Successful vs Failed actions.

There are a few meanings of a "successful" action, but I can answer in several ways.

If the goal is to find measures that a computer can easily evaluate as good or bad, then one example would be paying debt with cash while someone else holds your stock. Since buying back stock before paying debt is preferable, paying the debt first indicates that a player is in desperate straits. If a player has cash reserves, then losing in the market is also an indicator of poor play: if someone buys food and you buy food right after, then you should have bought food first. Losing on sells is an indicator of poor play regardless of cash reserves. But care should be taken that such losses are based on true market pressure; if a player buys and sells 100 food every half second, then this would indicate "poor play" for every other player holding food, having lost on every single trade. Clearly, no such poor play exists, and this price manipulation is all artificial, with no actual effects.

If the goal is to evaluate a player's strength, then a good indication is moving into an industry before everyone else. Players usually only think about something like power or water when they see "uh oh, power is at $300, time to build solar panels". But recognizing the upward trend beforehand is more uncommon, and better players can grasp the future better. It's an even stronger play when the resource can be stockpiled, such as food: get into food first, build a stockpile if you don't have pressing cash needs, and then sell the stockpile after everyone else moves into farms. But this skill becomes less important as ROI gets higher, because excess cash goes into investment instead of winning markets. Another indication is how much time they spend looking at everyone else's stuff, instead of their own stuff.

If the goal is to improve one's own strength, then the most I can say is looking for arbitrage. This could be simple, such as if an Offworld makes $500/s vs an additional PD at $600/s. Or, if buying a player yields 100%/day ROI, and a transition yields only 80%. But it could be more complex, with opportunity costs of eliminating players, or exclusive patents, etc. The arbitrage then revolves around significant timepoints, such as stock-buying time/tile-grabbing time/patent time, and then the worth of everything is decided by these timepoints, adjusted appropriately. But this is more of an AI issue; humans find it difficult to do such calculations.

I guess I could add a few more to the list.

Successful: getting a high tile, getting good resource adjacency, buying into a short appropriatly, succeeding in a research, making good use of a hack, spending your claims in a way to mitigate black market, removing debt, being in many different markets, punishing opponents without access to basic resources, selling useless starting resources first, identifying opponents despite masquerade, getting cheap goon squads, building an offworld.

Unsuccessful: unsucessful black market attack, commiting resources that you need in the short term (for example upgrading when that could allow someone to buy you),  losing a patent to someone  else because they finished the research first, not adapting to changing weather conditions, being unable to gain access to at least half the basic resources (silicon, carbon, water), failing to finish a buyout, running unprofitable buildings, gaining debt at a fast rate (700 a tick), losing the power race, investing heavily into research on a particular resource that does not have high demand, having to 'join up' claims making them a better black market target, getting a building mutinied and ahving no response, not obtaining half of your own stock.

Now that I'ev been writing it, I'm having trouble understanding what exactly your asking for, your example list seems really broad in that both really small things and really big things are mentioned, as well as things both in and out of the players control. Could you perhaps give us more or otherwise make it more clear what your asking here?

@Jaiwera All of these things contribute to why I'd like a discussion to be started. Especially with weightings attached to the actions.

@Gameslayer I'd say my examples were successful (although I likely should have weighted them myself) as it was intended to be a rather broad request.  Actions a player can take or allow in particular are a point of interest. How good is it to buy some stock? Better than researching tech on average? How bad is it to have some of your stock bought? "Why" is also an important question, but not necessary for now. Could make for an interesting argument though.

I know it's a rather broad potential list, and because of that there are likely plenty of factors I would miss if I tried to come up with and sort it alone. So I thought an attempt at a discussion was appropriate.

1. Evaluating the map and when available, claiming a monopoly or a semi-monopoly over a resource with just your found (and later, with BM).
2. Founding in a location with the best opportunity to upgrade.
3. Keeping the appearance that you're not really a threat by buying claims and not getting in the lead in upgrades.
4. Evaluating what everyone has in stock (by looking at their production, monitoring resource prices, looking at their wealth reserves and how much of it is in resources and making a few clever inferences along the way).
5. Identifying a key point of weakness in the opponents production chains and exploiting it.
6. Ensuring that your opponents must ship in their resources from far away (or simply not produce a resource at all to save on shippingl).
7. Ensuring that you ship in resources from nearby.
8. differentiating between upgrading 2 minutes later and using an EMP or Power Surge at HQ2 (a common scav mistake).
9. Recognizing which markets are going to raise in price and which are going to fall over the next 2 minutes.
10. Knowing when to rush for an OW and when ONW is more valuable.
11. Buying someone out when you can upgrade instead.
12. Not putting too much stock in the revenue displayed in a structure its mouseover information.

I hope the value of these actions is self evident.

On point 9, lots of people often build additional farms just because at that very moment food is expensive even though fuel will be too. By the time you can in that situation sell food, reactors will be more profitable because people won't have sold any of it because they havent produced any.

