Hello, Newbie trader here, I have a few questions in my mind, can someone please help clear these.
I would like to know what advantages does an institutional trader has wrt a retail trader. do they have better tools which convey more information (like insights about buy/sell volumes or any extra data related to trades), what does their trade base on.
I heard some things like HFT, quant, algo etc . does these make sense for retail traders considering 1-2 lot size (intraday) and no or regular tools/platforms. If it makes sense, how does one implement them( any strategy), is it through some platform like zerodha through api (apart from streak) or need some special application/software and data feed or separate connection to data? and what is tick data used for?
Wouldn’t be able to comment much on your 1st point as I have not been with the Institutional side. But obviously they have a full-time team who works day in and day out.
High-Frequency Trading is done mostly by prop trading firms and has a good amount of capital. The idea behind HFT is to take the price advantage. One of the strategies could be to take an arbitrage benefit say buy on NSE and sell BSE when the price of same stock is different at a given time and difference is higher than the cost of transactions. Execution is more important here. Hence, HFT’s have their own server in exchanges so the time to execute could be minimal. Having a server is an expensive task. One need to lease the server and also pay for the data feed.
HFT’s doesn’t make sense for retails.
Quant/Algo - is sort of trading or investing wherein you build your strategy either by coding yourself or use some UI based tools. The idea is to test your strategy on historical data and let the computers do the execution for you.
Does it make sense to do an algo for retail - It totally depends on the need of the trader. If you are focusing on long-term investing you could use excel. But if you are an active trader than quant could help you not just in execution but also in building better trading strategies. Quant models can also be built for long-term investing.
Here are some of the resources if you are looking to build quant/algo’s model on your own.
Building infra:
This requires you to build a backtest engine, data management engine, order management. Your backtest engine should be robust enough to give the entire analysis on your portfolio such as returns, risk, money management, cost of transactions, slippage etc., You could skip few of the modules.
Data: Having the right data is of the utmost importance. You should make sure that the data are properly adjusted for various corporate actions such as dividend, bonus, split, merger, demerger. Also, if you are looking to trade in Derivatives you might have to check if the vendor provides historical data for expired contracts as well. So before selecting data vendor make sure what type of data they provide and type of adjustments they do. Also whether you need live tick level data or minute level, this would depend on the strategy you develop. For low latency trading minute level data should be sufficient. Further, the API request limit of the data vendor is also important. Generally, if you are not consuming much of data you could rely on data vendors to get live data every time you need. In case you consume huge data every second then you might have to save live data at your end. There are tools such as sql, postgress, mongodb to manage market data.
Data vendors: Zerodha kite, Global data feed, True data, Alpha vantage. These would cost you somewhere from 1500 - 5000 per month. If your requirement is just a daily candle data than Yahoo finance is the best tool and it’s free.
You need to check out their API’s for integrating (google it).
Cloud: Once you build your engine, to trade live you would require cloud services. AWS, Google Cloud, Azure, Digital Ocean are few. They also provide free machines for a certain duration. You could check out at their websites. Cloud hosting could cost from $10-$100 per month.
Building quant models.
Setting up the entire infrastructure is one challenge and building a profitable portfolio is another challenge. You could learn by your own experiments or learn from platforms like https://quantinsti.com/. If you don’t want to pay you could check https://www.quantopian.com/, where you could test your quant models, learning modules and a great quant community all for free. Or you could read Trading Evolved by Andreas, an amazing book with practical guides to build a quant model.
Selecting Broker:
Now you need a broker who has made API available to execute orders. The API could cost depending on the broker. Some Brokers have free API. Below are few brokers whose API are available:
Paid - Zerodha, Upstox…
Free - Fyers, Interactive Brokers…
Most of the brokers are working on to open up their API. Hopefully, this would be free across all the brokers in future.
After reading above if you have made up your mind to build your quant model, Kite Connect developer forum is the place to go. Nithin and Kailash have done a wonderful job in enabling Algo trading for retails in India.
In case, you feel it’s a too much of an effort and might take at least a year to build your quant model and not sure whether your model would make money than it would be good to try out UI based platform like https://streak.zerodha.com/, https://chartink.com/, https://inuvest.tech/.