algosort's profile picture. High Frequency Trading algo developer.
Futures, Equities & Crypto markets.
Former UBS Quant.

AlgoSort

@algosort

High Frequency Trading algo developer. Futures, Equities & Crypto markets. Former UBS Quant.

置頂

MEGA THREAD. HOW TO BUILD HIGH FREQUENCY TRADING ALGORITHM. In this thread I will share my knowledge regarding to building robust trading strategies based on my 12 years modelling experience. Looking forward to your questions and comments. Enjoy! #HFT #algorithm #trading


AlgoSort 已轉發

I received a lot of surprised comments about me saying 60k lines code is not a lot for a trading system. So let’s dive into that today!

Post mortem of a (almost) 1 million dollar profit HFT strategy In the summer of 2023 I started posting some monthly screenshots of this (at the time) 40+ sharpe ratio HFT strategy. I used to only publish % returns, but if you ever wondered what the $ amounts were, there you go:

crypto_hades's tweet image. Post mortem of a (almost) 1 million dollar profit HFT strategy

In the summer of 2023 I started posting some monthly screenshots of this (at the time) 40+ sharpe ratio HFT strategy. I used to only publish % returns, but if you ever wondered what the $ amounts were, there you go:


AlgoSort 已轉發

I have worked in AI research prior to trading, deployed machine learning models to prod both in industrial applications & in trading the only models other than OLS i have used that actually make serious money trading, are quantile regression & ridge regression curious.

One for the “OLS assumes normally distributed errors” crowd.



AlgoSort 已轉發

basically every 'quant' library in python is a waste of time.


AlgoSort 已轉發

In 2024, Hyperliquid 1) redefined the crypto meta to put community above all. With Hyperliquid, there are no investors, no market makers, and no fees paid to any company. Billions of dollars of value were created by a common vision and relentless hard work. This value went…


Long on BTC, ETH, SUI, HYPE, SUSHI, ENA


AlgoSort 已轉發

Very useful reference sheet for exchange locations. Some are not in AWS (Alibaba Cloud) or this is outdated so not 100% accurate but still v. useful for most: docs.google.com/spreadsheets/d…

quant_arb's tweet image. Very useful reference sheet for exchange locations. Some are not in AWS (Alibaba Cloud) or this is outdated so not 100% accurate but still v. useful for most:

docs.google.com/spreadsheets/d…

AlgoSort 已轉發

Order State diagrams like this look complex, but are battle tested Attempts to simplify means you're accepting nested/coupled states somewhere else in the stack

0xLightcycle's tweet image. Order State diagrams like this look complex, but are battle tested

Attempts to simplify means you're accepting nested/coupled states somewhere else in the stack

market flat to FOMC, #altcoins very week. don't dream on another +30% now. $BTC $ETH new ATHs -> very strong, but Imaging what altcoins do when btc will down -10% #Crypto


Closed entrie lev positions today. 3 weeks trading, highest return on open positions 250%, finally around 100%, for me enough. Tommorow we can start again. It was huge oportunity to earn easy money. Small spot position still open (entry in sept).Need to finish mid frequency algo.


Now might be a great time to buy $btc $eth. 🚀 #Bitcoin #Crypto


Backed to the crypto world, lets make some new analysis.


AlgoSort 已轉發

🇵🇱 The Polish diaspora, with an impressive 20 million individuals residing outside of their homeland, stands as one of the largest in the world—surpassing even India's 18 million expatriates! This incredible pool of talent presents a golden opportunity for Poland to harness the…


AlgoSort 已轉發

To jest zdjęcie z ekranu ukraińskiej OPL w pierwszych minutach inwazji. Ktoś wrzucił, wtedy kiedy jeszcze nie wiedział że nie powinien.

emulatorsiersci's tweet image. To jest zdjęcie z ekranu ukraińskiej OPL w pierwszych minutach inwazji. Ktoś wrzucił, wtedy kiedy jeszcze nie wiedział że nie powinien.

AlgoSort 已轉發

"You're a fraud. You're an absolute fraud." - @CMEGroup CEO Terry Duffy recalls his encounter with Sam Bankman-Fried last March. Listen to the whole interview with Terry, @RiskReversal & @GuyAdami: podcasts.apple.com/us/podcast/on-…


AlgoSort 已轉發

So far applying linear regressions to many more statistical problems in HFT has born me so much more fruit than toying with and tweaking some boosted trees or NN architectures on a few selected problems.


AlgoSort 已轉發

Change in round trip order placement response times during the dump earlier today: FTX: 100ms --> 300ms Kraken: 25ms --> 500ms BFX: 5ms --> 15ms


⬇️⬇️⬇️

One mistake I used to always make was having totally separate ML and production pipelines/infra. I used to do all of my feature engineering in Jupyter notebooks using handy pandas.rolling() methods and easy visualization. In my opinion, this is the wrong approach. 1/5 🧵



AlgoSort 已轉發

Market orders are optimal when inventory and depth imbalance are at extremes, to add more flavor to @IDrawCharts point 1/3

0xLightcycle's tweet image. Market orders are optimal when inventory and depth imbalance are at extremes, to add more flavor to @IDrawCharts point

1/3
0xLightcycle's tweet image. Market orders are optimal when inventory and depth imbalance are at extremes, to add more flavor to @IDrawCharts point

1/3

remember kids: if you're bidding a level and betting on a pullback, if the market is strong you're unlikely to get filled and will probably get left behind when the market rises but if the market is weak and you misread it, you're guaranteed to get filled when the market drops



Loading...

Something went wrong.


Something went wrong.