MarketFeeder Pro 8.0 - BETA testing

So something's not right with the maximum (of all) condition in that it wouldn't appear to accept there might be more than one selection, which the V7.2 version does.

Wez, the trigger must have acted according to your logic, I don't see any contradiction:

Selection's Trigger Expression matching_number is equal or less than 3

That means 1 is OK.

and Selection's Back Price is between 2 and 66

So even if there were 2 in the previous condition, one of them could have been filtered out because its price was not right (i.e. less than 2 or greater than 66).

Although V7 placed two bets, it does not necessarily present a fault in the logic of V8, as prices can change every millisec, and the two copies of the program do not refresh markets synchronously. You need to check the logs of that trigger.

2. The exact same trigger and settings on V7 and V8 produce files on V8 that are some 30% larger than V7.

I'm not sure we can draw any significant conclusions from comparing log files generated by different triggers, LarpHager.

Firstly, the logic of these settings ("once per market" and "once per selection") is different and they will produce different logs in different volumes.

Secondly, one log file was being recorded for two and a half minutes, while the other one - for less than a minute. In one log the conditions of the trigger were met, but it had already been executed on all qualifying selections (and it said so in the log), while in the second log the conditions were false.

So I'm really not sure this matters.

P.S.
I've taken on board all the reports about possible memory leaks and obvious crashes, so I'm not commenting on them because I'm trying to find and eradicate them.

hi oxa thanks for replying about the thread issue, I have tested 8 again today with 4 markets refreshing, I started at 7.30 and not one bet was missed up until 12 av refresh rate was .22 and current bet refresh was .5. After 12 oclock though every bet is being left unmatched and current bet refresh average had gone to 2.16 with longest bet refresh a massive 15 seconds.This has happened to me the last 3 days ,everything fine till about 12 then lots of unmatched bets afterwards.I have noticed that once the average percent of cpu consumtion on the task manager goes above 4 percent in every 60 seconds the program becomes non responsive.surely something must be up as I am only monitoring 4 markets witn 4 second refresh rate, I have version 6 running next to it with 16 markets at 1 second refresh at it is still working fine, I have had to turn 8 off.

Been testing it all week to see how many markets I feel comfortable with it monitoring at a time, I have had to do this in real time mode as everything gets matched in test mode, my findings: I can only monitor 3 with refresh rate of 3 seconds, with a simple dutch trigger, anymore and bets are unmatched, with version 7 I could monitor 5 with refresh rate 1 and version 6, 20 markets with refresh rate 1.It is definetly to do with the size of the file, on the cpu version 8 runs at 48000 twice as big as 7 and 4 times as big as 6, is there so much extra in 8 that makes it so big, I will have to open 6 profiles every day to monitor the markets I need. I also have to turn it off at 2 as after 2 it becomes to unreliable

Interesting. Thanks for sharing, I wonder what other users' experience will be, as mine is different.
Let's also not forget that the API itself is different, so its performance and responsiveness (and thus the strike rate for matched bets) is different as compared to the previous API, and this is something we can't influence.

Coling, perhaps you could send your trigger file to the support for me to run it and see if it will increase the memory usage (it's not CPU, it's the RAM volume - just a minor correction so that we are clear as to where the problem might be).

hi oxa
any reply on the live bets showing when the uk and ausy bet list and account statement is unticked.

mcbee

hi oxa
any reply on the live bets showing when the uk and ausy bet list and account statement is unticked.

Hi mcbee, I've noted your message down as soon as you posted, already fixed that, the fix will be in the next update on which I'll post in this thread.

thanks oxa have sent the file to support. Coling

I had several bets so far today that were matched and won but did not show a profit on the market screen under Settled P/L, but did appear on the statement as a profitable selection.  Example attached - Thunder and Rain was placed.  This has screwed my staking system today - my first live test of Version 8.

Hi Oxa,

I'm not sure we can draw any significant conclusions from comparing log files generated by different triggers, LarpHager.

The triggers are the exact same Oxa, differing only in name  and the Market Execution instruction.

Firstly, the logic of these settings ("once per market" and "once per selection") is different and they will produce different logs in different volumes.

That is the point I am making. I presume "Once Per Selection" keeps looking for other selections even when there are none, while "Once Per Market" stops when it fires the once.

Secondly, one log file was being recorded for two and a half minutes, while the other one - for less than a minute. In one log the conditions of the trigger were met, but it had already been executed on all qualifying selections (and it said so in the log), while in the second log the conditions were false.

Again, the "Once Per Market" fired and the logs stopped after less than a minute. The 2 1/2 minute "Once Per Selection" fired early in the log, but ran up to the off time searching for new selections that were not there.

So I'm really not sure this matters.

Well, I think it does matter and I am sure there are many others who run triggers with "Once Per Selection" without realising the difference. In fact, the total trigger log file for the trigger on one dog race running a 2 second refresh for only 2minutes to the off totals to 3.95mb, too large to attach. This was reduced to 532kb, 0.532mb, by switching to "Once Per Market".So I have reduced the workload by a whopping 85% or so by changing the Market Execution Instruction. I don't understand how that is not important, not just the file size but the workload creating it?

I am attempting to record what happens if I crash again. I have created a dump file, but I don't really know if that will be any help. Is there anything I can set up to help locate any problems?
Thanks again
Larp

I have an annual subscription to MFP, cannot logon to Beta v8 app, it asks for license key, I enter the license key press ENTER and nothing happens.. please help.

Please read here how to install the beta-version.

click on the link in post one which takes you to a post oxa made about getting a new key sent via email

I had several bets so far today that were matched and won but did not show a profit on the market screen under Settled P/L, but did appear on the statement as a profitable selection.  Example attached - Thunder and Rain was placed.  This has screwed my staking system today - my first live test of Version 8.

Hi rubold, yes, I've just checked this and can confirm the issue. When MF Pro 8.0 was being prepared for the test, the real bets were being settled as soon as the market was finished. Now they are again settled with a delay, not sure if there have been any changes at Betfair. Very frustrating as it reduces to zero the advantage of knowing the winners/losers immediately after the market is closed.

I will have to think of some workaround.

Oxa, will the issue you describe have affected the variable market_settled_pl ?

rubold, yes.