My aim for AIM

AIM. The Ying to the AMM Yang. The Automated Investment Maker. The infamous capital vehicle to which you need 13 TRND as your ignition key, and a lot of xTRND as your fuel. Many asked what it will do and how it will work. It is still a deep work in progress, however I want to collect some of the things I said here and there into this single post.

One of the important features of AIM will be capital protection. Imagine having a tight stop loss on your every position. During market downturns AIM holdings would have been converted to DAI or USDC at minimal loss. Imagine the investments possibilities when the market dumps -50% but you get out at -3%. Of course, in this scenario the question always remains “Where’s the bottom?” The answer still is a tight stop loss exiting into stablecoins.

Another important aspect of AIM is its design against front running. The most obvious understanding of AIM is a trading bot. But AIM is much more than that, and in fact I would not call it a bot at all. Front running is really a big issue, however the impact is worst when one operates a trading bot that executes positions frequently. I believe AIM can mitigate this by entering less position, but of higher quality. In essence, I want AIM to act more naturally. There is an AI in AIM for a reason 😉

The stealth trial run of AIM predicts 60%-100% weekly returns. Indeed, I have been running this experiment for 10 weeks now, with four to five digit capital. I am at about 1100% overall profit over that timeframe. Looking at the crazy numbers many tokens presented in the last months, I think that is not that good. A very nice stress test happened this week when Sushi imploded along the stock markets and forex. The AIM acted pretty good, exiting into stablecoins at a minimal loss. It is interesting to mention that one of the first exists happened much earlier than I expected, with ETH still at $430. Well it turns out the AIM was right.

Finally, one thing I want to stress is that AIM will only partially focus on trades. What is even more interesting is exploiting opportunities on the blockchain. There is a lot going on in the DeFi space, and I won’t surprise anyone by saying that YFI is the leader that is easy to look up to. As always, expect more things to come.

About me

My name is C.
(telegram) (twitter) (email)

I am a fan of automation, data analysis and Ethereum. I have been working on Trendering for some time. I have seen many things: coin scams as they happened live, pump-dump schemes from coordinated addresses, whales moving markets…

I will never reveal my identity. If you have a problem with that then well… it’s not my problem. I give you Trendering to level the odds. It’s your choice whether to use it.

4 thoughts on “My aim for AIM

  1. ApopheniaPays

    I’m interested to follow this, but I have to admit to a healthy skepticism.

    Forgive the length of this, but this is a topic that is near and dear to my heard.

    I’m a trader and coder of many decades and spent many years of work trying to find a reliably profitable trading algorithm, and never succeeded. I was never able to find an “edge” that didn’t turn out to be as big or bigger a cost when circumstances different. Yeah, you can cut your losses quickly; and then whipsaw back and forth racking up trading fees as you get in and out when the price suddenly recovers after your stop.

    There are definitely more reliable ways of picking exits than just a price-only stop (I never stop out without considering volume myself and developed a complex software algorithm that finds “real” price drops, ones that don’t recover quickly, way better than I ever could by hand); but the cost of more certainty is they often introduce more slippage, and, on the odd fluke occasion that they mess up badly, wind up costing you _much_ more in losses.

    In short, there’s _always_ a trade off. The only truly profitable strategies I’ve ever found were high-frequency strategies, which meant they worked very well in theory, until you introduced even a very small transaction fee, after which the fees always very quickly added up to much more than the profits. And with gas fees often proportionally much higher for trading on AMMs than transaction fees are on CEXs, that effect is bound to be much worse.

    Personally, years of my own research ended without ever finding evidence that consistent algorithmic trading profits are even possible, because of the nature of trading: the extent to which a move benefits you if the markets the way you expect they will is the same extent to which that move will hurt you if the market subsequently goes against you. Any move that give you an “edge” will damage you just as much. And then the constant friction of transactions costs are the “house odds” that ultimately tip the game against any system.

    None of this is theory: this is my observations based on years of developing real software trading systems, watching how they work in realtime, watching why they fail to profit in the end, improving the algorithms to compensate for those failings, and starting over again with a new round o .

    Backwards-looking strategies, like “The market has fallen 3%, get out” don’t work because it is solely the market’s subsequent moves, not previous moves, that determine the profitability of what you do now.

    I’m not trying to FUD. In fact I’m confident to see if you actually have found a solution that I never was able to. It’s possible.

    But as I’m fond of observing… often, formulating winning strategy for the game of Go is considered the foremost computing challenge known to man. I believe- I don’t recall the figures exactly- that it took an AI 9 hours to teach itself chess well enough to beat a chess grandmaster. It took AlphaGo 14 days to teach itself Go well enough to beat the human champion. That should give you an idea of how hard Go strategy is. (And if you’ve never played, I encourage you to download a phone app and play against the computer a couple of times. Go strategy is much harder than chess, and a lot of fun to try to wrap your head around once you get into it.)

    But AlphaGo did eventually beat the human Go grandmaster, and it was probably the foremost feat of computer science in history.

    But, to my knowledge, no AI has yet found a way to be consistently successful at the stock market. Which means one of two things: a rigorous system of profitable trading is harder than beating a Go grandmaster; or, a rigorous system of profitable trading is not possible.

    Again, I still retain some small hope that I’m completely wrong, and that consistent algorithmic trading success is possible. I’m open to the possibility that you’ve really succeeded at it, but if some, I’m desperately curious at how you got around the natural obstacles, some of which I’ve just mentioned, but which I’ve got cataloged in greater detail than given in this late-night top-of-my-head rant here, and would love the chance to discuss anytime.

    If you ever feel like chatting about it further with someone who’s explored the same space, feel free to hit me up on twitter at the below username.

    Otherwise, I greatly look forward to seeing what you’ve created with the AIM. Thanks for everything you’ve done with TRND so far.


    1. C Post author

      I really like such long and deep comments, so thank you! I understand the scepticism, and I know where it is coming from. What I am describing is kind of a holy grail in the context of the traditional financial market. However, with crypto the game is a bit different. But either way, I have my share of scepticism as well. I am looking at the trial run and am still working on making it really scale and ensure it will continue to work like this in the really long run. Please share all your doubts now and in the future. Having different view really helps to look at the problem at a different angle.

  2. E

    This is very promising and thanks for all your hard world “C”. I’m coming from the biotechnology industry and must say I am intrigued. I was wondering if you are using knowledge graphs for the AIM as an intelligent interface that performs logical inference of data types, relationships, attributes and complex patterns within the marketplace too difficult for humans to spot, and in future pair this with natural language processing for social media monitoring. I am watching this space closely.

    1. C Post author

      To be honest I am not using the knowledge graphs per se. I did a lot of social media processing back in the days, semantic analysis, NLP and sentiment detection. My conclusion back then was that it is highly inaccurate and very easy to manipulate. Looking at CT (crypto twitter) today I think it only got worse. I do not think AIM should go into the direction of trying to spot the next gem and go all in. I think it must be based on statistical analysis of chances of winning, risk management, and protection of capital. The sources of information should rely more on the blockchain data and less on sentiment. I think deep learning has a big potential here, because if you look at the blockchain transactions graph, it is very similar to a game of Go mentioned by Apophenia earlier. Certain transactions have a big impact on the system, and certain movements have predictable outcomes.


Leave a Reply

Your email address will not be published.