Dai's Blog

Burritos

2018-07-03T00:00:00Z

While browsing a beach to visit for the 4th, came across this impressive github project for burrito ranking. It has interactive data via tableau and a very impressive ranking system. Will be a great resource when we move out to San Diego.

Happy Fourth!

Crypto Landscape

2018-06-23T00:00:00Z

This is a pretty good overview of the different projects in cryptoland.

Anthony Bourdain

2018-06-08T00:00:00Z

A sad day but inspiring to see the breadth of people touched by this man and the impact he left on this earth. Rest in Peace.

Mary Meeker 2018 Internet Trends Report

2018-06-07T00:00:00Z

Finally got around to reading the 2018 report which was good reading like always. The headline in the news was that internet penetration has now reached ~50% globally and mobile growth is stagnating (0% growth in new smartphone shipments). However, what was most interesting to me was the focus on China. Today the US has 11 of top internet companies by market value while China has 9!

It will be interesting to see how that list looks like in the next 5 years but my guess China will take > 50% of the top 20 and you will see at least 2 China companies in top 5 (US still takes all top 5 spots with AAPL, AMZN, MSFT, GOOG, FB).

Some points on China:

Government is very supportive of collecting user data and sharing this data. Users are also less wary and more open to it if they get a benefit.
Portals are very feature rich (ranges from news, to social to payments) leading to centralized services. For mobile payments which is growing at a ridiculous pace (reaching ~16 trillion in 2017), Alipay controls 54%, Wechat controls 38% and only 8% falls in other.
China's Mobile Data usage grew at 162%. Mobile Video was fastest growing segment (short form video similar to vine)
Innovation is spreading from online to offline shopping. My guess is similar to how Amazon is spreading to groceries via Whole Food purchase, you will see someone like BABA/Tencent dominate the retail experience in China.

NBA Finals

2018-06-06T00:00:00Z

Tough loss for the Cavs. They played tough and their hearts out but when KD shoots lights out like that, its pretty impossible to defend.

Learning Chinese

2018-06-05T00:00:00Z

今年我要练习我的中文。我每周都打算用中文写一篇文章。现在我正在使用 Duolingo 也读纽约时报的中文版。

Home Automation

2018-06-04T00:00:00Z

As a new parent, you'll find your kids want to watch the same videos/listen to the same songs over and over and over. For us, this has jumped between Moana, Frozen, Baby Shark, Lion King, StoryBots and Baby Einstein.

Being able to play these videos/songs on demand via voice has been a huge help. Below is what I used to make my life easier.

Hardware:

Amazon Dot - Very beneficial because of the Amazon music library included with Amazon Prime. I use Dot which has a 3.5mm jack to connect with my receiver. Has a number of other Alexa skills you can enable via app. Can also link to IFTTT via trigger command (I've used this to track babies feeding/poop etc).
Google Mini - Mainly useful right now because of ability to directly say which youtube video you want to play (Alexa can't do this)
Logitech Hub - Relays set of commands to your TV, receiver, etc like a universal remote except works with your phone via app and also with Alexa/Google.

Software:

https://github.com/rg3/youtube-dl - Downloads youtube videos to your computer. Useful for skipping ads. Also if your kids like a particular song unavailable on Amazon music, you can extract audio to mp3 and upload to your music library on Amazon to use with Alexa (youtube-dl -x --audio-format mp3 [link])
https://github.com/xat/castnow - You can cast any home videos in mp4 and downloaded youtube videos to your chromecast.
(Optional) https://github.com/sushilks/harmonyHubCLI - Command line interface for your Harmony Hub. Useful for really lazy people that want to bundle 3 voice commands into one bash command.
(Optional) Termius for IOS - Allows you to ssh into your server from your phone and issue commands from your phone.

That's it, hopefully this was useful for some new parents!

On a separate note, it was announced that MSFT acquired Github today. I really hope this doesn't lead to less projects on github else we'll need a search program for git projects.

Solar Panels

2018-06-03T00:00:00Z

Owning a house in San Diego, one of the popular things to add are solar panels. Plenty of sun and lower solar costs has resulted in wider adoption and California will actually require all new homes to have solar panels in two years.

