The average from the Values tab is the accurate one. It is computed from the raw values. You can compute it again by exporting the raw data of your transaction from the Values tab to a CSV file.

Regarding the average of the graph section it's the average from the values in that graph. Since on the graph you do not have the raw values but data aggregated by time interval the average on the graph is different due to rounded values.

A 250 VU license allows you to run 250 virtual users concurrently and not more. You need a 1000VU license to run 1000 concurrent users.

You should be able to manage that with regular expression even if the boundaries are not the same all the time. But without seeing the response and the values you would like to extract it's difficult to provide you more details.

In any case the recommanded way is to always use framework parameter when possible. First you won't have to do such work all the time and the framework will automatically replace the occurrences for you. In the framework definition you can specify where exactly the values should be replaced. For that you can use regexp or XPath and in our next version coming this week JSON path.

With a variable extractor configured to extract all those EntityId values and if you moved it as framework parameter NeoLoad will automatically replaced all the occurrences in your requests in the same order.

That means that it will do exactly what you do but automatically.