Search
Talks
Lightning Talk: Algorithmic Trading for Fun and Profit
Edit
Summarized using AI

Lightning Talk: Algorithmic Trading for Fun and Profit

Sheng Loong Su • June 27, 2014 • Singapore • Lightning Talk

Algorithmic Trading for Fun and Profit

In this engaging talk delivered at the Red Dot Ruby Conference 2014, Sheng Loong Su dives into the realm of algorithmic trading from an engineer's perspective, focusing on how to develop automated trading systems using mathematical and computer models.

Key Points Discussed:

- Definition and Process of Algorithmic Trading:

- Algorithmic trading involves creating a computer program with a set of predefined rules to automate the buying and selling process in financial markets, such as stocks, forex, and cryptocurrencies.

- The trading process incorporates standard scientific methodology including defining objectives, formulating hypotheses, backtesting strategies, and optimizing outcomes.

  • Financial Data Acquisition:

    • The very first component necessary for algorithmic trading is financial data. Sheng highlights sources for data, ranging from free platforms such as Yahoo Finance to commercial data providers.
    • Demonstrations include downloading historical stock data in CSV format and JSON response data from APIs to track market sentiments.

  • Strategy Development:

    • Strategies entail analyzing market data to generate trade signals using various techniques. Sheng focuses particularly on technical analysis, showcasing the moving average as an example.
    • The process to calculate moving averages, both fast and slow, is described in simple Ruby code.

  • Trade Signals Interpretation:

    • The crossover of moving averages generates signals for trading decisions: a long signal indicates buying opportunities, and a short signal suggests selling or taking short positions.
    • The importance of managing cash flow, market positions, and understanding portfolio returns and risks is emphasized.

  • Execution of Trades:

    • A broker component is introduced, which executes trades by placing orders in the market. Real-world trading involves utilizing brokers' APIs to manage live investments.
    • Sheng outlines the architecture of the trading system, which is comprised of various loops for continuous data fetching, analysis, signal generation, and order execution.

  • Project Overview:

    • Sheng introduces his project, Coahoma, which serves as a proof of concept for the discussed algorithmic trading concepts. He encourages attendees to explore this project for practical insights into algorithmic trading methodologies.

Conclusions/Takeaways:

While exploring algorithmic trading involves a deep understanding of financial data and programming, it also opens new avenues for engaging with the financial world in both a leisure and potentially profitable manner. With the right technical knowledge and strategic implementation, developers can effectively leverage Ruby to innovate within the trading landscape.

Lightning Talk: Algorithmic Trading for Fun and Profit
Sheng Loong Su • June 27, 2014 • Singapore • Lightning Talk

From an engineer's perspective, algorithmic (or quantitative/systematic) trading is about developing a trading system that applies mathematical and computer models for making transaction decisions in the financial markets. This talk will explore the algorithmic trading process and how we can use Ruby for research and development of profitable stock trading strategies.

Help us caption & translate this video!