Solar's growth has been tremendous (~30% in 2017, ~49% in 2016). China is the dominant player adding ~53 GW of power in 2017 (~53% of global total) vs US adding ~11GW of power in 2017. With costs coming down drastically in the past few years, it's become more "in the money" to switch to solar. Government has has also helped. The US has a 30% tax credit for and China has provided generous subsidies.

Getting quotes now but estimated usage for us is 15,000 kwh per year which at ~20 cents/kwh (SDGE retail rate) would cost ~3k per year.

Installing panels to offset that would cost ~22k after tax credit so breakeven would be ~7 years. Assuming constant 3k of electricity savings per year and 0 residual value on the solar panel after 25 years, irr of installing panel is ~13%, not a bad return on capital for 25 years.

San Diego Housing

2018-06-02T00:00:00Z

Since we are moving to San Diego in July, figured would play around with Zillow data to check out the market in our zip code.

Some metrics that I pulled:

Inventory Seasonally Adjusted: I pulled inventory both for greater SD and our zip code. Both show the story of rapidly shrinking inventory.

In [105]:

fig, ax1 = plt.subplots()

ax2 = ax1.twinx()

ax1.plot(inventory, 'g')
ax2.plot(sdinventory, 'b')

ax1.set_xlabel('Year')
ax1.set_ylabel('92130 Inventory', color='g')
ax2.set_ylabel('SD Inventory', color='b')

plt.show()

Prices: Small available inventory leads to rising prices. I pulled the Zillow 5BR Home Price Index and looked at the median sales and listing prices on a PSF basis.

In [166]:

ZillowHVI5BR = q.get('ZILLOW/Z92130_ZHVI5B')

In [167]:

plt.plot(ZillowHVI5BR)
plt.show()

In [66]:

psf = q.get('ZILLOW/Z92130_MSPFAH')
medianlistingpsf4br = q.get('ZILLOW/Z92130_MLPF4B')
medianlistingpsf5br = q.get('ZILLOW/Z92130_MLPF5B')

In [141]:

plt.plot(psf, 'r') # Median Sale Price PSF (All Homes) 
plt.plot(medianlistingpsf4br, 'b') # Median List Price PSF (4BR) 
plt.plot(medianlistingpsf5br, 'g') # Median List Price PSF (5BR)
plt.show()

Rental Yield: Zillow has a price to rent Index (Home value divided by 12 times its estimated monthly rent price.) We plot the inverse of this to look at the rental yield (or cap rate in RE lingo).

In [142]:

pricetorent = q.get('ZILLOW/Z92130_PRRAH')
caprate = 1/pricetorent

In [148]:

plt.plot(caprate)
plt.show()

Girl Like You

2018-06-01T00:00:00Z

Happy Friday! Loving this music video today.

2018 Market Outlook

2018-05-31T00:00:00Z

So with a rough start to 2018, I've had a number of people ask me where they should put their money. Here are my thoughts on general asset allocation for 2018.

Sectors:

Cash (30% allocation). Warren Buffet calls it a "indefinite call option on all other asset classes." Thus per option valuation, higher volatility in markets should translate to higher value of cash.
US Equities (10% allocation): Ever since the great money printing started in 2009, US equities have been propped up by a low rate regime. That is starting to turn as the fed is hiking rates. Inflation fears are rising and trade wars / isolationist policy only add fuel to the fire on higher prices. This is why managers are generally cautious on the equity market. The time value of money is the simplest and most important equation in finance and rates are the most important factor on asset valuation.
China (20% allocation): China continues to chug along at 6~7% GDP growth (vs ~2% in US) and the Chinese consumer tech scene continues to impress. I think BABA and Tencent are well positioned as the main customer portal from shopping to social to finance and value should continue to accrue to those two.
Rates / Credit (10% allocation) US Treasuries still don't offer much compelling value but if 10 year rates get closer to ~5% (currently ~3%), I would start allocating there. Floating rate corporate loans are L+ 200~300 and if defaults ticks up to 4~5% (historical ~2%) loss adjusted yields are 3~4% so not super attractive either. Most of my exposure is short duration in TSI and Doubleline.
Real Estate (10% allocation): Lower inventory levels and rising household formation should be beneficial for overall HPA across the US but select luxury markets could have difficulty with higher financing costs / lower tax deductibility / fewer foreign buyers.
Precious Metals/Bitcoin (5% allocation): Gold has been a laggard on returns for years but demonstrates good hedge value against big equity drawdowns. It's amazing how everyone has heard of bitcoin and the million other altcoins by now. I think a 1~2% allocation makes sense.
Alternatives / Opportunistic (15% allocation) Generally PE and illiquid investments.

