Hummingbot Community, 3 Jul 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Hackathon and bounties | Hummingbot Live

Description

Know the latest updates about hackathon and bounties.

0:00 Welcome
1:30 Start of Presentation
5:10 Community Population and Contributions

Reference Links:
Hummingbot Website - https://bit.ly/3qHoEtb
Hummingbot Miner - https://bit.ly/3FiZpS6
Hummingbot Blog - https://bit.ly/3qI2EhO
Hummingbot Docs - https://bit.ly/327EnYo

**Important Disclaimer: **
1. The content of this video does not constitute investment, financial, legal, or tax advice.

2. No guarantee of profit: CoinAlpha/Hummingbot does not claim that liquidity mining and participation in liquidity mining campaigns and crypto trading will be profitable, however measured, for the user.

A

And really try to show off what the community's done. So, um as you guys can see uh we're we've uh kind of improved our um setup so actually oops. I actually forgot to do the graphics, sorry one second, um I'm supposed to play this at the beginning of every return. So, let's start.

A

Again: okay, now that's an introduction. So uh once again welcome to hummingbot live our weekly show where we run bots and talk about all the contributions the tamiya community has made uh so this week um uh we want to have. We have something pretty special, which is uh we're going to go over um uh like some of the some bouncing hackathons that we've done uh one. Second, let me just check to see uh that I'm actually streaming here.

A

I think I am, but um uh if um let me just ask my team to, let me know if not.

A

Okay, all right, so I believe I'm live right now. uh It does look like there's two people watching. um I see some. Oh yeah. I see some people in the chat great so um so today I want to go over a little presentation. I made about um the some of the previous bounties we've done as well as looking for some future ones, and then um we just finished our first uh kind of bounty. We opened d5 and um and uh and bitcoin. So I wanted to show off the the winning contribution to my bounty.

A

So uh why don't we get started? Okay, one sec. Let me show you my screen.

A

A

All right um so bounties and hackathons at hummingbots. So first of all, um we're gonna be writing some bots today uh and um important disclaimer. It's really only for educational purposes. Only so um please, uh please uh yeah just consult your lawyers and financial advisors, you're actually going to run the bots and everything is really for developers purposes.

A

um So hummingbot has really kind of grown into a pretty large open source project. At this point you know we started a couple years ago. We only had connectors to finance and and uh one zero x exchange, but over the past couple years we've started to add.

A

um You know connectors to both centralized anti-central exchanges, and today um we feel like we're in a good position to kind of really serve the industry, uh because you know there's there's so many decentralized exchanges coming online uh and there continues to be more new products in crypto options, features um swaps. You know all kinds of crazy new financial derivatives on the blockchain and each of these new products has a different equity profile and so um and the trading is different.

A

um How settlement occurs is different and so the um yeah it's uh it's something we have to try and try to do so so because of that, I think um sorry, sorry once I it looks like my video is not very great here. Let me try to improve my videos now: okay, let's try this okay, I'm gonna bump up the video resolution on this thing and uh and see if we can do.

A

Okay, um let's see, I still don't think the screen is very high resolution. One second yeah.

A

Okay, uh well, first time, first time trying a new platform. So there's some some um some hiccups, but uh please you know please put in the chat. If you guys can't see the screen. um My face is not important, but uh I'm gonna show you some thoughts and everything. So uh it will be important for people to see what's happening on my screen.

A

So um so, let's see um okay, um so so yeah, so the overall. um If we bought we. um Let me show you my screen again.

A

Okay yeah. The resolution here doesn't seem to be great. It's funny. um Let's see.

A

All right um can't do much right now, we'll try to fix it next time, but uh apologies for the lack of resolution on the on the screen.

A

um This is gonna, make it harder to see what's happening when I run the bot, but uh I'm not sure, quite as traffic, so let's just keep going um so so yeah, but I think all the characters we've built is really uh really comes from the community and um today hummingbird is both the um you know: the open source project where we've had 15 000 people in our discord, community running and building bots, as well as a very decentralized set of people who are contributing into the project.

