AMA - Avalanche x Kronos Research

Understand the Cryptocurrency Market Through Quantitative Analysis and High-Frequency Trading Strategies

As the cryptocurrency market flourishes, many investors are seeking automated trading methods. Therefore, quantitative trading strategies have gained considerable attention in the crypto market. Why are quantitative trading strategies suitable for cryptocurrencies? And how can quantitative analysis be used to understand market trends in the crypto space?

Thank you for Avalanche's invitation, which allows us to connect with more partners interested in quantitative trading. This session features Hank, the CTO of Kronos Research, who will share insights on understanding trading markets through quantitative analysis and high-frequency trading strategies!

Host: Letitia (Avalanche)

Guest: Hank Huang (CTO, Kronos Research)

 

Here is the text record of this AMA:

Host: The AMA is about to begin! We are honored to invite Hank, Kronos Research's CTO, to join us for today's event! Hank, could you please briefly introduce yourself and Kronos Research?

Hank: Hello everyone; I'm honoured to have the opportunity today to introduce quantitative trading through this event! We'll also have time for questions later~

I currently serve as the CTO at Kronos, overseeing our technical department and leading our Market Making Team. Kronos' Market Making Team has become a top market maker on over 30 exchanges worldwide, providing liquidity to traders. I have been in the quantitative trading industry for 14 years, previously working at Morgan Stanley, UBS, and Allston Trading and founded three start-ups. Whether in trading strategy research or high-frequency trading system development, I have extensive experience. Kronos Research was founded in June 2018 by Mark and Jack and has grown from just two founders to a team of over 80 people spread across Europe, the US, Singapore, Shanghai, Taipei, and more. Our business scope has expanded from proprietary trading to market-making, venture capital, and asset management. Along the way, we incubated WOO Network. We continue to deepen our presence in the cryptocurrency industry, providing high-quality liquidity and investing in start-ups, aiming to contribute to the crypto industry's growth as much as we can.

 

Q1: Could you explain what quantitative trading is? What are some common quantitative trading strategies in the crypto market?

Simply put, quantitative trading uses mathematical models to "quantify" and predict price movements of financial assets, assisting or automating trading actions. Common quantitative trading strategies can be roughly categorised into high-frequency and medium-to-low-frequency strategies:

1. High-frequency strategies include alpha strategies, market making, and statistical arbitrage.

2. Medium-to-low-frequency strategies include quantitative CTA (Commodity Trading Advisor) strategies, such as trend following and mean reversion.

Kronos utilises all these strategies and continuously develops and optimises them.

 

Q2: What are the advantages and disadvantages of using quantitative trading in the cryptocurrency market? How can quantitative trading be used to analyse market trends in crypto?

Quantitative trading prediction models are typically based on statistics and probability, meaning there is a probability of success and a probability of failure each time the model predicts. The law of large numbers tells us that the actual success rate will approach the theoretical value more closely as the number of predictions increases. Therefore, even if the success rate of some models is only 51%, with enough predictions, the theoretical 2% gap between success and failure can accumulate substantial profits.

The cryptocurrency market operates 24/7, providing more opportunities for quantitative trading models to make predictions. The market's high volatility also provides more opportunities for trading strategies. These are advantages of quantitative trading in the crypto market. However, because cryptocurrency exchanges are globally distributed, the mechanisms for placing orders via APIs and receiving quote updates are not as mature as those in traditional markets, leading to issues like unstable and easily interrupted price data. Additionally, after placing a strategy-driven order, there is much higher uncertainty compared to traditional markets; for example, it may take several minutes to receive trade execution results. These challenges require innovative solutions in developing trading systems and formulating trading strategies. Kronos has invested significantly in overcoming these challenges in its early days, which has contributed to its current success.

Regarding analyzing market trends:

High-frequency quantitative trading typically relies on analyzing "exchange quote data" and "changes in transaction data" to predict short-term market supply and demand changes and thus price fluctuations. Such quantitative models generally have a certain degree of accuracy in analyzing short-term (minutes-long) trends but find it more challenging to predict longer-term trends.

 

Q3: Does learning quantitative trading require engineering and trading backgrounds? As a newcomer to the crypto world, where should one start learning, and what should they pay attention to?

Successful quantitative traders or researchers typically have a strong background in statistics, probability, and mathematics, combined with a certain level of software development skills. There are also increasing numbers of researchers with proficiency in artificial intelligence and machine learning. Mastering programming skills, starting with Python, is highly recommended, followed by a challenge with C++.

