Jesus Rodriguez: 10 Reasons Why Quant Strategies for Crypto Fail | Fintech Zoom

Jesus Rodriguez is the CEO of IntoTheBlock, a market intelligence platform for crypto assets. He’s held leadership roles in major tech companies and hedge funds. He’s an active investor, speaker, writer and guest lecturer at Columbia University in New York. 

The conditions “crypto” and “quant” appear to go perfectly together. Bitcoin and crypto resources were created during one of their most enjoyable times in funding markets coinciding with the gold age of organizational finance. The technological acceleration brought on by moves like cloud computing and large information along with the renaissance of machine learning have collided to induce the best storm in favour of this quant revolution. Billions of dollars are changing hands each year from discretionary funds into quant vehicles, and Wall Street can’t employ mathematicians and machine learning specialists quickly enough. 

Being a totally digital asset category, crypto seems like the ideal goal for quant models. And quant plans stay restricted to relatively simple techniques like statistical arbitrage (a set trade plan that seems to exploit market inefficiencies at a set of securities) and we haven’t seen the development of big dominant quant desks on the marketplace. Regardless of the appealing qualities of crypto resources for quant plans, crypto introduces unique challenges for quant models and the truth is that almost all quant plans in crypto neglect. In the following report, I’d love to research some of the basic but not clear reasons which can lead to the failure of the majority of quant strategies from the crypto area.

See also: Jesus Rodriguez – Crypto Needn’t Fear GPT-3. It Must Embrace It

By asserting that many quant approaches in crypto neglect, I’m referring largely to machine learning plans. Statistical arbitrage has turned out to be an effective mechanism to create algorithmic plans, but we ought to expect those chances to evaporate as the market increases in size and efficacy. In conventional capital markets, we’ve observed an explosion at the execution of system learning-based quant models along with also the entire body of study in the area is growing exponentially. 

But, the majority of the quant strategies proven successful in conventional capital markets are very likely not to work too when applied to crypto assets. According to a few of our latest experience at IntoTheBlock focusing on predictive models and quant plans, I’ve recorded a number of those elements that I think can get the collapse of quant models to get crypto assets.

Most of those machine learning-based quant approaches you see in study papers are trained in years of information from capital markets. The trading history of the majority of crypto assets could be counted in weeks, also, even for vehicles such as Bitcoin and Ethereum, the datasets remain comparatively modest. Most machine learning models will have difficulty generalizing any understanding from these smallish datasets. Let’s say that you’re working to construct a predictive model for your price of an advantage such as ChainLink (hyperlink ), which can be red-hot lately. It ends up LINK has a tiny trading history, which can be inadequate to train nearly all machine learning models in quant fund. 

Though the phrases “regular” and “outlier” shouldn’t be utilized in precisely the exact same sentence, I can’t think of a better phrase to describe what we encounter in crypto assets. Enormous price crashes or abrupt spikes which, at a lapse of a couple hours, alter the momentum in almost any crypto asset. All these “outlier” events occur quite often with lots of crypto assets.

In a machine learning perspective, many models will probably be puzzled with those price moves since they haven’t seen anything like during coaching. It’s not surprising that lots of machine learning quant models got decimated throughout the flash accident of mid-March or neglected to capitalize from the abrupt rise in volatility during the past couple of weeks. It’s challenging to catch knowledge for all those kinds of events throughout the practice of this model.

A negative effect of this little market datasets in crypto resources is that the propensity of the majority of machine learning quant models into overfit or into “optimize for the training dataset.” We continuously see quant models that function exceptionally well during backtesting simply to fail when applied to actual market conditions.

Consider this situation: You have produced a predictive model trained over a couple of years of Bitcoin trading background, you then encounter weeks of virtually no volatility followed with some loony volatile times (not that it’s happened previously ). You’d love to retrain the model to catch this knowledge, but how? If you merely retrain the model at the latest data, there’s a strong prospect of overfitting while in case you wait the comprehension may not be applicable any longer. 

This retraining issue is an immediate effect of this “regular outlier events” phenomena. Should you train a model at a dataset in the past 10 years of this S&P 500, you are able to design a plan to retrain the model often since it’s not likely that the index will deviate a lot from its conventional behaviour in brief intervals. This routine retraining of models which was well embraced in conventional quant plans goes out the window when it comes to crypto. 

