 2006 semi-finals. In that match, Nassif was a pretty big favorite, and at some point, after already having survived a Dragon (or a Dragonstorm, I don't recall), he tapped out to return a Martyr of Sands to play with Proclamation of Rebirth, when he already had a Martyr in play and a Remand in hand. Then on his turn Mihara topdecked the exact card he needed (jeez, I wonder where I've seen this before) which, together with the other three perfect cards in combination, enabled him to Dragonstorm for a bunch and kill Nassif. Was that unlikely to happen? Yes, it was - he needed to have those specific three cards in hand, and he needed to draw the last card. But it was, nonetheless, something that could have been avoided - Nassif did not need to return his Martyr that turn; all he needed to do was to sit on his Remand and on his Martyr that was already in play, and then make sure he was not going to die. As long as he did not die that turn or very soon, the game would eventually spiral out of Mihara's reach. What he did instead was try to put the nail in the coffin by bringing a second Martyr, but that was not really necessary at that point.
Playing not to lose
In my opinion, the best way to play not to lose is (like a lot else in Magic) to put yourself into your opponent's position and figure out what he can possibly draw to beat you. I don't care how lucky he would have to be to have all those cards, if you are going to win the game and there is a combination of cards that beats you, you should probably play around it. To do that, though, you must actually know how to devise a combination that cards that wins, which is what you have to do when you play to win.
When you are losing, the exact opposite scenario applies - you bet very little (a game that you were likely going to lose) and the upside is tremendous (you win a game you were not winning). Therefore, you don't need very good odds of actually succeeding to make the play profitable. The way you play to win is you basically take any small chance you have, and clutch to it as if your life depends on it (because, well, it does). You pretend like everything that has to happen for you to win is going to happen.
Imagine the scenario where your opponent has a 19/19 and you are at 20 and have a 1/1; They are at 10 life, and you have no removal in your deck, but you have an Armored Ascension with 9 Plains out. In this situation, you just have to take it when they attack - you are never going to draw something that deals with his guy, or a bigger guy to block it, so the situation is just going to repeat itself until they draw something or you blank, and then you're going to die. By not blocking, you have to get lucky and draw Armored Ascension as well as hoping they don't play a blocker or a removal, but that is the only chance you have! If the only chance is you drawing Armored Ascension, you play as if you were going to draw Armored Ascension - if you don't draw it, then you are going to lose anyway, so you don't lose anything. by playing like that.
Sometimes you not only have to hope you draw certain cards, but you also need your opponent to play in a certain way. In that case, act as if he was going to play like that! If you need him to miss a point with Cunning Sparkmage, then make sure you play in a way that, if he does miss the point, you capitalize on it. Basically, just figure out what needs to happen for you to win, and then make sure you do win if that happens. When you play not to lose, you do the same but for your opponent - figure what has to happen for them to win, and then try to play in a way that you do not lose even if that happens.
Though I generally like playing not to lose more, and I think it is more often the correct approach, I think I am better at playing to win - I can devise a lot of complicated things that have to happen for me to win a match. This often results in some awkward chump blocking, removing a card that seems worse than another card, attacking with my 2/2 only to be attacked by my opponent's 5/2 on the next turn - for an outside observer this might seem really random, but most of the times when I do that it is because I think I am not going to win the game, and I figured that my best chance requires me to make this sequence of plays this turn (or I am making a mistake, that is also an option).
A very good example of playing to win as opposed to not losing is Craig Jone's Lightning Helix play. For those who don't know, Jones was playing Olivier Ruel and Olivier attacked with a couple guys. Mike Flore's suggested play (he was doing coverage for some reason) was to burn the guy in the attack step, so that he could survive for an extra turn. Craig Jones didn't choose this line, though - he took the damage, going to a very low life, and instead burned Olivier. Then he untapped, drew Lightning Helix and used that to finish Olivier off (and in case you are wondering, this is where Randy Buehler's OH MY GOD IT'S LIGHTNING HELIX comes from).
Flores' play is a clear play not to lose - you are going to prolong the game for as much as you can. In this position, however, playing not to lose doesn't accomplish anything - you don't lose now, but you lose three turns later, and we don't get points for taking longer to lose (unless you are going to run out of time, in which case you actually do get points). Craig's play was a play to win - if he fails, he loses the game in that next draw step as opposed to in the next three draw steps, but he was going to lose the game anyway! He probably figured out that he had a much better chance to just topdeck a burn spell than to draw the three perfect cards in a row that he would need to win the game in the position he would have put himself by killing a creature, and then he got lucky and drew the card he needed, but if he hadn't played in exactly this way all his luck would have been for nothing.
