A 99% accurate polygraph has a 99% chance of getting the right answer. That means that for every 100 times you use it it will get the right answer 99 times and the wrong answer once.
A 80% accurate polygraph has an 80% chance of getting the right answer. That means that for every 100 times you use it it will get the right answer 80 times and the wrong answer 20 times.
But how accurate the polygraph is or is not makes no difference to the success criteria of your cat/ship test, which is the number of hits you need to get in order to prove that you have a paranormal ability and are not just making lucky guesses. That success criteria depends entirely on the type of test you are doing. For your cat/ship test, where there is a 50% chance of guessing correctly each run, you need to get the right answer at least 19 times out of 20 (or 26 times out of 30) for there to be a less than 1 in 10000 chance that you just guessed lucky. That's your target. Period. The only way to change it would be to change the type of test - for example have three words, say 'cat', 'ship' and 'book', which would mean you only have a 33% chance of guessing correctly each run. That would make your target 16 out of 20 runs, or 21 out of 30 runs.
The method with which you make your predictions, the tools that method requires, the accuracy of those tools - none of these things make a ha'porth of difference to the number of hits you need to get in order to prove beyond any reasonable doubt that you really do have a paranormal ability, and are not just making lucky guesses. They only affect how likely you are to meet that target. You cannot use the tables/your calculations to see what difference the accuracy of the polygraph makes to the hit rate required, which is what I think you're trying to do, because it makes no difference.
Do you understand this, golfy? Because until you do you cannot agree a test protocol with JREF. The success criteria is a vital part of that protocol, and JREF know how to calculate it.