2018 Goals and 2017 Review

2018-01-08T00:00:00Z

My 3 Goals for 2018:

Get back to working out daily. After kids, my daily workout routine went out the window. Need to get back on it.
Improve my Chinese. Repeat of my old goal but again going to commit to working on it daily.
Write more. Will commit to writing an article every day, even if its a short one.

Review of my 3 Goals for 2017:

Keep better in touch with family/friends. This was somewhat accomplished. I tried using social tools like FB, Whatsapp, Wechat, even starting a Slack channel for new parents but its still hard to replace face to face interaction.
Learn more about crypto. This was fairly successful. I was pretty active in reading and learning about the space.
Write more. Repeat Fail.

2017 Goals and 2016 Review

2017-01-04T00:00:00Z

My 3 Goals for 2017:

Keep better in touch with family/friends. After leaving NY and moving to LA and now CHI, I realized how the circle of people that I actually keep in touch with keeps getting smaller and smaller. I will make more of an effort to reach out.
Learn more about crypto. I've invested some amount in bitcoin but haven't spent much time with the protocol or other projects like ethereum. I plan to take the Princeton Coursera course and look into this area more.
Write more. Simply a repeat since I didn't write much in 2016.

Review of my 3 Goals for 2016:

Cook more. This was largely accomplished in 2016. I now have a rudimentary knowledge of what to buy at the grocery store (always keep garlic, ginger, green onion, and some Chinese vegetables on hand!) and what to put together to create a quick dish. One of my favorite books on cooking was The Food Lab. It was very scientific in its approach and helped me appreciate basic principles. Whereas before I would always just turn the heat to max setting, now I have a greater appreciation for heat at lower temperatures and different tools to use.
Learn more about the pharmaceutical industry and improve my Chinese. This was somewhat accomplished. I spent a fair amount of time learning about the pharma industry but eventually baby duties took over. Same thing with Chinese, checked out a bunch of vocab books but never managed to get through them all.
Write more. This was a failure. I didn't block off time to write and this blog went ignored.

2016 Cooking Tracker - Jan

2016-01-30T00:00:00Z

Starting with minimal knowledge on cooking, my gameplan was as follows:

Clear out the fridge
Start learning basic ingredients and how fast they spoil

Like starting with the prime numbers, I wanted to understand the building blocks before I jumped into random recipes online. There are handy charts such as this or this that provide guidance.
Set up system for buying, storing and cooking food

With a general understanding how fast items spoil, the next step was to develop a system to prevent wastefulness. The initial target is to do grocery shopping once a week and to buy the categories in the following amounts:
- Short Life Vegetables (buy ~3 Servings) Generally all the leafy greens. Plan to eat in next 2~3 days.
- Longer Life Vegetables (buy 2~3 Servings) Root vegetables, Carrots, Potatoes etc. Try to eat later in the week but ok if some go into next week.
- Seafood (buy 2 Servings) Eat these in the next 2~3 days.
- Meats (buy 2 Servings) Put in freezer and transfer to fridge 1 day before eating. Generally eat these later in the week along with the root vegetables.
As opposed to squeezing items wherever there was room in the fridge like I used to, I then designated sections for each food category. I had no idea what humidity drawers were for. This is how you should use them:
- Low Humidity Drawer: Apples, Avocados, Bananas, Kiwis, Mangoes
- High Humidity Drawer: Broccoli, Brussels sprouts, Carrots, Green beans, Herbs, Leafy greens, Strawberries

So that's version 1 of our cooking process. I think the key will be avoid the temptation of throwing everything that looks good into our shopping cart and stick with this system for the next few weeks. As we iterate, I'm sure we will have improvements.

