Speed - 2x automations on one instance or two?

[Fugazi]

So, when I add too many rows to my automation it makes it less profitable as speed is important.

Should I

A) Have two automations on one instance, Automation 1 back rows 1-6, Automation 2 back rows 7-12

OR

B) Have the 2 automations on 2 separate instances?

[jamesedwards]

So, when I add too many rows to my automation it makes it less profitable as speed is important.

Should I

A) Have two automations on one instance, Automation 1 back rows 1-6, Automation 2 back rows 7-12

OR

B) Have the 2 automations on 2 separate instances?

Option C strip down your automation. Reduce the number of rules. Reduce the number of rules triggered by time. Reduce the amount of log data being created.
Option D invest in a faster PC.
Option E run automations on more than one PC.

[Fugazi]

So, when I add too many rows to my automation it makes it less profitable as speed is important.

Should I

A) Have two automations on one instance, Automation 1 back rows 1-6, Automation 2 back rows 7-12

OR

B) Have the 2 automations on 2 separate instances?

Option C strip down your automation. Reduce the number of rules. Reduce the number of rules triggered by time. Reduce the amount of log data being created.
Option D invest in a faster PC.
Option E run automations on more than one PC.

Honestly, needs all the components. I'm using BA's best VPN (I know cheaper ones are out there but its been 100% uptime)

[Michael5482]

So, when I add too many rows to my automation it makes it less profitable as speed is important.

Should I

A) Have two automations on one instance, Automation 1 back rows 1-6, Automation 2 back rows 7-12

OR

B) Have the 2 automations on 2 separate instances?

Possibly better testing both by setting a signal up to log then check the timings v each other instead of someone having a guess 1 second is an absolute lifetime.

[Anbell]

So, when I add too many rows to my automation it makes it less profitable as speed is important.

Should I

A) Have two automations on one instance, Automation 1 back rows 1-6, Automation 2 back rows 7-12

OR

B) Have the 2 automations on 2 separate instances?

Option C strip down your automation. Reduce the number of rules. Reduce the number of rules triggered by time. Reduce the amount of log data being created.
Option D invest in a faster PC.
Option E run automations on more than one PC.

Option F: Optimize Restrict Refresh

[ShaunWhite]

What do you get if you psping from your vps to api.betfair.com ?

Another possible is to run exactly the same automation on several instances, the idea being that they run out of sync and one of them is always acts before the others. Picture a three-phase power supply vs single-phase. I managed to find another 10-15% that way. But it'll depend on how Guardian has been written wrt local caching of orders in transit.

[Fugazi]

What do you get if you psping from your vps to api.betfair.com ?

Another possible is to run exactly the same automation on several instances, the idea being that they run out of sync and one of them is always acts before the others. Picture a three-phase power supply vs single-phase. I managed to find another 10-15% that way. But it'll depend on how Guardian has been written wrt local caching of orders in transit.

Could cause carnage between my bots with transaction count as they'll annoy the shit out of each other BUT it may actually mean nobody else can get in on the action. This has definitely got me thinking going to give it a small test run later

[ShaunWhite]

Could cause carnage between my bots with transaction count as they'll annoy the shit out of each other

As far as I know Guardian can process a rule set every 20ms whilst transactions take over 100ms to be acknowledged. If it doesn't double up now then it shouldn't if a version of itself ran less than 20ms ago. I'd imagine a ruleset is executed in well under a millisecond. BA is compiled code as is mine and I process a market in about 30µs so two instances processing at exactly the same time should be highly unlikely.

Even if the code doesn't normally facilitate it via locking or similar then it might work? The odds are probably iro 20ms/50µs (400/1) you'll start them both absolutely in sync?

[Fugazi]

Could cause carnage between my bots with transaction count as they'll annoy the shit out of each other

As far as I know Guardian can process a rule set every 20ms whilst transactions take over 100ms to be acknowledged. If it doesn't double up now then it shouldn't if a version of itself ran less than 20ms ago. I'd imagine a ruleset is executed in well under a millisecond. BA is compiled code as is mine and I process a market in about 30µs so two instances processing at exactly the same time should be highly unlikely.

Even if the code doesn't normally facilitate it via locking or similar then it might work? The odds are probably iro 20ms/50µs (400/1) you'll start them both absolutely in sync?

Where it collides is when the ladder moves x ticks in x seconds (its a bit more complex than this as obviously you just change the tick movement +1). So an extra bet messes with ladder movement. But so far tested on a few races and massively increased roi (tiny sample ofc)

However Ive now quadrupled my transaction count because of conflicts. But thats the least of my worries and I'm sure I can deconflict that at a later point)

I contacted Betfair to see if any wriggle room on my transaction limits .. a firm no

[ShaunWhite]

I contacted Betfair to see if any wriggle room on my transaction limits .. a firm no

They clamped down years ago. Even those who where good enough customers to be on unlimited trn suddenly lost that. Going over doesn't necessarily mean you pay, commission paid offsets it.

[Fugazi]

I contacted Betfair to see if any wriggle room on my transaction limits .. a firm no

They clamped down years ago. Even those who where good enough customers to be on unlimited trn suddenly lost that. Going over doesn't necessarily mean you pay, commission paid offsets it.

Maybe I try run some on Betdaq too....

[ShaunWhite]

. But so far tested on a few races and massively increased roi (tiny sample ofc)

I doubt it will last, the most you'll gain on avg is 10ms. Maybe 1% max.
The 10-15% I referred to was doing that with all my strategies, and they're very latency sensitive but not all are so much. The limiting issue becomes the number of connections you can have.

[Fugazi]

. But so far tested on a few races and massively increased roi (tiny sample ofc)

I doubt it will last, the most you'll gain on avg is 10ms. Maybe 1% max.
The 10-15% I referred to was doing that with all my strategies, and they're very latency sensitive but not all are so much. The limiting issue becomes the number of connections you can have.

As I'm only on £1 stake, seems ive managed to scale to £2 this way without the usual reduction of profit