Monte Carlo Results table Cannot compute - error

N

#1

nathan

07.11.2018 13:59

Task created

Description changed:

Retester, crosscheck trades manipulation, method exact, simulations 100,000, startegies in databank 20.

I know, 100,000 simulations is pushing it, but wanted to see if it could do it (100,000 is industry standard for banks apparently).

It managed to test them all inside a couple of minutes which was amazing - just had trouble displaying the results tables afterward, see screen shot - entire program then locks up. Nothing can be clicked. Not Responding. SQX needs to be hard crashed inn task manager.

After opening SQX again, only 2/2500 startegies are in the builder databank which is stuck on 75% loading and only 2/20 strategeis are in the databank of the retester which is similarly stuck on 75% opening.

Retester gives an error

13:09:59 Loading strategies to databank 'Results' failed - Cannot load results from file '2018.11.5_LASTStrategy 5.19.125.sqx'. zip END header not found

Startegies are phsycialy stillt there in their respective results folders on the hard drive, SQX cant load them to databank though.

Second time this week all stratgeies have been lost - although on Monday 6000 were also physically wiped from the drive/results folder.

MF

#2

Mark Fric

07.11.2018 14:20

Status changed from New to Refused

SQ is not made to work with 100,000 MC simulations and we have no intention to support it. Such huge number of data would require a special approach, and SQ is not this kind of tool.

Besides, I don't see any reason for that, when just 100 or 1000 simulations will give you the same amount of information. I don't see how that many simulations would give you any edge.

N

#3

nathan

07.11.2018 14:38

Fair comment.

What I wanted to gauge is at what point the particular strategies in the data bank that passed or failed became consistent - at 10 simulations its almost random in the data bank, 100 better but not consistent. 1000 around every 2-3 runs is different. And 10,000 simulations are pretty much stable. 100,000 was pushing it, I know :)

I am not knocking the software, its incredible. Just developing a way of working and testing things out at the moment - if I see a bug or a crash I'll post it as it could help you guys out - I really want SQX to succeed and you guys to sell it by the truck full - my main worry is that SQX is the only software out there that can do what it does and you are perhaps only a small team dependent on a solid revenue stream.

Monte Carlo Results table Cannot compute - error

History

nathan

Mark Fric

nathan