I'm at around 2 months into this mm/hft rabbit hole.
I feel like you can learn by yourself most of what you will need to become a good dev, but finding good resources about trading is extremely hard
second day of testing live and it looks like I have some latency and fair price problems. Markouts are generally positive but once in a while I get very toxic fills from big taker orders
I will post my entire learning arc about crypto market making while building a bot from scratch. I will be trading on Hyperliquid with a $50M volume goal
I recently had to change how I model my bid/ask spread in order to quote a bit wider. Nothing special I just track volatility, trading intensity and inventory.
I will post my entire learning arc about crypto market making while building a bot from scratch. I will be trading on Hyperliquid with a $50M volume goal
Switch to Hyperliquid for a holiday gift 🎁
CEX traders get 50% off fees for your first $50M volume. Go to to get started.
Fees become 0.012% taker and 0.002% maker rebate - the cheapest place to trade perps. All you need is an email to start trading.
-
today I'm working on my theo model, never worked much on anything similar so it's a lot of stuff to process... I will probably spend a lot of time on this until I get back to other parts of my mm infra
Do you guys have any advice on where to start from/study to build a fair price model? What would you focus on if you started from zero? ideas, books, papers and repos are welcome
today I'm working on my theo model, never worked much on anything similar so it's a lot of stuff to process... I will probably spend a lot of time on this until I get back to other parts of my mm infra
I feel like I have a good idea on how wide my quotes should be in an optimal market condition. But when volatility spikes it seems like my quotes don't adapt fast enough. I'm trying to implement some simple feature to make my spread model more "reactive" in the short term
super busy with uni this week but I finally finished my spread model or whatever you want to call it. No more crazy things on volatility spikes, still feels far from perfect tho
super busy with uni this week but I finally finished my spread model or whatever you want to call it. No more crazy things on volatility spikes, still feels far from perfect tho
should I try to quote at multiple depth levels to get a better avg fill price on this large market orders? I will probably test something live and quickly see if it's better or worst
high volatility --> more inventory --> quotes get skewed to much on one side
opposite will happen when vol is little --> quotes remain almost symmetrical and I accumulate too much inventory
I'm building a very simple market making infra for a university project (classic A&S model). I update my quotes every 1s and I want a wider spread as volatility increases. How do you model the volatility parameter so that it is meaningful for such a short period?
@thodoha
mmh I see, also I guess rate limits could became a problem... but I guess that most of the times my theo changes I wouldn't have to cancel all orders at once
let's say I want to start quoting more then 1 level per side, how do you decide at what depth you are going to quote the second level bids/asks. Any suggestions on how to build this?
I'm not fumbling PURR here, would make sense if buying is rewarded more then just holding the airdrop.
But maybe this is just cope for not having a spot mm infra ready
ig it's time to move all the infra monitoring out of the main process, it's getting quite heavy on my ec2 instance. Do I keep error monitoring and move out the rest?
I'm trading on OKX but a lot of data I use comes from other exchanges (Binance, Bybit...) How do you decide where to put your infra? (ali cloud, aws hk, aws tokyo)
btw I removed any implementation of the A&S paper into my market making infra. I think it's a good theoretical paper but I don't find it useful for what I'm trying to achieve
HIP-1 and HIP-2 will soon open up for users to permissionlessly deploy on the Hyperliquid L1.
To recap, HIP-1 is a standard to deploy native tokens and onchain spot order books. HIP-2 permanently commits liquidity to a spot order book for HIP-1 tokens and synergizes with users'
question for all CT quants: is the microprice paper by Stoikov a good starting point to build a fair price model for crypto markets? I'm thinking of adding more features to it other then the ob imbalance;
I'm currently quoting 1 order per side but I'm thinking about starting to place multiple orders at different depth levels. Is this the most common approach when market making crypto?
the main problem I've found is that the microprice is always somewhere between the bid and ask spread which doesn't seems to be a realistic assumption in crypto perps
I've just realized that rather than trying to improve some of this bad fills I can just place less orders. Need to find the good middle ground so that I can grind some volume
about my last post: I added a request queue to my mm infra. I no longer use a fixed interval between order updates. It's not completed yet but it's better then what I had before...
@BeatzXBT
*from stackoverflow
The "T" or Trade time is the time of the transaction in milliseconds.
The "E" or Event time is the time value is inherent only to the sockets. It is associated with creating a socket object.
I still don't understand how there is such a huge difference in ms
@BeatzXBT
Hey, I'm building something similar for a uni project but I'm having some problems understanding how to model short term volatility. How do you define a volatility range?
I'm building a very simple market making infra for a university project (classic A&S model). I update my quotes every 1s and I want a wider spread as volatility increases. How do you model the volatility parameter so that it is meaningful for such a short period?
this is the first time I came to a roadblock in my project. I'm having an hard time understanding how to implement pricing features in my market making infra, I will have to do some studying before I can progress. Any resource or ideas are welcome
question for all CT quants: is the microprice paper by Stoikov a good starting point to build a fair price model for crypto markets? I'm thinking of adding more features to it other then the ob imbalance;
I have been working a lot on my simple market making infra in the past few weeks. Any tips on how to manage rate limits? Maybe there are some good repos out there but I can’t find any