A realistic evaluation of your "edge" using the AVM methodology (which has a probability of success of 1/3 based on chance) will probably require several hundred or more Trials. The reason for this is quantified by the binomial distribution.
The binomial distribution describes the probability of getting "S" successess when there are "T" Trials. When "T" gets large, a better measure is the Cumulative Probability of getting "S" or Less successess - this is your Percentile. The following examples for "T" = 10, 100, and 1000 Trials will clarify the situation.
If you were to successfully predict 4 out of 10 trials, your percentile would be about 79 (see the first graph below). This percentile means that if a very large number of people were to make 10 predictions with a chance probability of 1/3, then 79% of them would correctly predict 4 or less, or equivalently, 21% would predict more than 4 correctly.
If we are looking for evidence of a 7% "edge" over chance, then a 40% success rate with only 10 trials is clearly not sufficient. An edge is defined as a consistent increase over chance expectations of 33%.
However, a key characteristic of pure chance predictions is that as the number of trials increases, the likelihood of a 7% edge decreases dramatically. For example, if you were to complete 100 trials and be successful 40 times, then your percentile increases to 93.4. Thus, by chance only 6.6% of a very large sample would do better than you. See the following graph.
If you have 400 successes with 1,000 trials, then your percentile is 99.999557. Only 0.000443% would do better by chance. Thus, it is quite credible to believe that anyone who achieves 400 successes with 1,000 trials does, in fact, have a 7% edge. See the following graph.