Hacker News Reader

Comment by Veratyr | Hacker News Reader

Anyone interested in robo-investing might also be interested in Quantopian (https://www.quantopian.com/).

It's marketed as a platform that facilitates competition for the best trading algorithms but really, algos can be as complex or as simple as you want. A simple monthly portfolio rebalancer takes ~5 lines of code (handle_data is boilerplate required by the platform):

  def initialize(context):
      schedule_function(rebalance, date_rule=date_rules.month_end())
      
  def rebalance(context, data):
      order_target_percent(symbol('IVV'), .6)
      order_target_percent(symbol('MUB'), .4)
      
  def handle_data(context, data):
      pass

Plug that in, backtest it (it has a pretty comprehensive backtesting and research suite) and add your Interactive Brokers or Robinhood account details and you're off.

It's all free as well.

Note: I have no ties to Quantopian, I just discovered them after being disappointed by how little flexibility Wealthfront and Betterment offered me.

View on HN

What I would love is an open source implementation of the daily tax-loss harvesting strategies increasingly being offered. In addition to saving on fees, it would allow you to specify your own basket of stocks (and weightings among those stocks).

I'm far from an expert but as I understand it, the maximum deduction you can get from tax loss harvesting is $3000 (per the IRS: https://www.irs.gov/uac/Ten-Facts-That-You-Should-Know-about...). Does tax-loss harvesting actually increase performance significantly?

3,000$ per year for say 15 years is a good chunk of money saved, especially if you put those savings back in the market. Of course if your portfolio is huge you probably won't notice, but the average investor will definitely benefit.

$45k income is $20k in tax savings, which is not a whole lot over 15years.

you can claim at most a 3000 loss in a single year.. BUT

* you can offset unlimited gains with as many losses as you have - so that's the real value. * you can carry forward losses greater than 3000 to future years. so its not use it or lose it

Yes it's "offset up to 3000 against ordinary income per year, but offset unlimited capital gains", as far as I understand it. IANAFA.

Rebalancing your portfolio requires buying and selling shares. In most countries that will attract capital gains taxes, often affected by when you bought those shares.

If you care about tax efficiency -- which can severely deplete your on-paper returns -- then it's not really ~5 lines of code at all.

If you're looking for dumb-and-good-enough, delegating the fiddly rebalancing problem to someone with economies of scale isn't really so bad.

Rebalancing does not necessarily require selling shares - a different strategy is to rebalance at the same time that you want to add money to your portfolio by buying the shares necessarily for rebalancing. This would prevent the taxable event of selling.

Rebalancing is the service offered by Wealthfront and Betterment though. The same downsides of rebalancing should apply to them as well, no?

Quantopian won't be free forever. It seems the fees they are thinking of charging going into the future will mean that it only makes sense for active traders who derive value from Quantopian's market research and backtesting facilities.

I recommend using Interactive Brokers and coding your algorithm against their API directly, skipping the Quantopian middleman.

Quantopian's revenue model is to build a hedge fund and charge the fund investors returns/management fees. The algorithms in the hedge fund come from the Quantopian community. We work with the best algorithm writers on our platform, negotiate compensation, and then put their code to work.

We would be crazy to charge people to use our platform. We need thousands of algorithms, and charging for the platform would be one of the faster ways to kill our business.

Yes, you can code to Interactive Broker's API. But where would you get your free minute historical data for backtesting? Or corporate fundamentals data? Or the free IPython research environment? Or the community of 60,000 quants giving each other mentoring and advice?

I work at Quantopian, so you can imagine my answer to all of those questions.

In fairness, your (commendably candid) FAQ states the precise opposite "In the future we plan to charge for live trading, when you trade your algorithms through your brokerage account"

Time to go update the FAQ! The business plan changed in October '14 - evidently we missed that reference.

FAQ still reads:

Is Quantopian free? Quantopian's community and backtester is free for everyone to use. There will be a charge for connecting your algorithm to your brokerage. Pricing isn't finalized, but we're considering a flat monthly fee.

Looks interesting. Python...

Does it take into trading fees? I pay $9 per trade. Can you have fixed income instruments in your portfolio? Foreign stocks? Does it handle exchange rates? Does it account for dividends?

It's a platform, not a broker. At the moment it's totally free, and supports Interactive Brokers and Robinhood as backends (both very low commission).

It only supports equities at the moment, with futures on the way. It only trades the US markets at the moment, exchange rates aren't relevant and dividends depends on how you want them "accounted for".

The reason I mention fees is that if you rebalance frequently and have high fees the fees can kill your profit. Therefore when back testing any strategy one would want to account for fees. Similarly if you try do diversify using individual stocks you have to take trading fees into account and if you use ETFs there are also fees.

What you do with dividends also matters so ideally any algorithmic strategy you come up with takes that into account. For me not being in the US (Canada) currency exchange rate can at times dominate the returns.

There are a lot of newer ETF products on the market so it's going to be difficult to back test a strategy that uses ETFs.

At any rate, it looks interesting, I might play with it a little, but I don't think it works for me right now for investment purposes. I do my rebalancing yearly on a spreasdsheet :)

Quantopian has some built in features for modeling slippage, commisions, and dividends. These are described in the Help section: https://www.quantopian.com/help#ide-slippage

Reply on news.ycombinator.com