A

uh This includes our team, which is pretty spread out all over the world as well as a lot of the uh you know, um contributors in the community, so we've had um you know. At least half of these contributors have been outside of our organization, from exchanges uh from blockchain projects and from all of you guys out there um and then, in addition, with minor we've really seen kind of, like you know, the growth of a decentralized market-making army um we've had over 20 uh projects come to us for liquidity and um our our users.

A

Our miners have filled almost two billion dollars of orders since deception. So um I think a lot of the successes we've had as an organization is really uh from the community, and so that's why you know now that we're getting bigger and we've started to serve and build connectors for various projects. um So in the past we've built, uh you know kind of like connectors for xerox, balancer, uh terra cello, professional protocol and kucoin.

A

um You know going forward now we're seeing an incredible level of demand from both decentralized, essential exchanges for both connectors, as well as uh strategies that are customized to the idiosyncrasies of every protocol in every exchange.

A

So um and so, for instance, uh you know, there's um work to upgrade zeroth connectors, zeros matcha uh cello um has introduced a new amm on their exchange uh called uberswap that we planted the support support um tara.

A

uh You know we have a terror connector to their the core blockchain, but in addition, there's uh other other dexes like mirror ontario that will support and finally, um with a number of new partners uh like finance, xerox, um al grant, avalanche and others um who are uh you know who are partners with hummingbird and want to sponsor bounties um to to to basically have uh their custom connectors and their strategies.

A

So, uh to be perfectly honest, I think that while we could, our engineering team could serve maybe a small subset of one or two of these guys. uh It's it'll be impossible for us to serve everyone, and so that's why we've kind of made a choice internally to really focus on the architecture of homebot.

A

um You know we're still gonna fix bugs you know kind of like do things like generalize certain pieces of architecture, but we wanted to really encourage and and allow um the community of users out there developers to really really um basically, basically basically earn these bounties by building and maintaining um these different things. So that's why the first thing the first experiment we started with- um was the open, d5 hackathon. uh We started small because we know that something like codebase it.

A

What is um is still very much a project, that's written by us um and we have started to add documentation that shows people how to write their own connectors and their strategies, but it's pretty decent. So we wanted to start small and just see if our community uh would be able to write strategies. um You know basically, so basically, we work with bitcoin um during their open defined hackathon to launch a two thousand dollar bounty and we ask people to develop a hung out strategy.

A

So um so so honestly, we were blown away by the reception um you know like given a very uh given how small the bounty was. We really didn't expect people to to come up with anything. uh You know yeah incredible, um you know we thought that maybe they would create a strategy that you know. Kind of, like just you know sells, sells, um sells a block of uh you know of trades or it does something like scans the order book or nothing.

A

This isn't very simple, but what we end up with was actually this um this, so the winning the winner submission out of, I think we had three or four people who actually submitted strategies. Was this um a rune oscillator strategy that uh that um one of one of our users submitted so um I'm gonna post the github link in the chat right here, because I know it's not very, it's probably hard to see um so so I I think that um yeah, hopefully you guys can kind of like take a look at this strategy.

A

I know that my video is extremely. uh You know low resolution right now for some reason. um That is very hard to see. Let me try to fix this, because uh I don't there's no way that this is. You know the actual resolution, I'm scaring you so one second to watch how to fix the.

A

A

All right, let's try this.

A

Okay, uh that seems to be better. um Okay, I think that's that seems to be better. uh I can actually read the words on the screen now um still seems pretty blurry for some reason, yeah. I really wish I knew was going on. Actually it looks like yeah, okay, um okay, well, hopefully yeah. Let me move it here.

A

Is that big difference, but does that make a difference? Okay, all right! Well, hopefully, that's better, but uh we'll we'll see how it goes.

A

But, um as I was saying, uh we were really impressed by this submission because um the user not just created, you know it's very simple strategy: um the user added a indicator called the arun oscillator um and- um and he also kind of uh wrote, this very, very detailed description of how that indicator works and how um how this strategy uses the indicator to adjust spreads um as the market trends go either up or down.

