Trading for Influence II (FC Update)

[Jane Turner]

any exotic states or conflicts will also mess up the predictions, and under 200K will be iffy too - since its a rare even we haven't tried to model FDs fix for the log <1 being negative issue.

 

[goemon]

any exotic states or conflicts will also mess up the predictions, and under 200K will be iffy too - since its a rare even we haven't tried to model FDs fix for the log <1 being negative issue.

we obviously ran our tests in no state systems. no traffic, no states of any factions.

i'll check out some test systems later for their current states, and pick one for a test designed for maximum difference between our tested mechanics as above and the predictions your data driven model provides (thinking along the lines of single trade run, one source, different commodities, maximum tonnage, no demand, low total profit.)

that's gonna be interesting!

 

[goemon]

deal

System Population: 1,2 Mio
Market Controlling faction: 45,9 %
State: None. All factions states: None.

Trade: 4 different "zero demand"-commodities of 1 source.
Demand: 1 (the station produces the commodities itself).
single t9 run=720t

1 Commodity A: 180t with 100 cr/t profit = 18 000 cr raw profit
1 Commodity B: 180t with 250 cr/t profit= 45 000 cr raw profit
1 Comoodity C: 180t with 500 cr/t profit = 90 000 cr raw profit
1 Commodity D: 180t with 1000 cr/t profit = 180 000 cr raw profit

Total profit: 333 000 cr

___

What does your model predict? If i read it correctly, it should create a very minor influence gain following your databased model, and a significant gain according to our tests.

 

[Jane Turner]

We would expect the new inf to be 46.6 and the max in the system to be around 53.3.

 

[goemon]

We would expect the new inf to be 46.6 and the max in the system to be around 53.3.

we would expect >47,2.
max influence change draws from your excellent research, so same number.

 

[Chaz C]

Thanks for the work on this. Ever since the FC update I was beginning to wonder if there’d been a change as my usual 4x60t/750cr+ profit trades had hardly been making a dent in INF. This explains why - excellent work, cheers!

 

[Jmanis]

Ugh.... google-fu failing me.

Trading drugs on the normal market (whether beer, tobacco, wine, narcotics)... what's the Econ/Sec effects again?

For some reason I recall trading drugs has -ve economy effects (but +ve influence effects), someone able to confirm? If it's not that, it's -ve security.

 

[goemon]

Ugh.... google-fu failing me.

Trading drugs on the normal market (whether beer, tobacco, wine, narcotics)... what's the Econ/Sec effects again?

For some reason I recall trading drugs has -ve economy effects (but +ve influence effects), someone able to confirm? If it's not that, it's -ve security.

i know about secondary effects of food, medicines and weapons - but none of legal drugs.

edit: in context of elite dangerous, ofc ;-)

 

[Jmanis]

i know about secondary effects of food, medicines and weapons - but none of legal drugs.

edit: in context of elite dangerous, ofc ;-)

It's definitely a thing coz I remember being surprised by this when it first got mentioned and I tested it... because narcotics are one of my main trade runs to allied systems for sale on the legal market, and it explained why I'd been observing -ve something; it was myself doing that. But yeah, just can't remember if it was -ve econ or -ve sec.

 

[Riverside]

i know about secondary effects of food, medicines and weapons - but none of legal drugs.

edit: in context of elite dangerous, ofc ;-)

An example from this morning; I just delivered a cargo mission from one Pirate faction to another in a different system, the cargo was Liquor (legal, not stolen or illicit at either end).

The effect for the mission giving faction was positive of course, the effect for the receiving faction was negative economy, just as it would (more intuitively) be for narcotics say.

 

[goemon]

An example from this morning; I just delivered a cargo mission from one Pirate faction to another in a different system, the cargo was Liquor (legal, not stolen or illicit at either end).

The effect for the mission giving faction was positive of course, the effect for the receiving faction was negative economy, just as it would (more intuitively) be for narcotics say.

yes, but that's a mission effect, not necessarily a bulk trade effect.

 

[Blubbinio[GPL]]

Hey thank you for this time investment.
I've got a short question: You wrote in "Trading for Influence" in 2017 that buying comodities doesn't influence the station, only selling. Do you know if it's still correct?

Thanks in advance

 

[goemon]

Hey thank you for this time investment.
I've got a short question: You wrote in "Trading for Influence" in 2017 that buying comodities doesn't influence the station, only selling. Do you know if it's still correct?