http://amara.org/v/FGY6/

Red Dot Ruby Conference 2014

00:00:19.320 hey hello Ruby is nice to see you here
00:00:23.949 so today I will talk about algorithmic
00:00:25.779 trading for fun and profit so this is me
00:00:28.419 my nation long I'm a developer investor
00:00:30.759 and just a bit of disclaimers just to
00:00:33.789 make sure that I wouldn't get arrested
00:00:34.600 by monetary authority of singapore yeah
00:00:37.690 so okay you have heard of different kind
00:00:40.059 of terminology of her you have heard of
00:00:42.039 a algo trading high frequency trading
00:00:43.929 mechanical trading so and so forth but
00:00:46.239 as far as this store is concerned I'll
00:00:47.500 just stick with this definition so algo
00:00:49.960 train is about developing a computer
00:00:51.999 program that has got a predefined set of
00:00:54.910 rules in it that helps you to automate
00:00:57.160 the process of buying and selling in a
00:00:58.899 financial market such a stock market why
00:01:01.120 could be 4x or something else bitcoins
00:01:02.920 for example just like some other
00:01:05.980 scientific process there are few process
00:01:07.660 involved so we have the familiar
00:01:09.280 objective and hypothesis like how much
00:01:10.900 one you want to earn and then you
00:01:12.160 develop training strategy you back task
00:01:14.200 your strategy with the historical data
00:01:15.670 and you measure and optimize if it is
00:01:18.580 good you deploy to production if it's
00:01:20.410 not good then you have to start from
00:01:21.490 step 2 again and if you want to build an
00:01:24.820 algo trading system their very first
00:01:26.770 thing you need is actually financial
00:01:28.180 data so i call this component of feeder
00:01:31.120 so what it does is that he actually
00:01:32.680 fetches financial data feed from
00:01:34.630 external sources so there are a few data
00:01:37.420 sources I'm actually a hobbyist so I
00:01:40.090 hope for the free one like yahoo finance
00:01:42.160 but let's see if you are Quan trader
00:01:43.980 maybe you have a little bit more money
00:01:45.820 you might want to subscribe to like
00:01:47.500 commercial data provider for more
00:01:50.320 precise data and this is actually how
00:01:53.380 the csv file looks like when you
00:01:54.850 download it from yahoo finance so this
00:01:56.860 is actually the closing prices for apple
00:02:00.250 on a daily basis and this is actually
00:02:03.580 the JSON response data that we get from
00:02:06.130 stock please api so you can actually
00:02:08.019 find out what people are talking about a
00:02:09.820 particular stop on their social network
00:02:11.530 whether people are feeling optimistic or
00:02:13.740 domestic about stock and I basically
00:02:16.110 used high fees gem because I need to
00:02:18.330 download slow data for a lot of our
00:02:20.430 stocks so it's good for me to do these
00:02:22.800 requests in parallel and depending on
00:02:25.050 the data format i use csv DJ sonu kogi
00:02:27.150 to pass the data and after we have got
00:02:29.790 the market data we have to develop a
00:02:32.160 strategy the texting and market data is
00:02:34.050 input and then we generate some trade
00:02:36.300 signals as output based on the rules
00:02:38.370 that we defined so the question is how
00:02:40.470 do we define the rules so well a
00:02:42.360 different kind of techniques some people
00:02:43.500 for fundamental technicals news and sati
00:02:45.870 moon it really depends on your style but
00:02:48.260 today I'm going to talk about technical
00:02:50.970 analysis so as you can see on a church
00:02:53.130 this is actually the dailies prices of
00:02:56.550 Apple for the past like two years so I'm
00:03:01.050 going to talk about this technical
00:03:02.520 indicator call moving average this is
00:03:04.440 kind of like shallower example in the
00:03:06.270 technical analysis so the red color line
00:03:09.030 is actually fast moving average over 15
00:03:10.860 days whereas the green color line is
00:03:12.570 actually slow moving average over 50
00:03:14.940 days and it's actually not that
00:03:16.740 difficult to program this in Ruby in
00:03:18.600 fact as you can see just a few lines
00:03:19.890 what you have to what you have to do is
00:03:21.750 simply just you know calculate the mean
00:03:24.870 of the closing data for the past X
00:03:27.870 number of days so if you're looking at
00:03:29.370 simple moving average on 15 days they
00:03:31.620 just calculate mean of top prices over
00:03:34.530 the past 15 days and lower eyes for the
00:03:36.750 slow one and once we have the data what
00:03:39.750 do we do we actually have to identify
00:03:41.730 the crossover so whenever the fast
00:03:44.010 moving average crosses above the slow
00:03:45.750 moving average in trading time we say
00:03:47.910 that the generator the program can
00:03:49.590 generate a long signal so what does long
00:03:52.230 signal means it means that the curve is
00:03:54.060 actually on uptrend so on and on the
00:03:57.660 other hand if it's a shot signals that
00:03:59.610 means that the curve is actually on
00:04:00.840 downtrend and what do I mean by long I
00:04:04.790 yea sorry Karen I stories from you but I
00:04:07.920 really like go beep up whenever I think
00:04:09.690 of long there I thought it is yeah it's
00:04:11.250 pretty cool so like okay so this signals
00:04:14.130 actually advice portfolio in making
00:04:17.130 trading decisions so when you have a
00:04:19.140 long signals what you have got to do is
00:04:21.270 that all you think that the store is
00:04:22.740 actually bullish you think that the
00:04:24.270 price may go up in the future so you
00:04:26.710 got to stop yet you may want to buy some
00:04:28.210 and if a signal is actually shot this
00:04:30.910 indicate that the price may go down in a
00:04:33.190 future so you may want to sell it you
00:04:34.509 haven't got or you may wanna sell it if
00:04:36.490 you haven't bought it and popular
00:04:38.110 actually keep struggle for cash your
00:04:39.729 market position as well as the holding
00:04:41.319 values and most importantly actually
00:04:43.990 helps you to assess returns and risks
00:04:45.819 because you have to know your profit and
00:04:47.560 loss as well as your risk I mean you can
00:04:49.900 use a few matrix like Sharpe ratio and
00:04:52.210 you calculate your downturn things like
00:04:53.919 that and the last components actually
00:04:56.349 broker is actually a piece of code that
00:04:58.270 executes rate by placing order in the
00:05:00.610 market of course in the back testing
00:05:02.110 environment you want to simulate the
00:05:03.550 market behavior but in life trading
00:05:05.259 environment you want to invoke your
00:05:07.539 brokers API to actually place the order
00:05:09.520 in the actual stock exchange and when I
00:05:12.280 order gets filled you have two more you
00:05:14.259 have to notify your portfolio to update
00:05:16.479 the position and and holding the game
00:05:18.699 and all these components are actually
00:05:21.610 why out by using to what I call the
00:05:23.919 loops there's actually an outer loop and
00:05:26.289 inside that there is actually an ester
00:05:28.150 loop and this piece of code can be
00:05:30.400 represented by this diagram hope that is
00:05:32.440 easier to understand so basically you
00:05:34.389 have got a feeder that gets the next bar
00:05:36.310 a bar it's actually an abstraction so
00:05:38.620 it's like a trading activity over a
00:05:40.810 period of time so if you're looking at
00:05:42.099 daily data so this is the trade the
00:05:45.039 prices for that day by if you you can
00:05:46.960 actually look at frequencies like
00:05:47.979 minutes even up two seconds and then
00:05:50.229 once you go to data are then your
00:05:51.250 strategy analyze market data generate
00:05:52.870 signals for a portfolio place order and
00:05:54.669 then you execute orders and up their
00:05:56.740 holdings and game and for all of these i
00:05:58.719 actually created this project coahoma
00:06:00.610 which is a proof of concept of why i say
00:06:02.620 just now and if you are interested you
00:06:05.169 can go there and far more about it and
00:06:06.669 thank you
00:06:12.980 you
SLS
Sheng Loong Su
1 talk
Explore all talks recorded at Red Dot Ruby Conference 2014
+20