Dishes tried this month:

Basa Fish Fillet w Pickled Cabbage. Chop fish and add salt (one teaspoon to two fillets), baking powder, rice wine, ginger. Stir fry cabbage first then add fish.
Black Chicken Soup. Black chicken, dates, ginseng (soak and chop), goji berries, long yan, salt.
Water Spinach w Fermented Tofu. Add garlic and 1 chunk of tofu.
Precut Kalbi w Collared Greens.
Steamed Lobster. Salt Water, steam for 15 min.
Baby Bok Choy. Garlic and stir fry.
Lotus Root. Peel and slice. Add 2 tablespoon rice wine, 1 tablespoon soy sauce, garlic and stir fry.
Wagyu Steak. Salt, pepper. Lower heat than ribeye. Sear on both sides for ~2min.
Nagaimo pancake. 1 egg, green onions, herbs, salt.

2016 Goals and 2015 Review

2016-01-04T00:00:00Z

My 3 Goals for 2016:

Cook more. Having lived in NY for 10 years prior to moving to CA, my fridge largely consisted of water and various alcoholic beverages. Since I'm going to be a parent soon, figure it makes sense to start developing this skill which has largely been neglected. The current plan is to try to learn one dish a week. Approximating a ~25% success rate, figure that will give me 10~15 go-to dishes by year-end. If anyone has any tips or advice for best practices, would love to hear them! Particularly interested in inventory management as I think we currently throw out ~50% of the vegetables that we buy right now.
Learn more about the pharmaceutical industry and improve my Chinese. I have been helping Cat's dad as the finance director for his generics pharma business this year so I need to learn more about the field and improve on my Chinese reading and writing comprehension.
Write more. This will help to formalize and document my thoughts.

Review of my 3 Goals for 2015:

Start a family. We are having twins in April :)
Spend more time with family. Visited my family in Harbin for Chinese New Year and spent a good amount of time with both our parents.
Have fun. Having a break year before parenthood, I took this year to do some backpacking in Europe, picked up boxing as a hobby, spent a bit more time programming, and took a babymoon vacation with Cat to Croatia.

2015 Market Recap

2015-12-31T00:00:00Z

A tough year for the investment community. We've had a number of good years from 09 - 14 and as the saying goes, all good things must come to an end. Thankfully, I was defensively positioned in the PA. I wrote to a friend on my FB post back in Jan:

In terms of investment themes, I do think we will be entering into a period of higher volatility which this book sort of indicates as well since power is more transient. We've just been in such a low vol environment for such a long time and recent shocks in oil, fx etc should trickle into other parts of the market. Timing is always hard on these things but I currently just have a slightly cash/gold allocation than normal. Think it was Howard Marks who pointed out years ago he was worried about social unrest after the economic downturn and I'm keeping an eye out for this as well which ties into one of the themes of this book.

Finance people love baseball analogies and some think we are now close to the 8th or 9th inning, heading back into a recession. I am not quite as bearish but would definitely bet on a higher vol regime and the VIX (30 day S&P vol) trading > 20 on average next year after trading in the teens in the past 3 years. I think downside liquidity could be severely tested in 2016. Yes, bank leverage has declined since the boom years in 05-06 and there is less levered carry vehicles like SIVs, CDOs, etc so arguably money is in "stronger hands" but trading banks have reduced risk / headcount as well and there can be poorly designed ETFs (the new investment vehicle of choice) just like there were poorly designed market value structures during 08 that exacerbated downward spirals.

A review of 2015 by sector and current allocations:

US Equity (22% allocation): A flat year. Tech, healthcare were the leaders while energy took a major beating. Think 2016 will be another tough year for equities as growth is constrained and we are in the later part of the cycle where multiples come down rather than up.
In [78]:
pc.plot(kind='bar', title='YTD Return (%)') plt.ylabel('Pct Change (%)')
Out[78]:

<matplotlib.text.Text at 0x8e0b710>
International Equity (6% allocation): With the eyes of the world on it, China had a big run-up at the start of the year then an equally disappointing collapse in the later half to close flatish. Everyone's worried about the China slowdown and there may be opportunities to add China's big consumer companies like BABA, BIDU, Tencent, JD at much cheaper levels this year. Europe continued its slow decline and gains in the ETF were mainly from the currency short.
In [83]:
pc.plot(kind='bar', title='YTD Return (%)') plt.ylabel('Pct Change (%)')
Out[83]:

