Dynamic trade values (mayhap) coming after 0.70

[Zsar1]

That is a lot of events to maintain. 34x4x4=480 events vs. current 4x4=16, without increasing granularity as you propose.
If we limit this to the 20 tags above 30 IC, still 20x4x4=320 events.

We will have to benchmark whether there is any speed advantage and at these numbers, I am almost ready to wager that 320 single-country triggers will evaluate slower than 16 triggers on 87 tags.
... But most probably we will not gain significant differences, as other factors require much more processing power - as is evident by the major slowdowns at the most significant war declarations (ENG-GER, GER-SOV).

But I made my points. If they do not convince, so be it.

 

[dec152000]

Hi,

Yes, 320 is a lot of events. At least most of them would simply be cut and paste with a minor edit for level targets. But it is a lot of work to do all that. OTOH, getting some dynamic value to the pricing now is indeed better than static pricing. And it will give us some experience watching it. So how about we use the pricing you are proposing with somewhat lower stock targets and greater event frequency? Say double the event rate and half the stockpile target. This wouldn't require such giant Resource pools. Also add more tiers:

500: USA
300: GER, ENG & SOV
100: FRA, ITA & JAP
40: HOL, SPA/SPR, & POL
30: IND, CHI, CAN, BEL, CZE
20: AST, HUN, ARG, MEX, BRA, SWE, ROM, TUR & YUG
10: AUS, SCH, SAF, DEN, POR, NOR, FIN, GRE & CHL.

Everyone else could be static as they really are all bit players below this level. OR a few could be added to the 10 group if desired.

One other thing here. The 0.70 fileset is substantially different from 0.62 in trading balance. I spent a lot of time fixing the starting trades so they wouldn't fail early on. Part of the problem involved initial trades that failed roughly 1 week in d2 Convoy shipments starting lower than steady state. Another issue was sending too many Resources off to minor nations and having them pile up there. While things may get tight in 0.70 I''m seeing less issues with trading than there were in 0.62 IIRC.

mm

 

[Zsar1]

Hè. This is kind of what I was proposing in the first place... or thought that I was.
Anyway, very agreeable. ... Actually, but untested, there should be some lower limit based on the AI's trade frequency, below which stocks cannot be fitted to frequency anymore; I do not think we will hit it this time, but it is something to look out for.

Will bundle the package within about a day.

 

[dec152000]

Hi,

A minor shift here. I'm thinking 100 should be 150 instead since both JAP and ITA should expect trouble with imports in a war.

mm

 

[Zsar1]

Mmh. Was distracted by unrelated... distractions.
Also just found that your list uses a different measure than mine: I listed upper limits, you list lower limits; guess what I finished before noticing.
... Knew something looked fishy with those tags...
Luckily the difference is miniscule. I wanted a couple of low-IC countries inside the limits, notably EGY, PER and SIA, so I set the cutoff point to >5 IC instead of the proposed >=10 IC. Other than that, the specifications you have provided are met.

I also built the refined oil triggers. Cannot say anything about them yet, other than that mechanically they work.

Mulling it over for a bit, I figure that we may want to adjust the stocks for those resource events again. Also the frequency: One month is not a long time at all, half a month is probably not enough time for a price change to show any more-than-minor effect whatsoever.
Similarly, it seems suicidal to react strongly only when a mere one month of stock is left; I as a player would certainly grow restless long before.
- Contemplating it further, the original two months do seem like a fine line to start feeling panicked. Mayhap we should indeed have shortened the neutral segment instead.

In any case, we can deploy this as is for now. Mayhap my worries are unfounded and all will be fine - if not, we adjust, simple as that.

 

[dec152000]

Hi,

The thing here is that a "one month" stockpile based on the event tiered IC is likely to cover vastly more days than that in game. ITA being near worst case imports roughly 60% of it' energy. So that means a 1 month tiered supply is more like 2 months of usage. However ITA already is importing a lot of Energy. So the actual daily deficit when the event triggers is likely to be a lot less than 60% of consumption. So even more time to correct. It will be informative to see it work.

One other thing. In the past you made a comment about excess stocks being no big deal. I'd counter with the thought that everything has a price and thus extra stored Resources directly cost Divisions. It is far better to have the AI store enough Resources to keep running at 100% rather than running short. But anything beyond that point is just hurting the AI. The trick is finding the balance point. Hopefully we are close to it!

I'll try and develop some new initial stockpile numbers for you to test over the next few days.

mm

 

[Zsar1]

You do have a point with that. However, then it appears even less sensible to check in less-than-one-month intervals.
... I have to mull over this some more. In the meanwhile, some more live tests seem like a good idea.

Mayhap it will be best to switch to a non-linear function for calculating the price segments. More buffer on the low end, less at the top. My initial design concentrated on being as safe as possible, but a more greedy variant may indeed be superior... hohum, hohum...

Ah, something I want to clarify: It seems that you equal neutral (opposed to decreased) pricing with lack of sales - but of course countries with increased pricing will still be ready to buy for above-neutral prices, countries which favour each other in trade will be willing to pay more for less (so one country with neutral pricing may still buy from another, while at least one favours the other).
When those do not offer sufficient demand, the decision to dump price or to stockpile until the economic situation improves is not a simple one... that is what I wanted to model with a larger neutral-priced segment.
So, five months of stock before prices fall below neutral does not equal a "stockpile at least five months" directive but rather a "only sell it for what its worth unless the stockpile is truly excessive".

- Selling cheap in bulk may still end up being more beneficial than waiting for clients at deserved prices - it certainly works out for discounters in the real world. I wonder if somehow both strategies could be included by elegant means.