One of the greatest drawbacks of designing system learning quant models to get crypto resources is that the inadequate quality and trustworthiness of datasets. It’s not a secret that lots of exchange order publication datasets are filled with documents that signify fake volumes, wash transactions or spoofing behaviour. Evidently, training a machine learning model with those datasets won’t create any relevant outcomes. Furthermore, nearly every week we hear about exchange APIs having outages and shutting down for hours. When was the last time you heard a Nasdaq API crash? It definitely happens, but not that frequently. That lack of reliability can kill the accuracy of the most robust quant models.

Blockchain datasets remain one of the richest sources of alpha for quant strategies in the crypto space. But the anonymity of blockchain records makes it really challenging to design meaningful quant models. Let’s say, for instance, that one of the features in a quant strategy leverages the address count in the Ethereum blockchain. Well, addresses that are part of exchanges are fundamentally different from addresses of individual wallets and those are different from miners’ addresses. Labeling blockchain records is essential to design meaningful quant models based on blockchain datasets and, unfortunately, those efforts are still in the very early stages. 

Factor models have been at the center of some of the most successful quant strategies in the last two decades. Entire mega funds like AQR were built on the promise of factor investing quant strategies. From the original factors like value, momentum, or quality, factor strategies have grown to hundreds of factors that model relevant behaviors in financial asset classes. 

At least until today, most factor strategies have proven to be ineffective in the context of crypto assets. When it comes to crypto, factors like value and quality are not clearly defined and the behavior of others such as momentum defies conventional patterns. This causes many crypto quant desks to spend numerous hours trying to recreate factor-based strategies that are highly unlikely to perform in the crypto space. 

The field of quantitative finance is rapidly gravitating towards large and complex models regularly outperform simpler and more specialized models. This trend is a reflection of what’s happening in the entire machine learning space. The advent of deep learning showed us it’s possible to create highly complex neural networks that acquire knowledge in the most unthinkable ways. 

Funds like TwoSigma and WorldQuant are actively pushing deep learning research and incorporating ideas coming out of the AI labs of tech giants like Google, Microsoft, or Facebook. Yet, in the world of crypto, most quant strategies still rely on very basic machine learning paradigms like linear regression or decision trees. 

Simpler models are unquestionably attractive given that they are easy to understand, but they can have a hard time generalizing knowledge from a complex environment such as the crypto markets. As a machine learning environment, crypto combines the complexity of a financial market with the inefficiencies and uncertainty of a new asset class. Definitely not the best fit for simple quant strategies. 

Complementing the previous point, most quant infrastructures in the crypto space are relatively nascent. A robust quant infrastructure goes beyond good strategies and includes elements such as risk management, backtesting, portfolio management, strategy execution, error recovery and many others. In the crypto space, the quant infrastructure of most hedge funds remains relatively simple which makes it difficult to operate certain types of strategies. 

See also: Jesus Rodriguez – Myths and Realities: Sentiment Analysis for Crypto Assets

For instance, suppose that you have designed a beautiful deep learning quant strategy that forecasts the price of Bitcoin based on blockchain datasets. To operate that strategy, a fund would need an infrastructure that collects blockchain records regularly, has the computer infrastructure to run deep learning models, the appropriate retraining tool, and so on.

Today’s technology has certainly reduced the time and cost required to build a quant infrastructure to run machine learning models, but quant desks remain relatively basic compared to those operating in traditional capital markets.

I left the most controversial point to the end. As a financial market, crypto is still failing to attract top quant talent with relevant experience in traditional capital markets. We are still tackling incredibly complex problems such as forecasting the behavior of an asset class with relatively simple models, basic infrastructure and poor processes. Talent is a very important, and often overlooked aspect, to grow quant investment as a discipline in the crypto space. There are incredibly talented quant teams in crypto, but they are the exception, not the rule. 

These are some points that might cause us to reflect of the current state of quant investment in the crypto space. Crypto is an ideal asset class for quant strategies and, in the long run, quant funds should be the dominant investment vehicle in crypto. The path includes many challenges, but also fascinating opportunities.