<matplotlib.text.Text at 0x8e815c0>
Credit / Rates (17% allocation): Credit spreads widened with Loan ETFs down 4~5% and trading at ~10% discounts to NAV. On rates, everyone started the year expecting the fed to hike mid~late of this year. The Fed missed the Sept window but we did finally get the rate hike in December. While some like Marc Faber are calling this hike terrible in the face of a recession, I personally think it was due. The market has become accustomed to the Fed doing everything in its power not to rock the boat which distorts valuations when you keep rates low for so long and have the market expect a Greenspan/Bernanke/Yellen put whenever the S&P drops a few percent. 8% of my credit allocation is in a ETF comprised mostly of senior short-term floaters with minimal credit risk (TSI) that I would consider low beta.
In [86]:
pc.plot(kind='bar', title='YTD Return (%)') plt.ylabel('Pct Change (%)')
Out[86]:

<matplotlib.text.Text at 0x8e7da90>
Precious Metals / Bitcoin (1% allocation): Precious metals like gold and silver still had a terrible year despite the heightened volatility. Bitcoin, on the other hand, had a bit of a resurgence and conversation really starting picking up the underlying blockchain technology, in particular on its potential to transform clearing and settlements which takes weeks to months for some securities. I have high hopes for the space, just hope the current Bitcoin I hold isn't the Netscape Navigator of internet browsers.
In [89]:
pc.plot(kind='bar', title='YTD Return (%)') plt.ylabel('Pct Change (%)')
Out[89]:

<matplotlib.text.Text at 0x8d6e2b0>
Alternatives / Opportunistic (6% allocation)
Cash (48% allocation)

2015 was a tough year and would expect 2016 to be much of the same. Broadly speaking, expected returns across asset classes are still compressed and expected variance is high. To use a poker analogy, sometimes when the action gets too hot, the correct move is to nit it up and wait for better spots to put in your money.

Here's to a healthy 2016!

Vietnam

2015-11-16T00:00:00Z

I'm back from my trip to Saigon and it was exciting as this was my first trip to Vietnam. Besides being a huge fan of the cuisine (love the pho, banh cuon, banh hoi, banh xeo, banh mi, com tam), it was also very interesting to see the country in action. Made me a bit curious on if I should consider investing in the country.

PROs

Demographics Vietnam has good demographic trends for growth in the next 10~20 years. Using data from the World Bank, 2 charts signify this: 1. Low dependence from the elderly (Percentage of population > 65 is fairly low) and 2. Relatively high birth rate. Literacy rate in Vietnam is also very high (est 90~94% vs in the 70s for India).

In [1]:

%matplotlib inline
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

In [2]:

import wbdata

In [21]:

countries = ['VNM', 'USA', 'CHN', 'JPN', 'KOR', 'BRA', 'IND']

In [42]:

indicators = {'SP.POP.DPND.OL':'Age Dependency Ratio', 'SP.DYN.CBRT.IN': 'Birth rate (per 1,000 people)'}

In [43]:

df = wbdata.get_dataframe(indicators, country=countries, convert_date=False)

In [44]:

dfu = df.unstack(level=0)
sdfu = dfu[40:]
# sdfu.head()

C:\Anaconda3\lib\site-packages\pandas\core\index.py:4281: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison
  return np.sum(name == np.asarray(self.names)) > 1

In [45]:

sdfu['Age Dependency Ratio'].plot()
plt.legend(bbox_to_anchor=(1.05, 1), loc=2, borderaxespad=0.) 
plt.title('Age dependency ratio (Old/Working)')
plt.xlabel('Date')
plt.ylabel('Ratio of old (>64) to the working-age population')

Out[45]:

<matplotlib.text.Text at 0xa954dd8>

In [46]:

sdfu['Birth rate (per 1,000 people)'].plot()
plt.legend(bbox_to_anchor=(1.05, 1), loc=2, borderaxespad=0.) 
plt.title('Birth Rate')
plt.xlabel('Date')
plt.ylabel('Birth rate (per 1,000 people)')

Out[46]:

<matplotlib.text.Text at 0xaa027b8>

Hustle Ties into point 1 but the people in Vietnam work hard and hustle. I visited the Cu Chi tunnels by Saigon (farmland outside of Saigon where the battles took place) where I saw how resourceful the Vietcong had to be during the war to survive. They essentially were farmers by day and then dug tunnels and gathered/created whatever battle resources they could at night. Walking along the streets in Saigon, I saw many young adults hold multiple jobs for income. Vietnam must also hold the title of highest density of coffee shops in a major city. Literally, coffee is EVERYWHERE. All that caffeine must help :)
Trans-Pacific Partnership New trade agreement that should be positive for Vietnam.

