Trading for Influence II (FC Update)

This thread details the influence effect of trading at a factions market after FC update. It is a follow up to this outdated thread: https://forums.frontier.co.uk/threads/bgs-trading-for-influence.357715/

CMDR DaddaTheButt of Canonn research and me teamed up to revisit trading for influence. Total testsample was 67 trade runs without interference. Thanks to the BGS-Chat of Canonn and Dommaarraa, who blind tested some values.

I have also added a “How to test” for each fancy graph below, so you can see for yourself:
  • all tests have been conducted in no-traffic systems; tests where traffic was reported were discarded and repeated.
  • it’s better to test on low influence levels, as rounding occurs
  • it’s complicated to test low influence gains, as rounding occurs

one definition up front:

(influence-) “gain”/”gross”

refers to the gain of influence before the actual influence is taken into account. It allows to compare the effect of an action for a faction e.g. at 35% influence and the same action at 55% influence. i used to call this “raw influence gain” in the outdated thread, but Dadda made a reasoning to call it gross or simply gain, so for now it’s “g”. The original concept is of coatsilver 2015, and was crucial predicting the effect of superpower bounties patch 2017 (https://forums.frontier.co.uk/threa...-mechanic-and-bountyhunting-after-2-3.346177/) .
For calculating the effect of a faction on a single faction without any other action in system, we went with that well established
n : New_influence_delta
i: Old_Influence
g: gross, gain


n=(i+g)/(100+g)

transformed that’s

g=(100(i+n)-100i)/(100-(i+n))
So, which elements of a trade do affect influence?


Not: Source
There has been some discussion, whether source of a commodity has an influence effect. the idea here is, that buying a commodity from different markets has an effect on influence gain, while buying different commodities from the same market has no influence effect. Well...

kR160ym.png

fig.1

  • Source has no effect.
  • Different commoditioes have an effect, even if bought from the same market.

How to test:
- Buy 2x >150T of the same most profitable commodity from different markets and sell it
  • Next tick: Buy the same total tonnage from only 1 source, sell it, compare
  • Next tick: Buy two different same profitable commodities from same market, and sell it, compare.


Not: Total profit
There has been some discussion, whether total profit determines influence effect. It does not:

8WGdMHa.png

fig.2

- Total profit has no direct influence effect.

How to test:
- Buy 100t of a high profit commodity, sell
- Next Tick: Buy significant more t of a commodity, which you can sell for the same total profit.

So, if it is not source nor total profit, what is it?


Profit per ton

Profit per ton is one metric for influence effect. As you can see, profit per ton is pretty straight forward. Inflöuence gain increases, until it caps out ~2000 cr/t.

nftwZZ8.png

fig. 3

How to test:
- Sell the same tonnage of >180t with increasing profit/t each tick

Tonnage

The effect of tonnage (here with >2000 cr/t profit) has an interesting shape:

XosmBRu.png

fig.4

How to test:
- Sell increasing tonnage each tick with >2000 cr/t profit

If you combine “Profit per ton” and “tonnage”, it explains, why selling smaller batches of different commodities doesn’t yield extra influence. You first have to cap out the tonnage effect.

Effect of Population

This was something i wanted to test since forever. How effects population influence gain by trade. Turns out it doesn’t:

Bh2f66a.png

fig.5

Taking this excellent thread of Jane Turner on maximum influence change (which we didn't retest) https://forums.frontier.co.uk/threads/influence-caps-gains-and-the-wine-analogy.423837/, calculating that maximum influence change for a faction at 0.99% as 36-log2(Population), and calculating with that maximum gain at current influence as above ... - here is the percentage of Max Influence Change for the same trade in two different sized populations:

vy7p5Zx.png

fig. 6

This surprised us a lot. It leads to the contra-intuitive clue, that you need more trade(s) in small population systems, until you hit the cap.

How to test:
- Trade at two different population sized systems the same; compare.

Demand
Demand still has no direct effect on influence gain.

kx1wr3D.png

fig. 7

Cave-eat
- small tonnage sells (<50t) have always the same influence effect, or non. Those minimum influence gains are not additive if sold at the same docking. Here is a scatter plot of 1-50t:

7hBssq7.png

fig. 8

Take aways
- The current implementation leads to large trading ships having a function. Basically you should take ~150t of the most proftable commodity, and add a second one, until you reach ~150t again. That means, in a conda you gonna trade 150+150+100t, in a cutter 3x150t+1x210t, and in a T9 4x150t+120t. This also makes surface ports with markets valueable assets.

VBcnugJ.png

fig. 9

- For outpost trading, you simply bring the most profitable commodity. I have upped my BGS python from 128T cargo to 160t cargo in reaction to this test.
 
Last edited:
So something that isn't clear to me.

Oftentimes the most profitable trade from many of the stations I operate in is <1000cr/t.. mostly owing to many of them being hybrid economies eating themselves.

If it's less than 1,000cr/t profit, is 150t still the best aim?

Also, good to see cargo diversity being meaningful, shame it doesn't apply at lower tonnage though.
 
The tonnage and profit sigma are independent, so yes, it works the same below 1000cr profit.
Should note that the cargo sigma becomes linear rather than diminishing, so it's not the end of the world to go over 150. 300 is twice as effective as 150.
 
So something that isn't clear to me.

