How to Fix the Bot in Four Simple Steps, without Complicating Things.

[rando24]

The sales are hard on you guys, I'm sure. Once the servers stop firing at all frequencies it is the best opportunity to take the time to improve the already useful service, which SteamCardExchange (dot) net provides.

[1] Step one, stop adjusting prices.

Every time the bot changes prices it loses money, so stop doing it so frequently. It's virtually impossible to profit from differences between bot price and market price, because of how much of a pain in the Valve it is to buy or sell cards on the Marketplace. The Steam Marketplace is garbage, from crazy fees to 1 week delays, from wrong price reporting to shady currency conversions. I hate it, hate it, hate it. SCE to the rescue!
Change prices once a week, once a day during sale if you must. Better yet to pick random intervals.

[2] When you calculate price average don't include the one outlier. Calculate average of 4/5 cards excluding the expensive one.

Market squeezing has become very frequent and even took to regular cards of rare games. It is ubiquitous in foil card trading. However it is stupid to adjust all cards to the average when some sleazebag trader bought up all the stock and now charges an extortionate price for the one card everybody seems to be missing. You may not understand this or believe this is the case, but regardless, we should all agree that in statistics it is normal to exclude outliers. It's the difference between (geometric) mean and average, so I'm proposing to use the mean price (strictly the most common price out of the set) of a card rather than the arithmetic average.

[3] Don't adjust price when the change is less than 1 credit.

This is the gist of it, how exactly the mathematics works out is for a clever guy to figure out. It is necessary to remember the last weeks price exactly [Example 5.2] When next weeks price is [5.8] the price shouldn't change to a (6), when it's [6.2+] the week after that then it should be adjusted.
1 Credit is just an arbitrary threshold that can be raised or lowered.

We can hash out the maths when there is a willingness to do so, perhaps it can be as simple, as just rounding the prices down, instead of simply rounding them.

[4] Inflation (time sensitive)

The Steam Holiday Sale raises all card prices because of the additional "The Steam Awards" bonus received when crafting a badge. This year something extraordinary happened, Steam reduced this bonus by 20% by making the Badge bigger. It's a good step because gems were dropping value quite fast, but this lowers the value of badges and cards.
Once the sale is over prices are going to drop, even more so than after previous sales, because of the correction Steam made.

The bot however shouldn't drop its credit prices! 30 credits is a much better price for the cheapest badge. It also makes the bigger sets much more attractive and fairly priced. 4 credits for 1 card out of 15 don't look so ridiculous any more.

So yeah the bot should use the coming price drop to devalue "The Credit" covertly, and raise the prices without actually raising them.

Bonus Suggestion

Add a +1 credit payout for everyone who sells a missing last card to the bot. This is a legit and useful way for people to make credits. This will raise the rate at which last cards are sold, increasing the bot's profitability over-all. This will also heal the market a little making it less frustrating and help spoil the usual scams of market manipulators.

Thank you for reading. I will likely make several edits.

[GermanDarknes]

Like always thanks for feedback but again most of these things are already in use.

We won't pay +1 for a missing card, already discussed a lot in here.

[rando24]

OK, give one example, please.

[GermanDarknes]

1+2+3 are already there.

[1] It depends on the game when it refreshes prices.
[2] We remove outliners from the price calculation, we tried different algorithm for this and find a suitable one that works quit well for our data.
[3] This is already the case, there's a specific threshold until it changes prices.

[rando24]

Well [1] is the most important, and the goal is to break the dependency on the BS up-and-down prices of the marketplace, while having a reasonable price. As long as people can buy low, sell high with no risk, all is for naught. Not to mention that a trader can even trigger price change that way himself, in theory.
If the profiteers can do it only once a week, cha-ching bot is growing inventory again, 6 days out of 7 anyway.I suggest one week to reassess price, because that's the Steam market delay, but It would be an easy thing to test even longer periods of "price freeze".

Is this communism that I'm suggesting, prices that don't change!? Quite the opposite. Capitalism is based on resolving the tiniest conflicts of interest by negotiating price, I suggest that the bot becomes a TOUGHER negotiator, demanding its own FIXED price. Rather than accepting prices that some ignoramus triggers, when he sells his/her cards into a 3 cent low-ball buy order on Steam. Or when some gamer buys the last card for 10 times its price, because He REALLY likes this game called Oppai Senpai.

Right now the bot is being slapped around by the market. First step is to stand up for itself, people have no alternative, between the 2:1 forum traders and the sh!t marketplace, where are they supposed to go? Any price you ask they are gonna pay. More "normal" users will join once there are some normal cards in stock again.

[rando24]

Originally posted by GermanDarknes:

Like always thanks for feedback but again most of these things are already in use.

Even though this may seem frustrating at first glance to both of us, it shows that we are somewhat in agreement!

There are some differences. I suggest random intervals of price freeze, you say the bot is already keeping prices steady at varying intervals. But they aren't the same thing. Adjusting to the market price is the opposite of random, it is completely predictable.

[GermanDarknes]

You call it fundamental or frustrating, but be aware that not everyone has to be your opinion.

Most of the stuff you wrote it already implemented or tested in many different ways, always you find new stuff you want to implement that we completely disagree with, that's why we don't implement everything you write. Of course we are open to feedback, but don't expect we just go with it, it's not that we aren't testing stuff.

Of course you will now write how much better this would be and so on, but there's no reason to expect it, it's just another idea of you that will may work, or work not. So thanks for the feedback, most of the stuff you wrote here is already implemented and that how it should work. Other ideas are always welcomed.

[rando24]

Well, if you rather do things than talking about them, I can certainly respect that. Still, it is a missed opportunity to not discuss matters. The rigors of discussion can help uncover the truth ...

... but in the end - I agree - only the experiment can truly test an idea and show its veracity, an experiment that only you can implement. Me, I'm just a curious, humble observer, I'm thankful for the services the bot offers, but I feel guilty taking advantage of both your work and people who donate to the bot.

Your pricing algorithm is going to be a tremendously helpful tool to determine and compare the inventory value, in case of a bright and profitable (hypothetical) future in which the prices have been frozen. Again I can only guess about which tools you use to take stock (inventarisieren) and balance accounts. I'm not a accountant professionally, maybe you can find one to talk to.

Certainly It's an easy thing to criticize and even though my understanding did evolve, you'd be hard pressed to find either inconsistencies or contradictions in my suggestions. I certainly advised price stability before I understood fully why it is necessary.