HIP-18 Marketplace for Hummingbot Strategies (Phase 1)

Author: [0x1…0] (Snapshot)

Original Post Date: 3 Aug 2022

[Snapshot Proposal Link] (not yet)

Category:
Dev Grants

Summary:
For now, only technical guys can use Hummingbot to make markets. For this reason, Hummingbot is still very niche. To really “democratize high-frequency trading”, there needs to be a way for anybody to market make using Hummingbot with just a click of mouse.

We are planning to achieve that in two phases.

Description:
We are building Phezzan Protocol (http://phezzan.xyz/), the first perp DEX on zkSync 2.0 testnet. Since our DEX is an orderbook style DEX, it is very hard for retail investors to be LPs. So we figured we can use Hummingbot to run strategies for retail LPs, and retail LPs can just click a mouse and select the strategy they want to participate. Essentially, LPs put money in our vault, select a certain strategy, and our server runs the Hummingbot for them.

Anyone can upload Hummingbot strategies. Hummingbot’s upcoming bot management tool and private VM will make it a lot easier.

Our plan have 2 phases:
Phase 1 (This HIP-18):
Anyone can upload LP bot strategies to Phezzan’s backend. Retail LPs can just click a mouse and select the strategy they want to participate.

By completing Phase 1, Hummingbot’s reach will be way broadened, as now tons of retail investors can be an LP (using Hummingbot under the hood, without need to code/deploy/manage). This already democratized high-frequency trading.

Work involved in Phase 1 (This HIP-18):
Backend infra for supporting uploading LP strategies, run on VMs, etc. (20 man day)
Vault smart contract logic to allow LP bot to trade with retail LP’s money, and share the profit with LP bot uploader. (10 man day)

Phase 2:
Abstract away the code logic so retail LP can provide liquidity to not only Phezzan, but also any other orderbook style DEXs. Essentially, this becomes a market place for hummington strategies.

Our team are confident to deliver Phase 1. Anyone interested in Phase 2 can take the existing logic and go from there.

HBOT amount (for Phase 1, i.e. HIP-18)

  1. 600k to 0x4F787422d8db8a7619d45bCF669CafE6cc939218
    20k HBOT per man day. 30 man days in total. 20k * 30 = 600k
1 Like

I agree that improving access to HFT strategies is a worthwhile goal, but I think one of the ways that Hummingbot differentiates from web-based trading bots like 3Commas and BitsGap is by being open source and publicly auditable. Users can see exactly what strategies are being run and have full control over them.

Some questions:

  • How is the Phezzan approach different from web-based bots? How can users trust that their strategies are being executed?
  • Does this require a Phezzan connector for Hummingbot? If so, who will build it?
  • The scope of work in Phase 1 seems like it primarily benefits Phezzan, while Phase 2 would be more beneficial for the overall Hummingbot community. Why should the community allocate tokens only for Phase 1?

Thanks for replying!

  • How is the Phezzan approach different from web-based bots? How can users trust that their strategies are being executed?

Bots will be run in a confidential VM so the algorithms are kept secret. Also we have mechanism to ensure the bot will run as expected, unlike other web-based bot.

We have keeper bots to constantly check the hash of the VM images, if the hash changes that means the VM/code is corrupted and we will stop taking orders from that bot.

  • Does this require a Phezzan connector for Hummingbot? If so, who will build it?

Yes. Phezzan team will build it.

  • The scope of work in Phase 1 seems like it primarily benefits Phezzan, while Phase 2 would be more beneficial for the overall Hummingbot community. Why should the community allocate tokens only for Phase 1?

We will open source our code in Phase 1. So even if we don’t do phase 2, whoever interested can pick up our work.

Also, by opening source our phase 1 code, others DEXs can use our code and increase the presence/revenue of Hummingbot as well.

Finally, more Hummingbot usage (even just related to Phezzan) means more revenue for Hummingbot and thus its token, correct?

Thanks again for taking such a thorough look at this!

My 2 cents… A potential catch-22 situation… not trying to discredit or discourage marketplaces for Hummingbot, (in fact it’s a great idea) but do keep in mind

On one hand, someone who came up with a brand new strategy might not want the IP to be fully visible

  • does the owner upload pure Python files? Cythonized ones? etc.

On the other hand, if the uploaded strategy is obfuscated/‘compiled’

  • who verifies that the strategy contains no malicious code? (For instance, certain bad actors can code seemingly legit strategies but with time-activated malicious code, etc)

SoothsayerX
We can let user know whether the code has gone through an audit or not.

Some providers can choose not to audit (e.g. if they are Jane Street, Jump, etc.). They will lose their reputation if they cheat. So they have social stake in it.

Some providers can choose to go through an audit. Our team can audit this.

Some providers can open source the code. AMM style algorithm providers are doing this.

We will let users know which one of above scheme they choose. It is up to users to decide whether they want to trust the algorithm or not.

The market is efficient and smart to work out a way.

1 Like

The proposal is out:
https://snapshot.org/#/hbot-ip.eth/proposal/0xaf1541d48ba69a4b62947e1eb51895af393b2217d923bb4a0ec7262d40fc9380

Will there be any sort of fee share revenue for the Foundation that will help benefit the community even more? It seems like the Foundation will be allocating tokens to create your infrastructure but without getting any real benefit in return.

Also, since users just need to select a strategy to use, most users will probably not realize that Hummingbot is running in the back end or if this is advertised - what happens if the strategies/infrastructure doesn’t run as intended and the blame is placed on Hummingbot?

Its true that having a ‘one-click’ instant, Hummingbot deployment would have mass appeal.

However, these are some issues:

  1. Phezzan is a new Platform, at testnet phase.

    Personally prefer that Hummingbot be used at ‘live’ & well-tested platforms, to confirm the platform’s service functionality & reliability.

    Our existing community has quite high standard or demand on these aspects.

  2. Phezzan is a DEX with perpetual trading.

    a. Gateway v2, AFAIK, is connecting to spot-type DEXes. Perpetual Dex may need some special implementations that will require more time for testing.

    b. Currently, there are no perpetual trading strategies available for Gateway v2. Developing new, effective strategies will take time, maybe 1-3 months.

    c. Gateway v2 is still in heavy development, thus Phezzan team need to keep up with its updates for the next few months.

Thus, personally, the proposal is just too early, for Hummingbot integration. I would like to see the platform to be ‘live’, and see its users organically growing along with enthusiastic supporters. Perhaps, after 6 months, at least, from launch, we can review this proposal. By then, we will have stronger basis and stats to decide on.

Goodluck Phezzan team! :clinking_glasses:

1 Like