On point 12, the average value of resources depends on how much you do or do not need of it. For example a nonscav needs 800 steel to upgrade to HQ5 (and some 200-400 steel for other buildings). Even when steel is at 15$, the average price of steel (if you were to purchase 1200 units of it) would be well over 15$ and thus makes the mouseover revenue misleading, moreover there is an additional cost in time and resources when swapping buildings out.

Successful:

• Being patient to put down your first power plants taking into account the random power events that normally happen around day 2 or 3 (wind farm/solar flare)
• Using an EMP in a not so obvious spot when your opponent has purchased a goon squad and protects the center cluster of their resources
• Paying attention any time the underground nuke is in the BM and carbon is scarce (don't found scavenger or you'll defend your carbon all game)
• Crushing water on a low water map when you found robotic to put opponents at a disadvantage (particularly a late robotic found)
• Identifying that the high aluminum clear across the map is often not as good as the low adjacent aluminum near your base

Unsuccessful:

• Placing power at HQ 1-2 and sometimes HQ 3 (unless going for a complete power play)
• Ignoring cook the books on the BM until debt gets to D levels
• Being too aggressive in the BM (particularly as scavenger) to the detriment of your own expansion (those BM purchases start adding up!)
• Trying to rush to HQ 5 by only producing upgrade materials (carbon, iron, aluminum, glass) in a majority buyout environment. Frequently the prices crash and you will have no money to defend yourself from a majority buyout
• Placing 2 offworlds in a strong BM environment with dynamite and mutinies available (also see: final game of tournament- blackmagic)
  
  

I'll edit and add more as I think of it.

Off the top of my head: 

Successful: 

• Identifying opponents based on their play style and treating them accordingly. Skill level should be taken into consideration as well as in-game position. E,g, don't let a strong player recover from D debt. In general, identifying threats is crucial, but not so trivial in recent versions.
• Evaluating correctly how high to bid on auctions.
• Planning ahead when using claims. E.g. claiming dry ice when available, even if you don't immediately need it; expanding in certain direction to connect remote claims or possible future claims, such as geos; placing non-harvesters on basic resources so you can rotate in the future etc.
• Using cheap EMP's to spot opponents goons. Especially when holograms are in use. There's a high chance these goons are/will be protecting offworlds. Knowing about their location is vital. Freeze is a nice bonus, of course.
• Blasting through opponent's goons with cheap BM effects that can't really backfire, such as NV, CO, nukes...
• Combining BM items to increase the effect - double mutinies, blasting through goons + power surge, mutiny + goon squad etc.
• Keeping an eye on everybody's cash situation and identifying vulnerabilities - when someone upgrades and has nothing to defend with or debt spiraling. A cheap buy is better than expensive buy more often than not.
• Buying stockpiles of cheap resources even if you don't need them and selling when the price goes up.
• Upgrading and switching production when base resources are cheap. 
• Buying stock (including your own) when it's cheap. If you have a lot of debt and planning to pay it off, buy stock first, pay off the debt afterwards. 
• Watching the time and paying off the debt before the interest is added.
• Queuing research/patents when chemicals are cheap.
• Crashing power price asap when your opponents are trying to pay off debt and you are not.

Failed:

• Overbidding on auctions and as a result letting the debt get out of control.
• Making yourself an obvious target - usually just upgrading faster than everybody else is enough to draw unwanted attention.
• Not defending dying players. 
• Not being aggressive enough against the leader and expecting others to do the fighting.
• Being too aggressive against the leader to the point you let others catch up and surpass both of you.
• Trying to pay off debt when you don't have to thus making less cash.
• Attacking someone for non strategic reasons such as just to get even, and tilting in general.
• Switching production and upgrading when basic resources are very expensive.
• Investing into not very profitable patents.  
• Investing into patents and research when chems are too expensive,
• Being late to sell and/or buy into shortage.

The list will go on and on. Frankly, I don't see how we can weight all the actions since most of them are situational. Something that can win you one game will lose you the very next one. 

I'm finding this thread very interesting because I'm actually working on developing an agent that learns to play Offworld Trading Company using reinforcement learning.  What that means very simplistically is that as the agent plays the game, it will receive positive or negative feedback to help it pick better actions to take in the future.  Obviously, we can have the agent get big positive feedback when it wins and big negative feedback when it loses, but we expect that it would be helpful to provide some other feedback for actions.

Of course, the problem with that is exactly that having a computer program recognize a situation where the agent has clearly done the right thing or clearly done the wrong thing is . . . challenging.  And I would lean heavily toward having rewards and punishments only apply when they clearly are accurately reflecting movement toward or away from winning the game.

Since I'm a mediocre player, I would really appreciate any additional thoughts some of you better players have in regard to identifying what those clearly good or clearly bad situations might be.

And, btw, many thanks to the Mohawk team who are making it possible for me to already be working on this.

One other thought:

I've considered a reward for upgrading that was reduced by the amount of time it took to make the upgrade.  What do you think of that?  Should something like that only apply to levels 2, 3, and 4 or even only to levels 2 and 3?

Thanks in advance for any thoughts.