Warming up...!
So, you ( Quant traders & Geeks desirous of joining us ) have read --- “ Why this blog “ below…Before you send your bios to us, I suggest you read this so that we can make sure we’re on the same page.
Getting set…
Now that we are getting set into action mode, let us give you a primer on what we mean by trading strategies and how you can help us with your algorithmic skills. We want you to develop user defined and friendly software tools that bolster algorithmic trading strategies, which shall focus on the forefront of algorithmic trading and purging the lines between sell and buy side functions.
What is Algorithmic trading ?
A little reality check. To us, the true definition of algorithmic trading is a trader configuring and managing algorithms to automate a process that he or she would have to carry out manually, such as when a trader wants to place a large order. They can use an algorithm that is going to chop up that order through functions like volume-weighted average pricing (VWAP), and drip those smaller orders into the market. We also see a number of systems where there isn’t a trader involved at all, and we call that auto-trading. We’ll be working on projects in which the spot traders are continuing to handle all of the big deals that come through, but they have a lot of small deals that are now being sent straight through to an autotrading engine. The automated trader gets all the relevant data feeds, monitors its own positions to check whether at any stage, value at risk (VAR) or other key thresholds go beyond predefined levels, and then trades accordingly based on incoming flow.
How does this work…?
Simple. In these cases, they’ll be given a strategy to trade and strict parameters to stay within. However, there are lots of variants of auto trading— auto market-making systems, for example. For over-the-counter (OTC) products, customers might come in and request a quote for a particular bond. One can give them a price but you might want to have an autonomous algorithm that optimizes that price, based on past dealings with that customer. You might monitor how many quotes you have given them in the past and how many they have taken up. You can also look at your own internal feeds and the inter dealer feed. So, theoretically, you have this enormous brain that monitors everything that is happening—and has happened—and reacts accordingly. You also have systems that will automatically hedge your positions for you—systems that work behind the scenes and will gauge where you might be exposed at any given time. They will then fire off corresponding trades that close out that exposure to make sure you’re keeping it as risk-neutral as possible.
What are the latest trends that we need to reckon…?
Industry thinks we’re definitely going away from the traditional VWAP, and this is another reason why trading groups are putting their own secret ingredients in their algorithms and continuously evolving them as the market itself evolves. We’re also seeing some early interest in ‘self-evolving’ algorithms. Let’s say you’re doing a spread-trading strategy. You could potentially run 10,000 different variants of that same strategy, with each one having its own distinct P&L. You can then feed each one with market data, and even though you might only put one of them live at any given time, when one of your other algorithms begins to show better performance, you can swap that in and swap the other one out. You can grow variants and progress the branches that are successful. In this way, no one can reverse-engineer you because you’re constantly changing the live algorithm to look for optimal performance and killing off the less profitable variants. It’s quite cruel really. I feel sorry for some of the algorithms because it’s pure survival of the fittest out there. Darwin would be proud.
Is this a feature that is already available elsewhere…?
Yes. We would like this to be a module built into our system. This would be really interesting stuff which starts off with an idea that excites the user dealers.
How can technology neutralise the traditional distinction between the buy side and the sell side?
We think it definitely can. There’s always been some blurring at the top end of the buy side—a big quant fund might look like a sell-side firm in that it has a lot of technology staff internally that is able to code complex trading strategies into black box applications. The big difference is that people who are further down the chain are now being given the opportunity to develop their own strategies and use the platform themselves. Also, it’s becoming less viable to have an enormous IT staff in-house to code these strategies, because the time to develop them is definitely changing the metrics of the space. If you take too long to deploy a strategy, then someone else will have eaten your lunch. You’re also seeing banks setting up innovative proprietary trading desks, which are much freer and behave almost like a hedge fund within an investment bank. You’re also seeing hedge funds doing interesting and advanced things themselves that might make them resemble a sell-side firm.
So, technology is contributing to this blurring because both sides now have the same tools?
Absolutely. When you sell to large players on the brokerage side, they become your partners and roll out the technology for their buy-side clients. The smaller buy-side firms can ask their brokers to host their strategies, but the bigger buy-side players might get the platform themselves and run it internally.