A

So um I encourage you to read this in detail, uh because um this this contributor spent a lot of time actually both describing it, but also um you know, even when you run it, I'm not sure you can see this, but um the status command actually shows both the indicators uh and also the calculations of how they're how they're being used to adjust the spreads. So you can actually walk through the math and understand for yourself intuitively. um What's what the strategy is doing so um all right.

A

So, let's let's go ahead and um and um and start the strategy and see how it works. Our life.

A

So um I actually have the the strategy running over here, so I'm gonna, I'm gonna. Let's, uh let's see what it's doing um so, as you can see, um it's running live over here um and um it's it's kind of like you know it right now uh the the indicator up indicator value is 617 and the down enter value is 100.

A

uh So I believe that yep, it's it. um It's saying that the market is at a downturn and that's kind of why uh you know it's kind of like adjusting the spreads um based on based on that, so um so yeah. So, let's, uh let's go ahead and uh and actually kind of like start the box again uh and then I'll show you guys very gekko how how how this um strategy works. So one thing I've done here um as an fyi.

A

um I have pulled this um this person's branch, so uh this pull request um is is currently in our github. The reason it's open is because we're we're currently in the middle of creating a generalized indicator class that any strategy can import and so the way that uh that this developer uh he basically modified the peer marketing strategy to create a new strategy.

A

So what we plan to do is to um instead of merging this, as is we plan to um kind of like create any other class and add this room indicator to that class, so that uh any strategy can do this, uh but but overall, I think indicators are something that is sorely needed in humminbot and um and so now now that we're starting to kind of like collect um some data and memory, uh you know basically mostly price and volume data we can expose that to um this entertainer class and then and then dynamic class would allow uh developers to import that and use that to adjust their strategy in various ways, so um we'll be announcing more news of that later on.

A

But uh for now um I think this a room indicator kind of strategy. Also strategy could be an interesting one for users to try it. So um so what I've done here is is uh basically I've pulled. The um I've pulled this person's branch, the fork of heimbach, uh where he wrote the code and and then I've um I've also been compiled.

A

um I've installed and compiled as well in order to in order to basically build the custom source now afterwards, uh and then you can just run coming by um by writing command bin, slash dot py, which is how you would start the bot if you're, if you've installed the source uh and then, after that, it's the same as it would be if you um started either with the container or from the binary installer. uh So what I've done here already is I've um a test account from finance uh that we're using.

A

So, um I believe, there's like uh some funds right now and then what we're gonna do is we're gonna start a bot and um and run it on um using this run oscillator strategy that uh the hackathon would have created so to do that, um it's always a start command uh in hyundai, and uh so um the question for the strategy is so now.

A

uh You can pick from a number two number of different strategies that we have available and, um and the roon underscore oscillator is the one that this user is at, so for the choice of exchange um I'll pick finance. um If you guys uh have any suggestions for trading pairs, you wanted me to run. uh Please let me know, um I think, uh let's see um why don't we, you know why don't we do something? You know um something different here: let's, uh let's try something like doge.

A

uh We also have yeah- maybe maybe maybe some doge oit. Let's do that so to try to get things a little interesting. So um so the two spread parameters in this in this uh in this strategy are the minimum and maximum spread, um and so essentially what what this strategy will do is based on the market trend. It will um it'll kind of like oscillate the uh the spread between the middle of the maximum spread.

A

So let's do 0.1 for the minimum spread and then 0.9 for the maximum spread, so that basically, the bot's gonna kind of, like you know, on average, should be around 0.5 spread in this market.

A

um So uh the next one is the number of periods you can calculate the indicator, so the small number of periods, the more volatile this oscillator is indicator, is uh the more larger one is, is larger. So, um let's try something small, because uh we want to try to see some activity running the bottom, so I'll take 10 periods and then, instead of one minute one minute period, I'll pick something very fast like 10 seconds and then so.

