P2P Master AI [Machine Learning] [1-Click Account Builder]

[Aeglen 297]

1 hour ago, PPTouch2 said:

Banking/GE is still incredibly flawed. The buy/sell ranges seems to be based on nothing at all but speed trading. Buy at 4 ticks up (I think +20%), sell at 2 or 4 ticks down (I think -20%). I could be wrong. Is there any method to this madness?

Also, I was doing herblore for 1.5 hours. I had 500k cash. It buys 250K worth of herbs, so like 50 - 75 herbs, then cleans and sells them. If I'm doing herblore for 1.5 hours, why not spend the whole cash stack on the herbs and clean them? Add that to the bleed of G.E. buying above, selling below, plus taxes. Herblore at the GE is difficult to watch.

Really got me wondering how much money is lost from g.e. recklessness and how much inefficiency is introduced from poor purchasing mannerisms . Sometimes I have to hop on and clear off excess gold bars, steel bars, coal, fish, chaos runes, etc. Other times I'll see crafting for 1.5 hours. purchased 1500 gold bars and 250 diamonds, with plenty of gp to work with. Then they'll go back after and get more supplies. This behavior is consistent and just unreasonable.

I'm incredibly thankful for this script. I just hope to stop seeing responses that are "it'll do that sometimes." or "each account is different". Some mannerisms are just too inefficient and shouldn't be allowed to happen.

1) this is not the actual price you get, when you put an order at +20% it can pay 0-20% more, almost always close to 0%. The % increase/decrease is less for expensive items.

2) this is due to spending limits, the bot likes keeping some liquid gp. Imagine if it skipped task/stopped and all your gp was in herbs. Herblore is an inherently gp-negative skill and the alternative of potion making is at least 3x more money draining. Overall, the bot earns more than enough for its bonds under default settings and reasonable levels.

3) very little (see 1), and surpluses will be used for future tasks. As before, not all of the gp is spent due to liquidity limits, and the gp is distributed evenly towards buying bars and diamonds up to the amount it wants. Why not buy equal qtys? - there is nothing tying the bars and diamonds together and this is the default spending split for items. Might add grouping later.

4) this time you've asked some questions I can answer concretely, but a lot of the behaviours are "sometimes" because indeed each account is different lol. It's not a cop out answer, the script is genuinely coded for each account to behave substantially differently at pretty much everything, including those GE +5%s.

Edited March 6 by Aeglen

[PPTouch2 1]

18 hours ago, Aeglen said:

1) this is not the actual price you get, when you put an order at +20% it can pay 0-20% more, almost always close to 0%. The % increase/decrease is less for expensive items.

2) this is due to spending limits, the bot likes keeping some liquid gp. Imagine if it skipped task/stopped and all your gp was in herbs. Herblore is an inherently gp-negative skill and the alternative of potion making is at least 3x more money draining. Overall, the bot earns more than enough for its bonds under default settings and reasonable levels.

3) very little (see 1), and surpluses will be used for future tasks. As before, not all of the gp is spent due to liquidity limits, and the gp is distributed evenly towards buying bars and diamonds up to the amount it wants. Why not buy equal qtys? - there is nothing tying the bars and diamonds together and this is the default spending split for items. Might add grouping later.

4) this time you've asked some questions I can answer concretely, but a lot of the behaviours are "sometimes" because indeed each account is different lol. It's not a cop out answer, the script is genuinely coded for each account to behave substantially differently at pretty much everything, including those GE +5%s.

1 & 2) The only herblore training (aside from quests) I've seen the bot do is cleaning herbs. Half of the herbs are actually profitable with small margins, which probably cancels out when putting in offers at +20% and selling at -20%. It seems there's no methodology and it's just "Hope for the best". I guess I'll have to live with that. It's not comfortable to watch lol.

3) That makes sense that it's evenly trying to balance the purchase. It would indeed make more sense to evenly balance the quantity. The task itself is what should ty the bars and diamonds together. That's literally what they're being purchased for lol. I get that they can be used for future tasks, but when money is low and you notice thousands of unnecessary items, AND THEN THE BOT KEEPS BUYING MORE, it gets annoying.

4) I get that in some cases this makes sense, but to me these are foundational problems. The bot isn't making any calculated purchase or sale unit pricing. It's just clicking a few times higher or lower. That's not sophisticated and is definitely losing money. The bot is also purchasing unnecessary items and not selling when needed, probably because they're deemed 'valuable' or 'necessary materials'. These are not account based chance issues, it's a design problem. At least it would seem so.

[Aeglen 297]

6 hours ago, PPTouch2 said:

1 & 2) The only herblore training (aside from quests) I've seen the bot do is cleaning herbs. Half of the herbs are actually profitable with small margins, which probably cancels out when putting in offers at +20% and selling at -20%. It seems there's no methodology and it's just "Hope for the best". I guess I'll have to live with that. It's not comfortable to watch lol.

