SqueezeTheorem

Almost always a perm ban following a 2 day ban. Like the other guy said, don't risk it if it's a main account. Whenever my farming accounts get a 2 day ban, I just suicide bot on them after that until they're permed.

Absolutely no reason to buy/rent your own server unless you 100% know what your doing. OSRS is a very lightweight game, and the low CPU mode on OSB makes it even more so. If I wanted to, I could run like 30 accounts at once on my PC. Experiment with the hardware you already have before investing in more.

You're right, they've definitely changed something. I first got smacked by the updated system on on March 3rd or 4th and it's only been getting harder since then. I think gold farmers are feeling the pressure from all this. Look at the price of bonds, they went from 5M to 4.4M and they've been dropping in price since March 11th/12th. I assume the price drop is a combination of gold farmers having increased ban rates and more players from the steam update buying bonds. Retail gold sellers are offering more for gold and selling it at a higher price too. My last ditch effort at botting is going to be extremely long (4+ hours) macro recordings to see if that helps at all. Fingers crossed.

March 20th, 2021: RIP Account #1, you nearly had 99 crafting and you had 99 fletching. (~8M loss) 4 accounts are left, 2 of them have prior temp bans so I'm muling them frequently. I've started a 6th account that I'm investing a lot more manual playtime into than usual. The goal is to make it appear more like a main account and have it do a variety of profitable activities (rather than just one) to see if that helps. Total cash stack is now 30M + assets on remaining accounts. Would be higher but I invested some in my 6th account. I've also recently discovered how to scale up to ~30 concurrent accounts by using an FPS limiter.

With agility, the best thing to do is make a recording on a lower-level course so your fail rate is relatively low. Add in some kind of a kill switch that stops your recording when you've fallen off and you should be good to go. Gnome agility has a 0% fail rate and I've been tempted to try botting it for a while despite the garbage XP rates. No one goes to that course really so the odds of being found are quite low.

If you're botting something extremely common like green dragons, expect to be banned almost immediately. There's probably hundreds of thousands of accounts that have been banned for killing green dragons that Jagex has been able to record data on. If you want a lower banrate, pick something with lower GP/hr but that's also done by actual humans like cooking, smithing, fletching, etc. It's a lot less of a headache. Using my own custom scripts and macro tools, I've had accounts last hundreds of hours doing more human-like training methods botting 12 hours per day. Right now, I have an account with 35M cooking XP that hasn't been banned yet.

My first bond cycle complete, looking like it's gonna be around 18-19M profit. This is in spite of the 2 day temp bans on 2 accounts. 3 accounts were used. I have 2 days membership left on one account, 3 on another, and 4 on the third. They should be able to make up the cash to pay for their next bonds during that window. Account number 4 was recently created and will be rolling out tomorrow hopefully. I have the proxies needed to run 8 accounts concurrently. I can scale up to 16 accounts total using 12 hour shifts if I make some needed adjustments to my tooling. With 8 accounts, I'm on pace to make like 300M per month. I can dramatically increase that amount to around 1B, but I need GP to invest back into the farm to unlock better money making methods, and I also need to develop the tooling for those better methods.

New year new gold farm!

It already is, but it needs a couple of stability improvements. I'd post the link here but I'm not sure if that's allowed per the OSbot rules.