A

Let's cancel orders every 15 seconds as usual um and then you know I'll add some doge order and then save the configuration file. um I'm also going to just make some changes here to make it a little bit um a little more exciting. It's like removing the field order, delay and then I'm also going to.

A

um So uh yeah this is good. So let's start the box here and uh and see. What's uh what's gonna happen when we run this on the dodge, doji usb player.

A

So here's the doji pair uh looks like it's been pretty sideways for the past uh a few few few hours um but uh yeah so we'll see, let's see so. Let's start the bar and uh we'll see what happens.

A

Okay, so, let's see so created created, spreads on here on, you can see the the ask and the uh the bit spread here, uh it's creative again, uh and so, if we want status in the box, we can see that uh right now, it's starting to calculate kind of like these aruna up and down um indicators. So there's three indicators calculating the up uh the down and um this oscillator.

A

So as it's calculating, um let's, let's take a look at what what this oscillator root oscillator strategy actually does. So um the developer was kind of to put a link in this um in this github issue, and- and so there's an article here in investopedia that talks about how this works and essentially what it's doing is, is kind of creating two last two lines on the chart.

A

um You know an up line that that um that looks at the recent, like you know, uptrend and also another down in here, which is this is like this down um this down. So so, um overall, it's a very simple indicator, and, um and so typically um what what what you want to do is when the down indicator is very high.

A

uh You know it indicates a bearish trend and when the up indicator is very high, initiates a- and so you probably want to you- know kind of like take that account as you're um as you're using strategy all right. So so right now um uh the bot's going to wait a total of 10 periods. Now when each period is 10 seconds each and then after that, it will have enough data to then um kind of use this this indicator to adjust the spreads. So let's see how many periods it has done so far, 10 periods.

A

So um let's see it was that um yeah. So so then yeah it is. It actually should be adjusting spread. So you can see that um right now, because um the the up is this, it thinks there's an uptrend in the market, um it's kind of like it's. It's basically adjusted the spreads. uh You know so that the the the bid is higher and the up is lower, um so uh yeah.

A

So basically um you can see how it's adjusting here, uh but essentially it's kind of like using it's it's basically using the uptrend and the crosstrend, um and then which is so. I will say, however, that um um when I looked at this strategy as a user, I'm actually not sure if I would uh kind of do exactly what this strategy does, and the reason is because, in the description of of the strategy, it says that it will adjust the bit spread closer to minimum spread. You know the closer.

A

This down indicator gets to 100 and it'll adjust the aspect. uh You know kind of like push the aspirate closer to them, spread when the up indicator is approaching. 100. um To be honest, I think, as a user as a market maker, I think I would actually be opposite here, because I think that um if that a ruined down indicator is going to 100 and if that indicates a bearish market as in the price is going to go down, I think you actually want a larger bid spread.

A

You actually want you want the you know you want the you want the spread between your buy orders and the mid price be wider, not tighter. um So if anyone disagrees, please like put them in the chat, uh but but I I believe that I think this strategy might actually work better.

A

If this is flipped, um if the, if you're, actually increasing the bid spread when the down indicator is approaching 100 and you're increasing the ask spread um when the the you know up in the other, so um so right now um right now, I think it's actually not doing that. So I think you can see that um you know like right now the down is higher than up, um and so the bid spread is lower than that spread, uh but I would actually expect this to be the other way around.

A

So so, let's um so what do you do this? What are you doing this case? So let's say that you're writing you're using hummingbird, and um you know it's doing something that you you disagree with. uh You want to change out how the strategy behaves so because hung bot is just python. The entire code base is in python um in python. It is actually pretty easy to understand language. I think that most most users, who are somewhat technical, uh can actually you know, convincing modified human bodies uh to their own needs.

A

So um since I'm doing that live, let me show you how to do that right now um live so uh so this is. This is awesome code, so I'm going to open up uh my code editor and I'm going to go and edit I'm going to edit the um the file that uh that was used. So the way the honeywell works is that every every every strategy is basically each each um its own file, its own folder in the um hummingbot uh strategy, folder.

