We took the liberty of performing some scenarios to test the GMAT testing algorithm on Edvento vs GMATPrep. There were
some more and awesome tests done by GMATClub, which prompted us to use the data to show validation of our platform.
Scenario 1:
Quant: First 10 questions incorrect; next 27 correct
Verbal: First 20 questions correct; next 11 incorrect
The result:

Results from Edvento:
Analysis: As seen by the above result, the first questions turned out to be vital. Quant 38 is actually 48th percentile, which is not so good. Once you are in the high penalty zone, it is quite difficult to recover till the end. We did the same number of questions incorrect in Verbal and got 37, which is 81st percentile, an awesome score. So in Verbal, we had reached high reward zone and we didn't go as down as we did in Quant with the same number of questions incorrect.
Scenario 2:
Quant: First 27 questions correct; next 10 incorrect
Verbal: First 11 questions incorrect; next 30 correct
The result:

Result from Edvento:

Analysis: This time we reversed the order. By marking last 10 questions incorrect instead of the first 10, the quant score jumped from 38 to 50 and percentile jumped from 48th to 92nd percentile. A significant improvement. Similarly for Verbal, we marked first 11 incorrect instead of last 11 and the verbal score dropped from 37 to 22. Both in GMATPrep and Edvento, it's the same. The percentile significantly dropped from 81st to 27th percentile.
From the above two tests, it is clear that:
1) The first few questions are vitally important for your score
2) Edvento tests' adaptive logic closely mimcs that of the official GMATPrep.
We also did some miscellaneous scenarios to see how those affect your scores in the GMAT and Edvento Tests.
Scenario 3:
Quant: First 14 questions correct; next 10 incorrect; last 13 correct
Verbal: First 35 questions correct; last 6 incorrect
The result:

Result from Edvento:

Analysis: In quant, we did 10 out of 37 (1/3rd) incorrect in Quant. But we still reached a quite high score. That implies that though we were going down in ability in between, but at the stage (till 15th question) the damage is not that much that we could not recover from it by the 37th question. In Verbal, we tried something new, instead of 11 incorrect at the end, we did last 6 incorrect. That added a verbal score improvement of 6 points from 37 to 43 (95th percentile). That makes Verbal quite difficult to improve if even at this stage there is significant improvement.
Scenario 4:
Quant: Every 3rd question incorrect
Verbal: First 6 questions incorrect; last 35 incorrect
The result:

Result from Edvento:

Analysis: In quant, this time, we marked every 3rd question incorrect, as you can see from the plot. This did not seem to affect our high score from scenario 3. So 12 questions incorrect at equal intervals or 11 questions incorrect exactly in the middle, are essentially the same. So mixed incorrect in first few won;t affect the scores much. In verbal, we reversed scenario 3 and did first 6 incorrect. We scored 36 (79th percentile), not bad. But instead if we had done 11 incorrect (scenario 2), we would have scored just 22 (27th percentile).
Final verdict: First few questions are extremely important to mark correct consecutively. But if it is taking too much time (>2 mins per question), just guess and move on. A few questions incorrect won't matter much.