 

Q4: How do you decide on the targets for quantitative trading amidst the volatile crypto market? How do you determine the timing for using quantitative trading?

Generally, quantitative trading uses extensive historical data to conduct "backtesting" of models and trading strategies, simulating the profitability of strategies over a certain period in the past to select the most reference-worthy parameter combinations for the future. Therefore, any target for which the backtest results show the trading strategy's profitability is suitable.

Kronos employs quantitative trading around the clock and relies almost entirely on quantitative trading, rarely engaging in subjective trading.

 

Q5: Can you share your views on the future development of the quantitative trading industry in the crypto market?

In recent years, many institutions from traditional markets and established quantitative trading companies and teams have entered the crypto market. The quantitative trading industry in the crypto market will only become more competitive. This long-term competition is beneficial for the crypto market, improving trading volume and liquidity and accelerating the development of the crypto market into a mature financial industry. As an opportunity to recruit, we offer not only technical and research-oriented job opportunities but also logistical and marketing positions. If you are interested in joining us, please check our openings here >> https://boards.greenhouse.io/kronosresearch Related reports: https://www.inside.com.tw/article/26944-kronos-research

 

Here are the live questions:

First question: Hello 👋 I would like to ask, many strategies now emphasize AI and ML, but it seems like trading is done in a black-box manner, with problems like overfitting to overcome. So, I would like to ask the expert, at which stage in strategy development do you introduce AI or ML, or do you prefer to hand over the program after filtering specific features for rolling updates?

This is an excellent question! I've heard many people trying to use AI/ML models directly to predict price changes. The quality of the model depends significantly on the data fed into it; typically, unprocessed data tends to be very noisy. We start by analyzing some trading behaviors (e.g., how large holders buy and sell large quantities of coins), economics, and financial knowledge to develop basic statistical models. We have thousands of such statistical models, and in the final stage, we integrate multiple models using ML.

 

Second question: Most cryptocurrency projects are built to accumulate wealth for themselves. So, I would like to know what values you intend to add to the cryptocurrency industry to bring more wealth to the industry?

In the capitalist market, investors are willing to invest in industries that they believe will generate value in the future, which is crucial. Investors are willing to invest because they expect returns, and one way to achieve returns is through resale investments in the secondary market. One of the greatest contributions of quantitative trading is to maintain liquidity in the secondary market, allowing buyers and sellers to freely trade anytime, anywhere. In addition, our Venture Arm has already invested in numerous new projects within the industry, participating in many seed rounds and private sales, and assisting these projects in coin issuance and exchange listings.

These are direct or indirect supports for the future of this industry~

 

Third question: I have some experience in algorithmic trading and am curious about how companies like Kronos decide whether to implement a strategy based solely on back-testing results. When a strategy malfunctions, what criteria are used for elimination? I would like to understand more about how large organisations handle this aspect.

In fact, analysing backtest results alone is a significant discipline. Each backtest result differs in terms of "how much money is earned" and can be analysed comprehensively by many other metrics to find the most "stable" parameters, which are not necessarily the most profitable in backtest results~

 

Fourth question: What kind of people are suitable for quantitative trading?

In my opinion, quantitative trading is a highly competitive industry where your actions are virtually instantaneously reflected in the market. If you are someone who loves competition, faces different challenges every day, enjoys solving the same problem with new ideas continuously, and can patiently address new issues, then you should enjoy quantitative trading.

Of course, lucrative returns are also a reason 🤩

 

Fifth question: The cryptocurrency market is highly volatile and more susceptible to news. Does this make it less suitable for quantitative trading compared to the stock market?

In fact, the stock market is also greatly influenced by news, but for our high-frequency quantitative trading team, "news" is typically not considered because we analyze very granular data; even a few tens of milliseconds involve a lot of data. The holding and pricing times are not too long, so they are not likely to be suddenly hit by news. Instead, because we analyze market data very close to the millisecond level of market "supply and demand," our strategies can more quickly respond to market changes caused by news. When volatility is high, it is actually a good time for us. So, in this sense, we actually prefer targets influenced by a lot of news; the greater the volatility, the better~

 

Sixth question: After implementing quantitative strategies, what unexpected drawdowns are there? For example, maximum drawdown, longest drawdown time, and whether the strategy is considered ineffective?

These judgments can all be made

 

That's all for today's AMA. Thank you Hank and everyone for your participation.

In the future, there will be more AMA events. Please continue to follow the community for the latest updates!