I can't comment on any particular company's emulation of the GMAT algorithm, but do be aware that:
-no company test implements the real GMAT algorithm;
-there are many reasons why you can see questions on your test that are not at your estimated ability level. The experimental/diagnostic questions certainly can be at any difficulty level, of course, but you can also see questions which count which are above or below your level. Your response pattern to past questions is not the only factor that determines what question you see next. The test also needs to ensure a certain content balance, and since there is only a limited number of questions on each topic, there may not be a question that is both at your precise level and in the right subject area. Further, there is an exposure criterion that contributes to question selection - as a security measure, they want to ensure that no one question is overused (so that there isn't a great likelihood that you'll see a question that other people have seen earlier in the month).
The description in test prep books ('if you get a question right, the next question is harder, and if you get a question wrong, the next question is easier') is a significant simplification of how things really work. I don't have a problem with that, because it captures the details that test takers really need to know, but the test implementation is a little more complicated than it might sound from that description.
For online GMAT math tutoring, or to buy my higher-level Quant books and problem sets, contact me at ianstewartgmat at gmail.com
ianstewartgmat.com
