DislikedHi all, Has anyone else noticed that different ADR values get calculated depending on your broker's server time? It might not be that common. Out of 8 brokers that I checked only one was different. So for the 7 that had the same server time the ADR varied from 0 to 1 but the one with a 3 hour server offset from the rest caused the variation to be 4 to 10 pips from this group of 7. Attached image shows a side by side comparison. I've added a 2nd chart of AJ on the right for each borker to add 2 extra ADR indicators. The one in the yellow box is R4M-ADRIndi...Ignored
Let me try to make things more confusing...
At the top of your Market Watch window there is a time stamp that keeps changing -- that is your broker's server time. That server time is unrelated to your broker's time zone. The server-time is the time zone of the data feed. My broker my be located on some little island in the Pacific Ocean but receives its data feed from a bank in London -- so the server time is that of the London bank and not the location of my broker. In fact, we could be using two different brokers located in the same city -- and our accounts might have different server times.
ADR -- Average Daily Range -- the difference between the highest and lowest price during the day. Someone -- a curious person -- might ask "Well, when does a day start?", or "Does it make a difference if the day starts at different times for different brokers?"
When you are speaking solely of the ADR it probably does not make much difference at what time you define the start of a day. Since we are averaging -- typically over 20 days -- then it all comes out in the wash.
But what if you are monitoring the percent ADR reached so far today -- does the start time make a difference? Yes -- of course. A forex market session begins with the Asian Open, followed by Frankfurt, London, NY and Sydney. Things don't start popping until around the middle of the London session. So, if my start time were the middle of the London session then my %ADR might reach 100% within 4 or 5 hours. But if my start time were the Asian Open, then it would take many more hours to reach 100% ADR.
Ideally the ADR should be calculated beginning with the Asian Open -- but none of the ADR indicators I've seen do that. They all start at your server's time 00:00 (i.e. midnight). I have several different accounts -- one at GMT0 and the others at GMT2. The GMT2 accounts show similar %ADRs at the same time of day -- but those are slightly different than the GMT0 server.
A while back I thought I would modify the ADR computation to take into account that everyone's server might live in a different GMT time zone and in that way our stats would be more comparable. And it seemed to do the right thing. But when I began testing I started getting some unexpected results and, in fact, overall statistics declined. I'm not sure why. There's a lot of code and ADR is used in many different places -- not only my code but also in the code that I've borrowed from other sources. So, at that point I decided it was not worth it to force everyone's ADR to align if it impacted results in a negative way. After all, it's not like I routinely compare charts from different brokers.
So, in the end, I think this issue doesn't make that much difference. What is most important is that the ADR computations are internally consistent -- that is, within a particular account. It is not necessary to compare the ADR across different brokers. Furthermore, ADR is just an approximation -- there is variation each day regardless of when you define the start time. So, even if ADRs were to match across different servers there wouldn't necessarily be any improvement in performance and, in fact, actually introduces some risks because of potential side effects. And since this issue is not related to a performance bug, then I've decided to take a pass and no longer worry about it.
@rbs270 -- if you were confused before then I imagine I made things worse. But at least you should know you are not alone...
There is no effort without error or shortcomings (T.R.)