Why Quantitative Hedge Funds Fail

We get tons of questions regarding the right infrastructure to succeed in this business, and the number one reason always will be “having a profitable model”. Without that, nothing else will matter.

So this post will be assuming small funds and startups trying to enter into this space (under $100M AUM) where the organization (whether is in the buy-side or sell-side) has one or more profitable models to run, so we can describe what makes them fail, and hopefully, you can avoid these common pitfalls.

This top 10 reasons list is by no means “the fact”, but is based on our own experience serving such firms with our Advisory service and those using our low-latency HFT platform.

1 – Not enough capital commitment

Setting up a quantitative fund with all the resources needed is the easiest part. You just need to budget whatever you need and then execute it. But the ongoing maintenance and expenses are where all the capital goes. For a detailed list, you can read our previous post on this.

2 – Relationships

Accessing quality liquidity pools and order flow is one of the most important things.

Also, finding a broker-dealer that will work with you is very important. The broker-dealer will be your middleman between the public marketplace and your funds.

3 – Technology and the Team

Technology and the Team: those who don’t have the tech team in place, will certainly fail without a doubt. Having the best tech and engineering teams in place, you can afford to take risks with more confidence, and the statistics show that it will likely achieve success rather than not.

4 – Data access

Those who don’t have direct access to the market (DMA) are going to be at a clear disadvantage from the beginning.

Having collocated servers, next to the exchange and live/direct feeds are the key to have market intelligence, providing you with what you need to stay ahead.

5 – Lack of innovation

Nowadays, being the fastest is no longer the “only” factor. Algos needs to be smarter and more creative.

Constant research is paramount. for the continued success of a trading firm.

As such, things like machine learning, big data, and artificial intelligence have become prevalent in the development of algorithmic trading. Applying these methods to the analysis of order flows, execution patterns, and market microstructure will give enough intelligence to your models.

6 – Market Impact

Not having the right picture of how the orders you are sending will impact the microstructure of the market, potentially, will make your strategy fail. To get the right picture, you need to consider how orders are distributed across the market. In this case, you would want to see a real-time order book picture and its current dynamics.

It is paramount to know the behavior of orders you are going to send into the market beforehand. How your order will impact other orders already in the market, how they will shift, etc.

A lot of people get bad advice when it comes to trading. This is because they do not have a good understanding of how all the pieces fit together and how the technology could help with this.

7 – Rely on simulators/backtesters

In high-frequency trading (hft), there is no way to simulate your strategy with enough confidence. Models must be tested in LIVE with real money.

The most important thing about a model is that you KNOW how it would have performed in LIVE. And In order to know how your strategy would have performed live, you must run it many times under the same conditions.

8 – Risk Management

Lack of risk management is one of the reasons companies go out of business. Rare events happen all the time. Poor risk management can lead to over-or under-reaction, with predictably poor results.

Some common techniques could be position sizing, real-time diversification, and portfolio-level hedge.

9 – No Monitoring

Having the right monitoring systems to let algos know that everything is ok. From monitoring market data, networks, exposures, and bad behaviors. And of course, the human factor is very important as well. Human surveillance will give you much better decision execution when things go bad. Keep in mind, that these monitoring systems should not be intrusive to the main system. They need to run independently.

10 – Resciliesense

Having systems and procedures in place to recover from unplanned situations could save the business.

What happens to your trading system if your servers got disconnected from the exchange?

What if your market data is corrupted (because of an internal problem or the exchange sent bad data)?

What if the market closes and you still have open orders? or worst, open exposure?

Your trading system must be able to recover from any potential disaster, and if that can be handled automatically, businesses can run with no problems.

Conclusions

Again, we built this top-10 list based on our past interactions with clients that had failed and succeeded. It is coming from our personal experience.

We want to encourage our followers and readers to share their experiences and add to this list more “factors” that make Quant Funds fail.

👉 PLEASE, share your thoughts!! 👈

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.