CONs

Corruption Corruption is a big issue (as with a lot of developing countries). Sounds like every company has a government aka bribery department.
Credit Vietnam is a very cash heavy country. Credit cards are rarely accepted (supposedly ~10% penetration rate). However, smartphone penetration is very high (~40%) which provides opportunities for mobile payment services.
Infrastructure. A lot of buildings are hastily put together and apparently have a short shelf life in the hot and humid weather. Roads are majorly congested (crossing an intersection in Saigon is not for the faint of heart). As people get wealthier and migrate from mopeds to cars, this will only get worse so there is a definite need for mass public transportation in Saigon.

Overall, I am very excited at the potential in Vietnam. In a lot of ways, Saigon felt like Beijing 10~15 years ago. It'll be a interesting spot to keep an eye on.

Moving to Chicago

2015-11-03T00:00:00Z

Cat matched to UIC for fellowship next year and the family is moving over to Chicago (target living in the Loop for her work). With a new city comes the same old question: Buy or Rent? Let's look at the pros and cons:

PROs

Chicago has lagged the recovery. Below is the Case-Shiller graph of Chicago vs the 20 city composite vs NYC. You can see Chicago has lagged both NYC and the composite. Chicago is only up 34% since 2000 while NYC is up over 80%.

In [9]:

%matplotlib inline
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

In [10]:

caseshiller = pd.read_csv('caseshiller.csv', index_col = 'Date', parse_dates=True)

In [11]:

caseshiller.plot()

Out[11]:

<matplotlib.axes._subplots.AxesSubplot at 0xf2b390>

In [28]:

ratio = caseshiller["Chicago"]/caseshiller["New York"]
ratio.plot(title="Ratio of Chicago Index to New York")

Out[28]:

<matplotlib.axes._subplots.AxesSubplot at 0xd48cf8>

In [24]:

caseshiller[-1:]/100-1

Out[24]:

	Chicago	20-City	New York
Date
Aug-15	0.3413	0.8247	0.8116

Reasonable rental yields/cap rates. Scanning across properties in the Loop, it's still possible to get 5% cap rates vs it was hard even finding anything > 3% in NYC. Rents seem to be ~60% of what it costs to live in a similar quality place in NYC so there's some rent appreciation potential there as well vs I don't see how rents can keep going up in NYC.

CONs

Returns suck. Real estate returns have lagged S&P equity returns. The main benefit really is in the financing (low long term rates with put optionality). However knowing my general risk-averse nature, I doubt I would use much leverage or take full advantage of the put. If I wanted levered market exposure, I could buy S&P futures with a lower leverage ratio to account for the vol difference and still come out ahead economically.
Rates. Fed may raise rates next year which following a line of reasoning may lead to lower house prices. Not too concerned about this as the general correlation of housing prices to rates hasn't been that strong and I think other factors like inventory levels / household formation will dominate over rates.
Convenience. More things to worry about with owning.
It's cold out there. Despite favorable economic reasons of living in Chicago vs other cities, doubt Cat wants to live there long term.

So on a relative basis, Chicago seems to be a decent city to buy but on an absolute basis my feeling has always been biased towards renting vs buying. With that said, anyone know any good houses/brokers out in Chicago?

Hello World!

2015-11-02T00:00:00Z

So after abandoning my CS degree for the past 10 years, I figured I would get a bit up to speed on what the cool kids are doing these days. Here's what I looked at this summer.

PYTHON

The first step was to decide which "new" programming language to learn. I went on Github and looked at which languages had the largest number of repositories. This helped me narrow it down to Python and Ruby though there was a ton of other languages I've never heard of like Haskell / GO / Swift etc. I briefly tried out both, then chose Python mostly because of IPython/Pandas (see below) and because animals are cooler than jewelery.

But it doesn't end there.

