Hi Glen This is an issue that catches virtually every newbie. The cause is that, unlike most other moving averages, an EMA uses ALL historical data to generate a result. Do some basic research on EMAs and you'll see why that is. In practical terms what it means is that you should either use the same amount of data in a system test or exploration as you normally use for an open chart. As a rule of thumb I suggest that five times as much data as the longest EMA Periods parameter be available for an exploration. Where you have one EMA nested inside another then "times five" should be applied to the SUM of both Periods parameters. With a systems test this amount of data should be available prior to the starting point of the test. One way to load more data without including that extra data iin the test is to add a "date filter" to your entry signals. The down side of adding more data is that execution speed slows down, but the upside is that the results are much more accurate. Roy