Another example is my Top 8 match in San Juan against Josh Utter-Leyton. It was game five, and I attacked with a Nirkana Revenant along with some other guys. Josh blocked one of the small guys, and then I pumped the Shade and he took lethal damage. Did he miscount my mana to lose the quarterfinals of the Pro Tour? No, he didn't. Then why did he not chump block the Shade? Because he was playing to win. By blocking the small guy and eating it, he gives himself the scenario where I either forget to pump or fear some random card and don't pump, and then he can topdeck a bunch of guys in a row and I can topdeck a bunch of lands and he can actually win. Of course, the chance of this all happening at the same time is probably smaller than 1%. If he chumps the Shade, however, then he lives for a while but dies on the following turn, and there is no amount of things that can go his way that will make him win other than me having a heart attack, which is a much smaller percentage.
Yet another example comes again from Gabriel Nassif’s match in the Worlds 2006 Top 8, the same match against Mihara in the semis. In this scenario, Nassif had 6 lands in play, none in hand and had not played a land for the turn (he had in fact not played a land for many turns). Mihara had lethal damage in play, and Nassif had a full hand that included Compulsive Research and Wrath of God. Nassif felt that he needed to keep making land drops or he would not win the long game, and decided that he would rather play Compulsive Research there and try to find a seventh land to play Wrath of God that turn - if he misses, he dies. He hit, and then Wrathed Mihara's board (I wish I could say "and won the game" but I don't remember who actually won this game). The point here is that he felt, from the way the game was going, that if he played not to lose and just Wrathed he would actually lose in the long run - he figured he would have better odds to win the game by hitting a land there, and the fear of an immediate loss didn't stop him from going for the "to win" play.
One thing you should worry about, though, is not overdoing this so that, by trying to play around all the bad scenarios, you cause a bad scenario to happen - sometimes you think you are so favored that you can't lose unless they have Sunblast Angel, but by not attacking you now put yourself into a position to lose to Shatter instead. Whenever you play not to lose, make sure you are not giving up too much of your advantage to the point where it's no longer worth it.
One example of this comes from our last team draft in GP Toronto; Luis was on my team and playing against Gabe, and Luis attacked with a bunch of guys, including Golem Artisan, and enough mana to pump for lethal damage. Gabe had five lands and two Myrs, and UR open, and didn't block anything. We knew from the draft that Gabe had Platinum Emperion (since it was teams and the person feeding him saw it, but not the person next to him), so Gabe's play made perfect sense - he was so behind on the board that he could not possibly win if he blocked, so he had to hope that Luis would not pump for fear of something, and then he would be able to untap and play the Platinum Emperion and try to win the game with that. If the Myr blocks, he is back to 6 mana and dies next turn anyway, so he made the "play to win" that he had to.
Then, Luis didn't pump for lethal damage - he said he was trying to play around Disperse or Shatter, and he was so far ahead that he didn't need to push it. Then Gabe (who didn't have an instant) untapped and didn't draw a land to play Emperion, and died anyway (justice). Later on, Luis said he should not have tried to play around anything - because he was too far ahead, he could afford to get blown out by Shatter/Disperse, and by playing not to lose instead of going for the throat he actually gave Gabe an out for the game.
I know this seems kinda obvious and me saying "only play around things when you can afford to play around things" does not help much in this matter, but there really is no definite way to know when you should or shouldn't do it, and all I can do is give you some general scenarios to try to help you think for yourself when the situation comes up - basically, if playing not to lose makes you actually lose to more (or more likely) things, don't do it.
So, to sum it up, whenever you think you are in a very bad position, assume that everything that has to go your way does go your way, and then play to win in that scenario - no use prolonging your loss. If you are in a very good position, try to do the same for your opponent - assume that everything that has to go his way goes his way, and then try to prevent that. If you do both those things, you will certainly reduce those "Jesus my opponent drew the three cards he needed to win the game", and at the same time you will increase those stories for everyone else.
I hope you've enjoyed this, and see you next week!