 

[dec152000]

Hi,

I tend to look at this more from a buyer perspective. I know it goes both ways. But the really critical trading behavior is getting Tags to buy resources before they run out. To many times you see the AI fail to do this when there are possible trades waiting to happen.

Check interval: Checking frequently is great as long as it doesn't slow things down. Just gives the AI more time to make trade offers. But it indeed may not be needed.

mm

 

[Zsar1]

Grr. Forgot to include setting puppets to aggressive pricing per default. Reuploaded. Only core_AI_trade.txt has changed.

 

[dec152000]

Hi,

One other thing here. We should statically set the value of Money and Supplies to a custom setting. This should bbe based on the actual production rate at game start. So an FI would normally value these IC produced "Resources" the leaast while a SB would value them the most. This would tend to make the game better mimic RW trading patterns where industrialized nations exported finished goods and imported raw materials.

mm

 

[Zsar1]

True. This can be done directly within the AI files, I think, while there are no countries which transition between these states.

Another slip-up: There should probably be some handling of oil-producing puppets (IRQ, ROM, PER...), but while there is not, oil price should be neutral for puppets.
Going to fix the latter later (and notify once done); if you want to use the code in the meanwhile just move the "ispuppet = -1" in event 35413 from NOT to OR.

addendum: done

 

[dec152000]

Hi,

Shouldn't Puppets always be in Sell mode? Since they only Trade with their Master they should always offer the best price if the Master initiates trading. Most likely if a player needs a quick resource infusion and hits his Puppet up for the 1000 point stash.

mm

 

[Zsar1]

That would prevent them from buying if they need. For supplies or money which they do not automatically contribute.

...
Countries which net consume oil and may be puppets include: HUN, MAN, SIA.
Countries which net produce oil and may be puppets include: IRQ, ROM, PER.

I do not want a setup where either side is disadvantaged automatically and without good reason. At neutral oil price (which needs to trigger if the country was not on neutral before becoming a puppet), whether they buy or sell will be dictated by their master's pricing.

Admittedly, there is a weakness when the master has an abundance of oil and sells cheap, while the puppet also stockpiles oil - but this is an uncritical case, I suppose.

... Mayhap I will just remove the puppet clause from oil trading entirely, should this arise as an issue:
We already know that, unlike IC resources, the AI actually aggressively stockpiles oil while the supply allows it. At the same time, no resource is as close to being a "dump good" as oil is, especially for puppet candidates. So, if they end up stockpiling oil which they pay their master for in money and supplies, I am enticed to just let them, as long as the master's supply is not threatened by this - which it should not due to pricing changes and causally cancelled (in more extreme cases even reversed) trades.

Incidentally, as they maintain highest pricing for all the IC resources, the relative value of those and oil should forbid trades of IC-resource-to-master-for-oil-to-puppet.

 

[dec152000]

Hi,

What I'm suggesting is they should be in Sell for all Resources. The Master(AI or Human) may gift the Puppet any needed Resoureces. But if a player wants something form his Puppet he should get the best price for it. Puppets definitely need to be hard set on some level d2 the limited stockpiles they are allowed.

On a different note, I was thinking of the tiers again. Once we get some testing on this I propose that we eliminate the top two tiers and do custom events for those nations. Simply no need to trigger check all those other Tags and the odds of any AI vaulting to 100+ IC is miniscule. This will also allow us some customization for these nations. So we should also get a better result.

mm

 

[Zsar1]

On the first point: Definitely for the IC resources. For oil, let us leave it at neutral for now and wait for the first few tests. I do not want a puppet to buy all the oil it can get and then be unable to trade it away again, because it has no net income (which limits the amount tradeable regardless of stockpile).

Concur with the second point. Custom events should start from the top, we can iteratively work downwards until overall behaviour is pleasing.
... There is a chance that internally an OR evaluation stops as soon as the first true value is encountered. I sorted the checks accordingly. In that case adding tiers atop would not change much in term of performance impact.
- Probably not worth to profile for that though. As I noted before, the event impact to performance is still a fair bit away from being problematic (at least on my hardware... but we would probably read from people for whom this is different).

 

[dec152000]

Hi,

Since USA is so much bigger than the rest we could just start with that one a do USA only. That also averts the issue of tier swapping suddenly forcing a big shift in target prices. Instead we can gradually ramp up the desired stockpile targets.

Puppet Resources: Not why you are concerned with this. AFAIK Puppets are limited to stockpiles of 1000 with all of the excess transferred to the master. On that note, in my test game IND is right at the limit and has been gifted a daily 0.1 Rare trade from AI ENG to keep it's total Rare trade at 0.0 exactly. So the AI will certainly offer trades to support Puppet economies.

mm

mm

 

[Zsar1]

Ugh, until just now I was convinced that oil was not among the capped resources (supplies and money are not), but a quick test showed that it is indeed not.

That of course changes things. Feel free to make event 35415 instead of 35413 the default for puppets. Going to do so on my end as well.
- Will fix the upload later.

... I hate when this happens.

--

As far as I am aware, gifting is a feature of trade/favored entries. GER may offer free trades to ITA, HUN, ROM, once they are allied and at war. Similarly, MAN on occasion offers free trades of metal to JAP, although the latter is the master.
Never delved too deeply into it, though. Mayhap it would also work for master/puppet-pairs without trade/favored entries of each other. Being at war and allied certainly seems to be one requirement, however.

 

[dec152000]

Hi,

Being At War is definitely not required as I've seen it happen plenty of times as ENG prior to war and my current test game shows a "gift" in place with no war. The Alliance part almost certainly is.

mm