Getting set…
Now that we are getting set into action mode, let us give you a primer on what we mean by trading strategies and how you can help us with your algorithmic skills. We want you to develop user defined and friendly software tools that bolster algorithmic trading strategies, which shall focus on the forefront of algorithmic trading and purging the lines between sell and buy side functions.
What is Algorithmic trading ?
A little reality check. To us, the true definition of algorithmic trading is a trader configuring and managing algorithms to automate a process that he or she would have to carry out manually, such as when a trader wants to place a large order. They can use an algorithm that is going to chop up that order through functions like volume-weighted average pricing (VWAP), and drip those smaller orders into the market. We also see a number of systems where there isn’t a trader involved at all, and we call that auto-trading. We’ll be working on projects in which the spot traders are continuing to handle all of the big deals that come through, but they have a lot of small deals that are now being sent straight through to an autotrading engine. The automated trader gets all the relevant data feeds, monitors its own positions to check whether at any stage, value at risk (VAR) or other key thresholds go beyond predefined levels, and then trades accordingly based on incoming flow.
How does this work…?
Simple. In these cases, they’ll be given a strategy to trade and strict parameters to stay within. However, there are lots of variants of auto trading— auto market-making systems, for example. For over-the-counter (OTC) products, customers might come in and request a quote for a particular bond. One can give them a price but you might want to have an autonomous algorithm that optimizes that price, based on past dealings with that customer. You might monitor how many quotes you have given them in the past and how many they have taken up. You can also look at your own internal feeds and the inter dealer feed. So, theoretically, you have this enormous brain that monitors everything that is happening—and has happened—and reacts accordingly. You also have systems that will automatically hedge your positions for you—systems that work behind the scenes and will gauge where you might be exposed at any given time. They will then fire off corresponding trades that close out that exposure to make sure you’re keeping it as risk-neutral as possible.
What are the latest trends that we need to reckon…?
Industry thinks we’re definitely going away from the traditional VWAP, and this is another reason why trading groups are putting their own secret ingredients in their algorithms and continuously evolving them as the market itself evolves. We’re also seeing some early interest in ‘self-evolving’ algorithms. Let’s say you’re doing a spread-trading strategy. You could potentially run 10,000 different variants of that same strategy, with each one having its own distinct P&L. You can then feed each one with market data, and even though you might only put one of them live at any given time, when one of your other algorithms begins to show better performance, you can swap that in and swap the other one out. You can grow variants and progress the branches that are successful. In this way, no one can reverse-engineer you because you’re constantly changing the live algorithm to look for optimal performance and killing off the less profitable variants. It’s quite cruel really. I feel sorry for some of the algorithms because it’s pure survival of the fittest out there. Darwin would be proud.
Is this a feature that is already available elsewhere…?
Yes. We would like this to be a module built into our system. This would be really interesting stuff which starts off with an idea that excites the user dealers.
How can technology neutralise the traditional distinction between the buy side and the sell side?
We think it definitely can. There’s always been some blurring at the top end of the buy side—a big quant fund might look like a sell-side firm in that it has a lot of technology staff internally that is able to code complex trading strategies into black box applications. The big difference is that people who are further down the chain are now being given the opportunity to develop their own strategies and use the platform themselves. Also, it’s becoming less viable to have an enormous IT staff in-house to code these strategies, because the time to develop them is definitely changing the metrics of the space. If you take too long to deploy a strategy, then someone else will have eaten your lunch. You’re also seeing banks setting up innovative proprietary trading desks, which are much freer and behave almost like a hedge fund within an investment bank. You’re also seeing hedge funds doing interesting and advanced things themselves that might make them resemble a sell-side firm.
So, technology is contributing to this blurring because both sides now have the same tools?
Absolutely. When you sell to large players on the brokerage side, they become your partners and roll out the technology for their buy-side clients. The smaller buy-side firms can ask their brokers to host their strategies, but the bigger buy-side players might get the platform themselves and run it internally.
posted by Krish at 7:08 PM
0 Comments:
Post a Comment
<< Home