3) That makes sense that it's evenly trying to balance the purchase. It would indeed make more sense to evenly balance the quantity. The task itself is what should ty the bars and diamonds together. That's literally what they're being purchased for lol. I get that they can be used for future tasks, but when money is low and you notice thousands of unnecessary items, AND THEN THE BOT KEEPS BUYING MORE, it gets annoying.

4) I get that in some cases this makes sense, but to me these are foundational problems. The bot isn't making any calculated purchase or sale unit pricing. It's just clicking a few times higher or lower. That's not sophisticated and is definitely losing money. The bot is also purchasing unnecessary items and not selling when needed, probably because they're deemed 'valuable' or 'necessary materials'. These are not account based chance issues, it's a design problem. At least it would seem so.

That's not correct, the + and - are needed to buy/sell in a timely manner. The bot pays a small premium (normally 0-3%, up to 20%) to get the item now, and that is the "method". An alternative is putting things at GE price and skipping half the tasks due to items not buying. A high-effort and worse alternative would be to buy in advance, tying up a bunch of gp and requiring stop handling. If you don't like this, you should know that the meta most bots use is to overpay +1000% if an item won't buy.

There is no functionality for grouping items right now. It works well enough as-is. It's as sophisticated as it needs to be, and there are dozens of more important features to add before item grouping.

Selling later-needed items would incur GE penalties; the bot prefers to keep liquidity in GP and typically does not suffer from money shortages assuming you gave it a reasonable skill balance.

With the exception of grouping gold-gems, I have not heard such well thought out but silly critiques in my entire time developing bots. I would strongly advise that you get the necessary domain knowledge, either by creating your own bot, or using others, before declaring that the bot's non-trivial systems are "badly designed", "not sophisticated", "incompetent", "foundational issues" etc.

Edited March 7 by Aeglen

[SDN Bot 167]

P2P Master AI has been updated and is now live on the SDN!

It's currently at v1.132 and is available in the client!

Changes:

• ADDED MLM BETA
• Fixed all reported Slayer quirks
• Added Ferox and MLM banks
• Improved banking optimisation
• Misc fixes

Time since request was made: 1 hour, 1 minute, 19 seconds

Thanks!

[jared462 0]

Hey Agelan. Been training some combat specifically training with gargoyles. For the most part it works great, my only issue is that when banking it will sometimes make multiple trips not banking the adamant platelegs and adamant boots (obviously this takes away from how much food i can bring with me). Also it will equip 1 rune full helm . Bank my Obsidian helm and the rest of the full helms. 

[Aeglen 297]

Just now, jared462 said:

Hey Agelan. Been training some combat specifically training with gargoyles. For the most part it works great, my only issue is that when banking it will sometimes make multiple trips not banking the adamant platelegs and adamant boots (obviously this takes away from how much food i can bring with me). Also it will equip 1 rune full helm . Bank my Obsidian helm and the rest of the full helms. 

Ideally, join the discord below and send the logs to the bugs chat so if there is a bug I can fix it. Thanks!

[PPTouch2 1]

1 hour ago, Aeglen said:

That's not correct, the + and - are needed to buy/sell in a timely manner. The bot pays a small premium (normally 0-3%, up to 20%) to get the item now, and that is the "method". An alternative is putting things at GE price and skipping half the tasks due to items not buying. A high-effort and worse alternative would be to buy in advance, tying up a bunch of gp and requiring stop handling. If you don't like this, you should know that the meta most bots use is to overpay +1000% if an item won't buy.

There is no functionality for grouping items right now. It works well enough as-is. It's as sophisticated as it needs to be, and there are dozens of more important features to add before item grouping.

Selling later-needed items would incur GE penalties; the bot prefers to keep liquidity in GP and typically does not suffer from money shortages assuming you gave it a reasonable skill balance.

With the exception of grouping gold-gems, I have not heard such well thought out but silly critiques in my entire time developing bots. I would strongly advise that you get the necessary domain knowledge, either by creating your own bot, or using others, before declaring that the bot's non-trivial systems are "badly designed", "not sophisticated", "incompetent", "foundational issues" etc.

Clearly I've upset you and that's not my intentions. I started developing OSRS bots around 15 years ago and released several publicly. Now I actively work in C#, C++, Python, Java, Powershell, .NET, and even do some Web Design if it's needed of me in my occupation. I would strongly advise you don't make assumptions of others' experience to validate your opinion. 

With that being said, your overall bot is very sophisticated. I've expressed several times that I like it. That, however, does not mean that everything about it is sophisticated. There are multiple APIs for gathering current and historical G.E. pricing, and the script seems to not consider any of that. It instead randomly selects an arbitrary number, probably from 1 to 5, for amounts of times to click either the up or down arrow for changing the price. It doesn't matter how much worse other bots are Aeglen. That is unsophisticated. 