Introduction Sporadically over the past year or two, I've been experimenting with unorthodox botting methods hoping to figure out how BotWatch works. It's a complete black box. We have no insight into precisely how it works, and it likely changes over time, but some of the experiments I've done lead me to believe I've figured out a significant component of it. For me, figuring it out is the ultimate engineering challenge. Initial Testing - Multiboxing (~1 year ago) I began my research by making a very primitive multiboxing script using another botting client that offers finer control over mouse movement. For those who aren't familiar with the term (since it's far more common in other MMOs), multiboxing software allows you to have your actions mirrored across many accounts. If you click at a certain coordinate on one OSRS client, all of the other clients click at the exact same coordinate. If you move your mouse on one client, the mouse moves in the exact same way for all the others. I used this script to control 8 accounts at once in free-to-play making golden jewelry and get enough money for a bond. None of the accounts were banned for multiboxing despite the fact that I was using an injection client to control them. The total play time was around 6 or 7 hours, which for free to play seemed fairly good to me. I was confident enough in this initial anecdotal evidence to try more experimenting with control over the mouse as a means of evading bans. Phase 2 - Macro Recording (~1 year ago) Next, I cobbled together a very basic macro recorder. It could track mouse movements and play them back on a loop. In designing the macro recorder, I made sure that the mouse playback was as precise as possible. Inputs are played back within a few milliseconds of the correct timing, which is about as good as it's going to get. Using one of the accounts I had multiboxed on, I cautiously made an hour-long recording of me enchanting sapphire jewellery at castle wars. It was slow profit and xp, but it allowed the account to increase its playtime using a very non-intensive training method. Within a day or two, the account was at ~50 magic. Given my past experience with botting and getting banned extremely easily on newer accounts, I felt confident enough to try botting way more hours per day. I got the account to 55 fletching manually. From there, I made an hour-long recording of me stringing bows at castle wars. I then played this recording back for 6 hours a day, and then eventually upped it to 12 and finally 18 hours per day. I was able to bot this much on a daily basis until the bond on the account ran out and finally had to renew it. I knew I was on to something, so I decided to expand beyond just one account to see how other accounts did. Phase 3 - Expansion (~ 1 year ago) I took more of the accounts I had previously multiboxed on and bonded them up. I gave membership to 4 or 5 total, and tried to suicide bot using my macro recorder. Curiously, all but one of the fresh accounts were banned within the first day or two of them starting to bot while the original account continued to bot heavily with no issues. What was even more bizarre is that once an account hit a certain age (~24 hours playtime), I was able to bot continuously for ~18 hours per day without any problems. It seemed to be that there was an initial "screening" period where Jagex more heavily monitored new accounts. Either more computing power was dedicated to analyzing their behavior during this window, or they were simply viewed more suspiciously by the system due to their age. After an account survived this initial screening period, it became seemingly unbannable using my macro recorder. In total, 2 out of 6 accounts survived this screening period and I was free to bot on them basically as much as I wanted without consequence. Phase 4 - Bot Busting: Moderate (Recent) More recently, I got back into botting and decided to REALLY push the limits of my macro recorder. I bonded up 2 of the original accounts I had botted on previously using the cash I had saved on them and ran them for 20 - 22 hours per day. The recording that I used had a total runtime of 2 hours and 15 minutes. Keep in mind that right now, Jagex is going all in on stopping bots because of the Steam launch and ban rates have skyrocketed. After running them for 4 days straight, I finally got hit with a 2 day ban on each of them. On one of the accounts, I was able to get to 99 fletching by stringing bows with a 1 hour recording and 91 crafting by crafting golden bracelets (30k xp/hr) using the 2 hour and 15 minute recording. This means, in total, I was able to run the crafting recording for like 200 hours before a temp ban. I'm not sure how long I ran the fletching recording for, but I suspect it was around 100 - 150 hours because I was making maple longbows instead of higher EXP bows back when they were the highest profit. Conclusions & Future Experimentation While Jagex almost certainly uses a large variety of methods to detect and ban bots, mouse data seems to play a very significant role. The fact that mobile bots initially had such a low ban rate seems to support this. If you are able to incorporate human mouse data into your bot, that mouse data isn't publicly available, and you bot 12 hours per day or less, you should be able to dramatically decrease your ban rates. Going forward, I would like to collect more mouse data and make an even longer recording by combining a series of smaller ones. The ideal length is 4 - 6 hours worth of data. With recordings that long, moderate botting should be virtually undetectable. I would also like to experiment with creating new recordings every ~100 hours of botting time to see if this helps. I attribute the success of this method to the fact that over small periods of time, it's very difficult to distinguish a recording of human behavior from actual human behavior. I was able to take around 3 hours of recordings and get 300 hours out of it before a ban. Not bad imo.

I have a method of botting where I used to be able to bot 18+ hours per day, and just last night 2 of my accounts got hit with a temp ban. First time it's happened with this method. My guess is they've changed the sensitivity of their bot detection parameters in preparation for the steam launch and are issuing temp bans just in case they accidentally ban real players. Ban rates should hopefully be back to normal soon.

Have you guys noticed any changes in banrates since Weath left? Last time I tried botting Weath was still around, but I've been suiciding 18 hours per day on 2 accounts for the past 3 days. Am I lucky or has it gotten easier since he left?

Using nothing but my own homebrew macro recorder, I was able to get 99 fletching botting 6 - 12 hours per day. The recording was 55 minutes long, and I still have the account a year later. Got nearly 90 crafting using the same approach.

The owner of a competing bot website has shown that Jagex collects data about garbage collection from the Java Virtual Machine. Most modern clients spoof this data to avoid being detected. This is one of the reasons why mirror mode is preferred for lower ban rates -- the garbage collection data looks completely normal in mirror mode and not like a bot client. We also know from this thread here that Jagex collects mouse movement data from the client. How exactly this data is analyzed we don't know. We only know that it's collected. I'm in a Discord where some people have been experimenting with privately collected human mouse data to a relatively high degree of success. If you can provide human-generated data and avoid directly modifying the game client (like with injection mode) you're off to a decent start. You can still be banned, but combining these two approaches has worked well for me so far.

June 14th, 2020: I gave up on f2p gold farming because I found a way to avoid bans in p2p. After some pretty extensive research and making some custom tools, I'm finally at the point where I can run two accounts for 20+ hours per day without getting banned. This is accomplished using some highly unorthodox methods and tooling. The first account has been running 6 - 20 hours per day with no issues, and the second was just deployed. They're both p2p accounts. They don't make much, but it's consistent and they pay for their bonds within 24 hours. My current bot farm won't scale well, though. The resource they're botting has a relatively small daily volume on the GE. As little as 10 accounts would start to undercut my profit. Going forward, I'm going to prioritize antiban above almost everything else. The consistency of low banrate methods is better than higher GP/hour, higher banrate methods, at least to me.