Thanks in advance

yes, that still is correct.

 

[Jmanis]

An example from this morning; I just delivered a cargo mission from one Pirate faction to another in a different system, the cargo was Liquor (legal, not stolen or illicit at either end).

The effect for the mission giving faction was positive of course, the effect for the receiving faction was negative economy, just as it would (more intuitively) be for narcotics say.

yes, but that's a mission effect, not necessarily a bulk trade effect.

Nah, it's also a bulk trade effect. This sticks out to me cos it did surprise me, then i independently verified later.

Thanks @Riverside

 

[goemon]

We would expect the new inf to be 46.6 and the max in the system to be around 53.3.

we would expect >47,2.

max influence change draws from your excellent research, so same number.

Controlling Faction goes from 45,8706 -> 47,3657.

zero traffic besides me, no bountie redeems etc.

raw profit per ton has been 100 cr/t, 250 cr/t, 502 cr/t, 1004 cr/t. 180t each, demand: 1, single source, single docking.

 

[Riverside]

When your model makes a good prediction obviously that's a good sign

But when stuff happens in a way you didn't expect that's the more interesting result, because it means there is more to learn & discover

 

[Vandaahl]

Controlling Faction goes from 45,8706 -> 47,3657.

zero traffic besides me, no bountie redeems etc.

raw profit per ton has been 100 cr/t, 250 cr/t, 502 cr/t, 1004 cr/t. 180t each, demand: 1, single source, single docking.

Awesome result. Thx a bunch for that

 

[goemon]

When your model makes a good prediction obviously that's a good sign

But when stuff happens in a way you didn't expect that's the more interesting result, because it means there is more to learn & discover

i'm in no positions to make serious predictions, speaking of which, this test landed in the lower bond of what i expected. i would have expected anything between 47,1555 and 48,1555...

the kind of edge case testing gives numbers for very specific sets of numbers, but doesn't cover normal gameplay. so if you'd ask me "how much will i gain from a conda trade 2*150+100t @1302, 877, 642 cr/t profit, i just could give you a very roundabout guesstimate (+/- 0,5).

i look at those tests a bit like the role of twin studies in context of genetic heredity. you naturally can't get huge databases, but it keeps conclusions in checks.

i think it's quite understandable why an analysis of (lots of) day-to-day gameplay data can potentially lead to wrong conclusions on mechanics, if not tested in experimental settings (no traffic systems).

to spell it out here:

influence gain and demand

• in day to day play demand will create profit, so you'll see a correlation between demand and influence gain.
• a clean test shows that demand has no effect (Figure 7).

influence gain and total profit

• in day to day play, total profit, which is a multiplication of tonnage and profit per ton, will correlate with influence gain.
• as we are limited by shipsizes on tonnage, and in day-to-day gameplay we'll bring commodities of regular profit (500-1200 cr/t) this adds to the correlation.
• a clean test shows that influence gain is not bound to total profit (Figure 2).
• Further testing shows that there are seperate effects of tonnage and profit per ton.

none of the above has much relevance in day-to-day gameplay (trade with as much profit per ton as possible = influence gain).

but drawing conclusions on the mechanics from analysing day-to-day gameplay data, even a lot of it, without testing the conclusions in an experimental design, can potentially lead to wrong assumptions about the mechanics.

 

[rdc000]

I wonder if you have any plans for further testing .... in particular testing of more than single ship-loads of cargo.

In particular I would love to see a test of: two single commodity trades vs two two commodity trades each of which result in the same total tonnage of each commodity being traded that day.

 

[goemon]

I wonder if you have any plans for further testing .... in particular testing of more than single ship-loads of cargo.

In particular I would love to see a test of: two single commodity trades vs two two commodity trades each of which result in the same total tonnage of each commodity being traded that day.

i think that question has been answered already in fig 1 and fig 9. trading 2x2 commodities of same profit per ton and same tonnage >150t will result in more gain than trading 2x1 commodity of same total tonnage and profit.

i'm currently think about several other tests ...

a) testing the size of the trade bucket in two different population sizes
b) testing whether trade missions add to the trade bucket or whether there is a different
c) testing 4, 3, 2, 1 commodity with a total tonnage of 600

alas, currently playing the bgs and some other ingame acticities have more focus, so i'd be as happy to read anybodies test on it as ours