Oftentimes the most profitable trade from many of the stations I operate in is <1000cr/t.. mostly owing to many of them being hybrid economies eating themselves.

If it's less than 1,000cr/t profit, is 150t still the best aim?

Also, good to see cargo diversity being meaningful, shame it doesn't apply at lower tonnage though.
you are getting ~80% of the max gain by profit/t at 800 cr/t profit (~80% of what you'd get at 2000 cr/t). and still 25% at 150 cr/t profit.
 
The tonnage and profit sigma are independent, so yes, it works the same below 1000cr profit.
Should note that the cargo sigma becomes linear rather than diminishing, so it's not the end of the world to go over 150. 300 is twice as effective as 150.

the factor is around 1:0,7. so after ~150 t you get for doubling the tonnage around 1,4 times the influence gain and so on.
 
I'm probably misunderstanding but wouldn't 3x 50t (3x 0.6g = 1.8g) be better than 1x 150t (1x 1.25g), assuming that all three have ~1000Cr/t profit?
Maybe allowing for the cut-off at the 50t mark by upping that to 3x 55t or 60t.
 
I'm probably misunderstanding but wouldn't 3x 50t (3x 0.6g = 1.8g) be better than 1x 150t (1x 1.25g), assuming that all three have ~1000Cr/t profit?
Maybe allowing for the cut-off at the 50t mark by upping that to 3x 55t or 60t.
ah, interesting, and you are right - the curve looks like it. but we haven't tested that combination at all. we only tested smaller samples (10 times 15t). i'll put it on the list of tests to do!
 
ah, interesting, and you are right - the curve looks like it. but we haven't tested that combination at all. we only tested smaller samples (10 times 15t). i'll put it on the list of tests to do!
:) Sorry to make more work for you.
And I have some self-interest as 3x 50/60t fits better in my Python.
 
does this work for Black Markets, as well?

or does it only apply to legal goods traded?
we have not tested black market trade at all.

the 4 year old test on that showed profit playing no role, but tonnage. might have changed.

@Dommarraa tested his "hydrogenbomb" (selling for loss) against positive trade, looks as those are symmetrical (and both capped at ~2000 cr/t/.
 
I'm probably misunderstanding but wouldn't 3x 50t (3x 0.6g = 1.8g) be better than 1x 150t (1x 1.25g), assuming that all three have ~1000Cr/t profit?
Maybe allowing for the cut-off at the 50t mark by upping that to 3x 55t or 60t.
i now remember, why we didn't tested that further: because it looked practically irrelevant.
you'll struggle finding 2 or more commodities making most profit the same, especially with the bulk tax. so practically you will be better off by trading a full load of the most profitable single commodity at ~150-200t.

even if you use your FCs market to set the profit to max profit at sell, you have in most economies the problem, that the commodities with >1200 cr/t galactic average don't have enough demand at M-Pad stations to generate for exampel 1000 cr/t profit with 3 different commodities, and more so for those commodities with >2200 cr/t galactic average for full 2000 cr profit effect.

i anyway will look into it, goes well with some other testing i intend - but it's more an academic question as far as i can see. i think, i'll test 800-1400-2000 cr/t profit and a combination of those as well.
 

Jane Turner

Volunteer Moderator
We've found only two factors matter. Total profit and Demand, and the same log base 2 term applies up to 1 billion profit. It doesn't matter how you make that profit. Clearly the effect varies with stating influence and population.
 

Jane Turner

Volunteer Moderator
Its a few thousand data points from a number of groups, normalised to population and starting influence.

We've shown to our own satisfaction that the effect of all activities in the game follows the form

inf gain is related to log base 2 [value] x [modifier]

where the value is greater than one unit.

In the case of trade profit is measured in 100K, and the modifier is affected by demand 0.5 when there is a demand, substantially less when there is no demand.
 
Its a few thousand data points from a number of groups, normalised to population and starting influence.
i think, that's a problem then.

because, while i'd say datadriven is perfect for many things, it's not set up to provide a test of the mechanics.

exampel from above:
-in normal gameplay you are not going to do edge cases. for exampel selling with the same profit of high demand and no demand, or selling 2 different commodities for same profit and compare that to one commodity.
  • so you'll get no datapoints on those mechanics.
  • you interpret your data points.
  • we tested the mechanics (some of those assumed by your impressive datadriven studies - it was great to have something to test!).

i'm pretty sure, you can interpret your data points to "demand has an effect", but it does not hold up to the scrutiniy of testing the mechanics.

please run a simple test in a no traffic system with for exampel total profit as above, 100t vs 350t and same total profit - i'm pretty sure you'll get the same results as we did. total profit is not what constitutes the influence effect.
 

Jane Turner

Volunteer Moderator
All I can say is that we have been able to predict accurately the effect of trade in systems to within our ability to measure (ie the rounded influence figure) and the only factors we need to know are starting influence, population, demand and raw profit. I can't really say any more other that the results for demand and no demand fall on their own separate curves.
 
All I can say is that we have been able to predict accurately the effect of trade in systems to within our ability to measure (ie the rounded influence figure) and the only factors we need to know are starting influence, population, demand and raw profit. I can't really say any more.
how about i give you system stats, and what i'm gonna do, and you tell me your prediction, so i can test against it?
 
Top Bottom