I've literally worked with a developer for a G.E. flipper bot that uses the same exact APIs. I am actively working with the developer of a new GE tracking website. This stuff is well within the realm of possibility. If you have other priorities, or it's too complex, or you never thought about doing that before, or it's not worth the time, or whatever, that's fine, but there are other routes. You can make calculated, sophisticated, item trades. That's all I was saying.

As for the other point about item grouping, that's fine if you have other priorities. It's understandable. At least those words lend some credence to the validity of my point. Sometimes a little bit of acknowledgement goes a long way.

[Aeglen 297]

10 hours ago, PPTouch2 said:

Clearly I've upset you and that's not my intentions. I started developing OSRS bots around 15 years ago and released several publicly. Now I actively work in C#, C++, Python, Java, Powershell, .NET, and even do some Web Design if it's needed of me in my occupation. I would strongly advise you don't make assumptions of others' experience to validate your opinion. 

With that being said, your overall bot is very sophisticated. I've expressed several times that I like it. That, however, does not mean that everything about it is sophisticated. There are multiple APIs for gathering current and historical G.E. pricing, and the script seems to not consider any of that. It instead randomly selects an arbitrary number, probably from 1 to 5, for amounts of times to click either the up or down arrow for changing the price. It doesn't matter how much worse other bots are Aeglen. That is unsophisticated. 

I've literally worked with a developer for a G.E. flipper bot that uses the same exact APIs. I am actively working with the developer of a new GE tracking website. This stuff is well within the realm of possibility. If you have other priorities, or it's too complex, or you never thought about doing that before, or it's not worth the time, or whatever, that's fine, but there are other routes. You can make calculated, sophisticated, item trades. That's all I was saying.

As for the other point about item grouping, that's fine if you have other priorities. It's understandable. At least those words lend some credence to the validity of my point. Sometimes a little bit of acknowledgement goes a long way.

It is rude to use all the quoted language to describe a process you appear to be unfamiliar with, which is surprising given how long you've worked with GE things, so expect grumpy replies on that. Perhaps you've not used it much by hand in-game. It's certainly unrelated to how long or what you've coded in, but it's somewhat equivalent to a "market order" if you are familiar with trading. That said, I could tell you were experienced at - something - (if not GE workings, nor politeness) due to the quite rare level of competence in your critiques.

Whilst there may be APIs with pricing data, they are not useful for acquiring items in a timely manner given specific % budget constraints. The only way to do this is to buy them in advance or place a "market order", the latter of which will usually pay the same price whether it's +10% or +1000%.  Knowing the price does not change this strategy in the latter case, nor the former unless you are strongly erring on the safe side as with flipping. Speaking of assumptions, the bot most certainly does not randomise its 5%s with a random number generator from 0-5. As with everything it does, it pulls the number of times it presses + or - from a distribution unique to your account, typically with 2 or 3 possibilities for cheap items.

Edited March 7 by Aeglen

[PPTouch2 1]

 

4 hours ago, Aeglen said:

It is rude to use all the quoted language to describe a process you appear to be unfamiliar with, which is surprising given how long you've worked with GE things, so expect grumpy replies on that. Perhaps you've not used it much by hand in-game. It's certainly unrelated to how long or what you've coded in, but it's somewhat equivalent to a "market order" if you are familiar with trading. That said, I could tell you were experienced at - something - (if not GE workings, nor politeness) due to the quite rare level of competence in your critiques.

Whilst there may be APIs with pricing data, they are not useful for acquiring items in a timely manner given specific % budget constraints. The only way to do this is to buy them in advance or place a "market order", the latter of which will usually pay the same price whether it's +10% or +1000%.  Knowing the price does not change this strategy in the latter case, nor the former unless you are strongly erring on the safe side as with flipping. Speaking of assumptions, the bot most certainly does not randomise its 5%s with a random number generator from 0-5. As with everything it does, it pulls the number of times it presses + or - from a distribution unique to your account, typically with 2 or 3 possibilities for cheap items.

Saying I'm not being polite while again insulting my competence and questioning my experience. I've done nothing but suggest that there's ways to improve your script. Disappointing.

You think that pricing data can't be used well, or in a timely manner, and that's all I really needed to know. Accuse me of being unfamiliar all you want, when it you who does not understand how to make this work for you. It can and is being used well in scripts to do flips non stop on the G.E. for a profit. There's no way to make that efficient and profitable unless it was using the latest data for the immediately best priced trades. There's not many good bots that do it, because they're not familiar with markets or how well the APIs can be used, but it's very possible. I would make some suggestions for some you can look into, but honestly I don't think you give a shit at this point. The function works 'well enough' and that's that for you. Forgive me for being so daring as to suggest that there's room for improvements.

Edited March 7 by PPTouch2

[SDN Bot 167]

P2P Master AI has been updated and is now live on the SDN!

It's currently at v1.133 and is available in the client!

Changes:

• Fixed issue with new accounts
• Fixed all known MLM bugs

Time since request was made: 25 minutes, 2 seconds

Thanks!