Since everything is done via the web now and no one mails installation discs anymore (remember formatting those AOL floppies?), you now need to decide on a web framework for your language as well. Python for example has Django, Flask, Bottle, Pyramid along with some other smaller ones. I picked Django simply because it was the biggest. After that, you need to pick a database which is typically between SQLite or a relational DB like PostgreSQL or MySQL. I went with SQLite since it was the easiest to start with (your DB is a simple file) but then bumped into issues when deploying to Heroku as they have an "ephemeral filesystem" so I had to rebuild the DB in Postgres.

So have fun deciding between the 30+ different combinations possible and then hoping you don't have to change your mind later on.

IPYTHON

A big reason why I decided to go the Python route was because of IPython notebook/pandas library. IPython (now Jupyter) provides a way to run live JUlia/PYthon/R code in a notebook cell by cell (similar to Matlab) which makes it great for building things and playing around with data. Numpy and Pandas provide a great analytic library along with some pretty neat data manipulation tools (sequencing data, merging data with holes/NaNs in your dataset etc). For example, this is one of the notebooks I made to look at YTD Returns on a selection of stocks:

In [81]:

%matplotlib inline

In [82]:

import numpy as np
import pandas as pd
import datetime as dt
import matplotlib.pyplot as plt
import pandas.io.data as web

from pandas.tseries.holiday import USFederalHolidayCalendar
from pandas.tseries.offsets import CustomBusinessDay, CustomBusinessMonthBegin

In [83]:

today = dt.date.today()
bday_us = CustomBusinessDay(calendar=USFederalHolidayCalendar())
last_bd = today - bday_us
start_year = today.replace(month=1, day=1)

In [84]:

ticker_list = {'MSFT': 'Microsoft',
               'AAPL': 'Apple',
               'AMZN': 'Amazon',
               'GOOG': 'Google',
               'BABA': 'Alibaba',
               'YHOO': 'Yahoo'}

price_change = {}

In [85]:

for ticker in ticker_list:
    prices = web.DataReader(ticker, 'yahoo', start_year, last_bd)
    closing_prices = prices['Adj Close']
    change = 100 * (closing_prices[-1] - closing_prices[0]) / closing_prices[0]
    name = ticker_list[ticker]
    price_change[name] = change

pc = pd.Series(price_change)
pc.sort_values(inplace=True)

In [86]:

pc.plot(kind='bar', title='YTD Return (%)')
plt.ylabel('Pct Change (%)')

Out[86]:

<matplotlib.text.Text at 0x8e37ba8>

In [ ]:

DEVELOPMENT / DEPLOYMENT

I decided to develop locally on my Windows laptop. While it was ok to try things out, I wasted a lot of time downloading separate builds or setting PATH variables etc just to get things to work. When the instructions simply say "pip install X", you may end up having to do 5 extra steps for Windows users. As for my editor, I went with Sublime which I like very much though I did miss the dog icon in my old Crimson Editor.

GIT is another "new" thing that I completely missed in the past 10 years. It is basically a version control system that everyone seems to be using for code development today. This xkcd comic summarizes how I feel about git.

I basically read a handbook, memorized 5 shell commands (git init/clone/add/commit/remote) and treat Github as a free ftp server to store my code as opposed to actual version control. Maybe one day I will figure out how to use it properly.

For deployment, I used heroku. Python Anywhere was another free option but it only supports python whereas heroku supports other languages.

WEBSITE / BLOGGING

I've always wanted to keep a journal but I never took the time to do it. After trying out-of-the-box offerings like Wordpress / Tumblr, I decided to go for the full customization route and build my own. I needed a programming project for Python anyway.

I first started with Django which generates dynamic web sites (html on the fly). I used the official Django tutorial which is quite good. I never learned HTML / CSS before but with resources like Bootstrap / Stack Overflow, it was pretty easy to hack something together quickly. As the website progressed, I found this to be cumbersome as a blogging platform so I turned it into a travel journal instead and deployed it to Heroku here.

For a written blog, I decided to use a static html generator called Pelican which pre-generates the html so pages load faster. You simply create a Markdown document and can easily embed basically any content (even ipython notebooks!) via liquidtags. Pelican then magically creates the html pages which I used Github pages to host. You are looking at the result of that now.

So that's a quick summary of my programming adventure and the tools I've used along the way. Now here's hoping I continue updating this blog and don't just drop it after a month!