A

So so um this you can also see this in github as well. So if you go to github and you kind of like look at coming by, uh you can see that you know there are uh in the hummingbird directory. um There are, you know, there's um folders for connectors for each uh connector used um as well as as strategies.

A

Each strategy is on counter here, and so, if you look at any of these strategies, there is there is this pyx file, and this pyx file is really where all the logic of of the strategy is contained. Now I know it seems pretty intimidating, but a lot of this is just kind of you know, functions and everything. So, overall, I think those of you with a decent understanding of python um or programming in general.

A

You can easily modify that strategies so, um as I mentioned um so, let's, let's try to modify this so that, instead of making the bid spread uh smaller when the roon indicator goes to 100, it actually increases the spread and um similar. Similarly, when the root indicator goes 100, it should actually um it should actually decrease it should adjust. It should make a bit spread closer to the minimum spread and vice versa, especially basically we're going to try to flip uh this. This logic over here.

A

So to do that um so go to hummingby and then um strategy and then a roon oscillator and then once you're in here you can open up this um oscillator.pyx file. um Now this user did he created a couple other like helper files uh to assist this calculation, uh but I think um you know so you may want to look at those, but I think the the bulk of this this logic is all self-contained in the main strategy file, which is obviously a dot uix.

A

Actually I'm sorry, I believe, especially this one here. Yes, this is this one here, so a roon also root underscore also dot. Pyx is the actual strategy file, so so this is pretty long, um but basically at the very beginning, it's really just initializing initializing a lot of variables and then the real starting point is. This um is really the start function right here.

A

So let's see um yeah yeah. So typically it's really the c-tick function here, where it's the strategy is really starting um and it's kind of like it's coming doing the main logic. So you can see that this is really the core of most coming out strategies over here, where it says it's gonna, um it's basically creating a set of order proposals um and then adjusting those orders based on uh what's happening in the market itself, and so there's a bunch of modifiers.

A

That's adding like adding order levels changing this price, changing the size, and so what this is. This is um essentially we're going to adjust the spreads uh given. uh These are all sliders, so so so, basically- um and here it is so- here's a function which is adjusting the spreads uh used by the bot based on the settings of this rune oscillator, you know a class that has been created, and so you can see here that it says the ask increase um is equal to the differences spread times, basically, basically minus um this up indicator.

A

So um I believe all we have to do is just change this up and down, and then this this word down up um and then it should work. So so, let's try that and um let's see what happens now I'll make that up. um Apologies! If my screen is my screen's visible, uh okay, it looks like it's. It's not quite readable, but uh you know um we'll try to improve the resolution next time, uh so yeah try to go okay. That should be a little bit better.

A

um Basically all I did here was um I just changed this word, you know from up to down and then this one down up um so and so basically now the the ask increase the acid increase, the ask using um using kind of like like a if this down indicator is low, so um so yeah.

A

A

Yep, um okay, so um so yeah. Let's, uh let's try to see how that works.

A

So if you modify code um then make sure to recompile after you, um you change any code, because uh basically, what this is doing is is taking kind of code. That's um in python, it's compiling the the c plus plus parts vector c so that um it can be.

A

A

After that, um we can import the um the strategy again. So here's the strategy we created and then let's start the bond uh and let's see now if it's behaving differently than it was before uh before we um so now what what the the behavior I'm expecting is that when, when, when spreads kind of widen, um you know, let's say if the if there's a downtrend, the bid spread should.

A

A

A

Once again, I think it's going to wait for see um yep it's going to wait for 10 periods of 10 seconds each uh to fill the fill the buffer, a price buffer and now again use that to make the make the.

A

A

Yep and there you go so now, you can see now, there's actually calculating again and now. I think it's acting more, how I would expect where, um thanks to the downtrend signaled by this 80 here, um it's widening the bid spread, and so you can see here even the chart here, that uh the bid spread uh the one in in green is is much lower than it was before, and this is because of this. You know just that over here.

