GMAT Strategy

Ideally, you would not need to guess at all- you would pace yourself so that you had time to finish the test, giving each question sufficient consideration. If you haven't read other posts here about pacing, you should be able to find some helpful advice.

That said, if you did know in advance of your test that you absolutely needed to guess at a few questions to conserve time, your strategy for the math section is likely a good one. You'd want to avoid guessing at a string of consecutive questions if at all possible. I'd also recommend guessing at DS questions rather than PS questions; on DS questions, if you take 15 seconds to read through the problem, you can often at least eliminate two or three answer choices, which makes guessing less damaging to your overall score.

I'm less convinced by your verbal guessing strategy. It's certainly true that you'll save a good chunk of time by skipping an RC passage, but you're risking a damaging string of consecutive wrong answers. And it isn't true that every RC question about a given passage is of the same difficulty; while some passages and question sets are more difficult overall than others, each individual question will have its own difficulty level, and it isn't relevant that the sets are not adaptive.

I'm not sure where you might have read that there is an algorithm for guessing correctly on the test, but I can assure you that's a complete myth.

I received a PM asking me to comment on this thread.

I agree with everything Ian said. A couple of other thoughts are below.

1) Go back over some of your recent practice tests and see what types of problems tend to "suck you in" (ie, spend too much time). Just being aware that you tend to get sucked into 3-D geometry problems and PS word problems (or SCs with the whole sentence underlined, or whatever) will help. If you're aware of your tendencies, you're much more likely to stop yourself from getting so sucked in that you cost yourself points at the end of the test.

2) If you have to guess randomly, don't choose the question number ahead of time. Actually look at every problem. If, after reading it, you are having the, "uhhh.... what are they talking about?" feeling, then guess randomly and move on. And don't start thinking about this only at the 30+ mark; if at any point on the test, you are more than a few minutes behind on your timing benchmarks and you hit one of these "Huh?" questions, just pick something and keep going.

3) Really, really make an effort to hold yourself to the per-question timing standards. If you go back and look through the data from recent practice tests, I'm positive that you'll discover multiple problems on which you spent way too much time and which you got wrong anyway. Going way over on the time is primarily an indication that you don't really know how to do that problem. The solution? Get it wrong faster. At least that way it won't harm your chances on a future problem as well.

4) You don't want multiple problems wrong in a row. So if you do have to guess to get back on track with respect to your timing, spread those guesses out. Ideally, have timing benchmarks for every 10 problems. If you find yourself more than 3-4 minutes behind, make a random guess on the next "this looks really hard" problem that you see. If you're still more than a few minutes behind, wait a few problems and then make another guess on a really hard one.

Re: your overall guessing strategy, you have two primary goals: (1) choosing to make guesses on hard problems, not just on the next problem, whatever it is, because you happen to be behind, and (2) spreading those guesses out to minimize the chances of having a string of wrong answers in a row.

You're not the only one who struggles with the scoring of this test! Retrain your brain to ignore the # or % right. That's not how it's scored. You WILL get lots of questions wrong but as long as you answer correctly the questions that are at or below your desired scoring level, you'll be fine!

Instead of thinking of it as % correct (which applies to static paper tests), think of it this way: I'm sitting across a table from you and giving you questions one at a time. I decide what questions to give you based on the performance I see as you answer the previous question. I start you out with something in the 500s and you get it right, so I think - okay, let's try something a bit harder. And I give you some more questions that you get right, so I try some harder stuff. Now, you start getting things wrong, and I think - oh, okay, now I'm finding the limit. Let's test that a bit, so I start giving you things above and below to see what you get right and wrong, and that allows me to narrow the range and figure out what your "true" score is. (I put "true" in quotes, because who knows how valid these kinds of tests really are... ) Anyway, the point is, part of what I use to get your score is the ones that you get wrong, not just the ones that you get right. I'm always going to give you stuff that you'll get wrong because that's how I've found your limit. If I gave you a static test (that is, all questions determined ahead of time), instead, on which you got everything right, it might be that I just didn't give you hard enough questions!

So, please do risk getting the difficult questions wrong! It's much worse on your score to get the lower level questions wrong. Let's just define some terms here, because "lower level" and "difficult" mean different things to different people.

Basically, you're going to see a range of question difficulties that cluster loosely around your current scoring level. We'll split these into "low," "medium," and "high" - which correspond to different difficulty levels for different people. Your goal is to get all of the "low" ones and at least half of the "medium" ones right. But wait - you want to improve your score, so you have to get harder ones right! Yes - but then you also have to nail the "lower half" from your current range just so that you're offered harder ones in the first place. Then you start to work on consistently answering all of those "medium" ones right, instead of just half. And you work on picking up some speed (say 15-20 seconds) on the ones you can already get right, because the questions are just going to get harder, and the harder ones take more time unless you've studied how to save time on the ones you can already do and then applied those time-saving lessons to harder problems.

If your current scoring level is around 650 and you get a 680 question wrong, well, that's not going to prevent you from getting a 650. But if you get a 620 question wrong, now you've got a bit of a problem. (And one wrong answer won't actually prevent you from getting that 650, but if you start to get multiple questions wrong in that low-600s range, then you do have a problem.)