Dear All
Most Brokers urging Clients to go the Algo Trading way.Many of us starting to feel that Algo is the perfect way to go and it is the future.
Recently came across this article in my weekend reading about Algo trading.
Written by Shrinivas Viswanath, founder, RKSV, www.rksv.in
Worth a read.
https://www.quora.com/How-much-feas...me-set-up-Algo-Trading-as-per-my-requirements
Most Brokers urging Clients to go the Algo Trading way.Many of us starting to feel that Algo is the perfect way to go and it is the future.
Recently came across this article in my weekend reading about Algo trading.
Written by Shrinivas Viswanath, founder, RKSV, www.rksv.in
Worth a read.
https://www.quora.com/How-much-feas...me-set-up-Algo-Trading-as-per-my-requirements
Before I explain how a typical algorithmic setup works, I would caution everyone of the following:
1) Beware of any person/firm that promises returns. No one has a magic formula that can yield consistent returns. If that were the case, ask yourself why they wouldn't pump more money into it themselves instead of getting in the business of selling it to you.
2) There is no algorithm out there that prints money for you without you doing any work. No software/company will ever make money for you. Especially if someone says that the algorithmic trading strategy has been "proven to make money" -- run far far away.
If you're looking for stable returns without much effort, look up investing.
Do your own research and your own backtesting and find something unique that others haven't spotted. If the "strategy" is out in the wild, it's probably not going to work. Finding a trading strategy that works is hard, but not impossible. But it is hard.
Okay, now that's that. There are typically a few options for algorithmic trading that a retail trader can use. Algorithmic trading is a blanket term used for any trading that's based on signals. Signals are usually market events (news, price changes, indicators, etc...) that trigger an action (buy, sell, hedge, etc...).
Algorithmic trading can be as simple as waiting for a stock to hit a particular price and telling the computer to buy it automatically. It's nothing more than just creating a series of conditions and then attaching a trigger at the end of it.
There are a lot of programs that can help you do basic algorithmic trading. You can search the Internet for names or ask an electronic broker for guidance. With most trading software, you can configure it to watch for certain conditions and then trigger an action when the conditions are met. That's called your "algorithm."
Now the next question most people ask is -- okay, but how do I know if I'll make money? Here's where backtesting comes in. Backtesting is just taking your algorithm and running it across several weeks or months of market data. Test run your algorithm over a period of three to four months to see if it's profitable. Check to see how long you need to hold your investments for them to make money. Maybe it's a few hours, or maybe you need to hold them for a few days. Backtesting gives you an idea of the following:
* Is it profitable? Specifically, how much capital do I need to put in and what's my expected return?
* Can it scale? If I put in 2X more money, does the yield (return) also double? Or does it taper off? What about 3X? 4X? ..
* Algorithms are not perfect. Sometimes they lose money. What is the maximum amount of money I could lose at any point in time (called drawdown)?
* Is there a latency risk? If I'm not able to get my trade at the exact price, will it drastically affect my return?
Remember that no algorithm is perfect. You may not make money all the time. You don't have to. You just have to be right 51% of the time to make money.
Background: I've been writing low latency algorithms for over 5 years
1) Beware of any person/firm that promises returns. No one has a magic formula that can yield consistent returns. If that were the case, ask yourself why they wouldn't pump more money into it themselves instead of getting in the business of selling it to you.
2) There is no algorithm out there that prints money for you without you doing any work. No software/company will ever make money for you. Especially if someone says that the algorithmic trading strategy has been "proven to make money" -- run far far away.
If you're looking for stable returns without much effort, look up investing.
Do your own research and your own backtesting and find something unique that others haven't spotted. If the "strategy" is out in the wild, it's probably not going to work. Finding a trading strategy that works is hard, but not impossible. But it is hard.
Okay, now that's that. There are typically a few options for algorithmic trading that a retail trader can use. Algorithmic trading is a blanket term used for any trading that's based on signals. Signals are usually market events (news, price changes, indicators, etc...) that trigger an action (buy, sell, hedge, etc...).
Algorithmic trading can be as simple as waiting for a stock to hit a particular price and telling the computer to buy it automatically. It's nothing more than just creating a series of conditions and then attaching a trigger at the end of it.
There are a lot of programs that can help you do basic algorithmic trading. You can search the Internet for names or ask an electronic broker for guidance. With most trading software, you can configure it to watch for certain conditions and then trigger an action when the conditions are met. That's called your "algorithm."
Now the next question most people ask is -- okay, but how do I know if I'll make money? Here's where backtesting comes in. Backtesting is just taking your algorithm and running it across several weeks or months of market data. Test run your algorithm over a period of three to four months to see if it's profitable. Check to see how long you need to hold your investments for them to make money. Maybe it's a few hours, or maybe you need to hold them for a few days. Backtesting gives you an idea of the following:
* Is it profitable? Specifically, how much capital do I need to put in and what's my expected return?
* Can it scale? If I put in 2X more money, does the yield (return) also double? Or does it taper off? What about 3X? 4X? ..
* Algorithms are not perfect. Sometimes they lose money. What is the maximum amount of money I could lose at any point in time (called drawdown)?
* Is there a latency risk? If I'm not able to get my trade at the exact price, will it drastically affect my return?
Remember that no algorithm is perfect. You may not make money all the time. You don't have to. You just have to be right 51% of the time to make money.
Background: I've been writing low latency algorithms for over 5 years