A

So um so you know, as you can see, getting this right is not just the you know, set and forget kind of thing. um You know market making is really about understanding both the market that you're trading, as well as understanding the algorithm that you're using so so. The algorithm here is not just the this kind of a simple setting spreads, uh but it's is using these indicators um and it's adjusting the spreads.

A

You know in responses indicators, but but I think that you know, as a user, um I would say my recommendation that you really understand um you know if you're gonna use a strategy um understand both kind of like how market making works, but also like how this root indicator.

A

You know how it works as well um and then, and then kind of like you know, try to play around with the definitions, the uh this oscillator values um and different things like that uh to see how it works, because um a lot of this stuff is really about kind of like fine-tuning um what your, what your body is doing to kind of match, the kind of like the the pattern that in the market, so the what we could find is that you know like um especially in range bound markets.

A

So I think in the last few weeks, um crypto has really been very range-bound, and- and so I think, market making in these range markets is really a great strategy, because you know, as you can see, for dogecoin, you know if you were kind of like you know. If you had a bot that was kind of like at the right.

A

You know um oscillation kind of size, then you can really kind of pick up and you know buy those how high violence so high during during phases like this, um the the I think, the you know a lot. A lot of the um a lot of this stuff is really going to protect you. You know, because when the market kind of shoots up or down uh that can really hurt you, and so the sooner that your body can kind of like spawn and move um the side.

A

Is you know, trending and make that slime a lot wider? The more that um you see it productive, so the market making a lot of it's kind of like just you know, making money buying low selling high, but avoiding these large moves in the market that might change your internet position and cause you to like give up a bunch of games you're making. So um that's why a lot of users?

A

You know um one thing we always hear from users is they're like you know, I'm making money making money, but all of a sudden you know boom. You know I I lose a lot and that's really because of trends, and so the reason we're adding indicators is to help users avoid those things, but but overall, I think, like um yeah kind of like it's really understanding how these indicators work is also quite important.

A

uh So with that. um Finally, let's see finally have our um because we're in this doge, I think, because we're using we're in a very tight market here. uh This is dojo, uh we're, probably not going to fill any trades, but um but so that's why? um But that's why? Overall, I would say if you're trading, some of the more smaller tokens, especially that's kind of why uh our customers are paying for liquidity, because the equity is not very good. um I think um yeah, I think, for these smaller uh tokens and trading pairs.

A

I think I do things. These spreads are pretty tight for the smaller pairs, uh so I would not use the spreads music here um as and just purely copy them before what you're doing, because uh for doge, it's a very active. You know highly liquid pair right now, uh given everything's happening, and so that's why I'm not feeling any trades, but I think if I want a smaller token, some of our liquidity mining ones, I definitely would have seen some trades being filled uh using this price.

A

So with that, um I hope you guys found this useful. uh You know we're going to try to do this. Every friday, run, bots and um and and also um kind of like going back to the theme of today, um we're really impressed by what this user did in terms of strategy. It went way above and beyond. You know um what we thought, this small kind of like a small bounty.

A

We tried to do and so um we're we're working with a bunch of other projects for interest in putting much larger bounties for uh for users to to build connectors and strategies uh customized for various washing exchanges um and and out there. So so I I think um you know there's some work for you to kind of like to set that up, but but look for look out for some announcements from us in the next few weeks uh with how you can earn some of these bounties by building on television.

A

um So and also, if you have a chance, um we are gonna, we we will be um adding this root oscillator.

A

uh You know kind of like indicator to our inner class, and so um please, but in the meantime, if you wanted to try it out, uh you can actually already run this pull request um and get a rebuild. So um with that. uh Thank you very much for um for for looking at us. um If you, if you um follow us on twitter, uh it's hummingbit underscore io on twitter um and uh and we'll we'll be here next friday again with another edition of coming out live.

A

So uh thank you very much and uh I do have a see.

A

Okay, no okay, okay, anyway,.