Equal-weight ETF allocation with automatic rebalancing | Hacker News Reader

Equal-weight ETF allocation with automatic rebalancing

quantopian.com

40 pointsfawce26 comments

The strategy equally weights the stock sectors in the S&P 500 using ETFs. In a rising stock market this would incur capital gains taxes in a taxable account. If you are saving regularly, you could direct your new investments towards the sectors that have underperformed to get closer to equal weighting without selling the outperforming sectors.

Very neat. After a brief foray into reading about HFT, I've started reading more about algorithmic (non-HFT) trading.

I'd love to hear if other HN readers are playing with algorithmic trading. I occasionally think quantitative trading would be fun because of the intersection of math/stats and CS. Of course, that's an "in the abstract" statement. Depending on it for my mortgage would be . . . different.

I do algorithmic trading, and write the underlying systems and algos, for a living, not high frequency, though I did once get to 200 messages a second:)

The stress is always there, though you become desensitized to it after a while, or you just quit in a stress induced rage:(

I'm Canadian so my knowledge is skewed to the Canadian markets.

My email is in my profile if you'd like to chat.

I'd feel VERY uncomfortable giving a 3rd party the password to my brokerage account. Too many bad things can happen.

Yes, I know you take property security. There's always some bug or flaw you don't know about. Because you're logging in to the clients brokerage account, that means your software has access to their UNENCRYPTED UNHASHED password.

Also, there are reasons to NOT automatically rebalance every 21 days. Long-term games are better than short-term gains (unless it's an IRA). In addition to commissions, there's the cost of the bid/ask spread, not included in your calculation.

Does it say somewhere that they store your IB username and password? Their documentation suggests otherwise.

You will need to authenticate to your Interactive Brokers account. (Note: Quantopian does not store your brokerage password.)*

https://www.quantopian.com/help#overview-livetrading

On the question of long/short gains, you are certainly correct to say there is a difference, but it's also likely that rebalancing will only trade a fairly small portion of your portfolio.

A neat way to do this would be to use your IRA only for the balancing portion. So suppose you have 100k "normal" money invest and 10k in an IRA. Allocate it all even (in both accounts), and when rebalancing needs to occur, do it in the IRA. The normal account stays unbalanced, but the IRA goes way out of balance in the other direction to make up for it. It'd be interesting to see how much "normal" money could be balanced by a $1 of IRA money, historically.

I don't see the fundamental distinction between this and a S&P 500 index fund. Instead of carrying out one rebalancing (at whatever frequency) there's now two rebalacning steps -- one at the sector level and then another one to account for market cap changes within the sector. The net result of either strategy looks to be an imperfectly realized approximation of the performance a continuously adjusted market cap weighted portfolio of those 500 stocks.

Is the claim that this is somehow a better approximation? Are there some tax loss harvesting benefits? What am I missing?

The 500 fund has sector weightings based on market cap. This one has equal weightings.

Yes, you are exactly right, that's what I was missing.

Is there any theory behind this sort of weighting or is just one of those things that someone figured was worth a shot?

Yes, the rationale is to maintain equal sector exposures across your portfolio.

That's tautological. Is there any reason rooted in -- for example -- portfolio theory to believe such a strategy will outperform market cap weighting on a risk adjusted basis?

Sector weighting equally is based on diversification. Market cap weighting will tend to bet more on winners (the market cap is going up), concentrating your investment there.

Another way to think about it: market cap weighting is a momentum strategy. Equal across sectors is mean reversion.

In the broader topic of portfolio management, equal weighting will outperform cap weighting due to the small cap effect (read up on Fama-French three factor model). tldr: historically small caps outperform large caps.

In this case, the equal weighting is abstracted away by 1 layer - each ETF is already cap weighted, so you aren't getting the full effect of equal weighting. Instead, you're reallocating to the ETF that has underperformed relative to all others, so as already pointed out, you are betting on reversion rather than momentum.

>equal weighting will outperform cap weighting due to the small cap effect

Just because the sector is big doesn't mean the companies in that sector are big.

The small cap effect has nothing to do with sectors.

If the sector is big because it has a lot of medium/small sized constituents you gain from this implementation because a cap weighted index would under allocate to these smaller names.

The costs of this strategy are hidden in the trading. Every time you trade, you incur a cost. Lets say your brokerage commission is $0 - you still pay the spread when you trade and taxes when you sell. Fees for this look reasonable, but why not hold VTSAX which holds approximately 3700 stocks at 0.05% (5bp) expense a year.

They claim that their backtesting mechanism accounts for the spread. Your guess is as good as mine whether they do so perfectly.

The rationale for equal-weighting is that cap-weighting will have you over-invested in stock sectors that are priced too high. If tech stocks have a huge run and double their weight in the S&P, do you want them to represent twice as much of your portfolio? If you think the market is efficient you do, otherwise maybe not.

You are correct that it is basically recreates an S&P index fund, but so is the EQL etf, and people have put $125M in it and pay 0.5%/year. I think this is more proof of a) how to recreate something that is already out there for cheaper, and b) the framework for customizing something to your own investment strategy. Maybe you don't need all 11 ETFs under there, maybe you just want Tech+Energy+Fin and you equal weight those - they're giving you the framework you just have to customize it to your needs.

The strategy is not attempting to approximate the S&P 500, it is seeking to manage your exposure to these industries.

The algo author wrote a detailed explanation of the outperformance of the S&P 500 here (source code included): https://www.quantopian.com/posts/equal-weight-all-sector-str...

Out of curiousity does anyone do algo trading on options? Maybe through IBs java api?

How does it execute the rebalance trades? Just MOC or something like that?

No, it places orders market orders in any of the ETFs we need to rebalance mid-morning. Orders go out around 10:15am NY time on rebalance days.

Also, the chart in the page is from actual trading on a roughly $25k account, and updates each day.

It would be very interesting to see the chart of actual trading next to a similar chart generated by the Quantopian backtester.

Very impressed with the product btw. Good work.

That's a great question, I've been working up a more detailed analysis of implementation shortfall (and thinking about ways to make this automated) - but your question prompted me to post a quick comparison here that you can look at for some current results: https://www.quantopian.com/posts/live-results-vs-backtest-re...

Full disclosure, I work for Quantopian and I wrote this strategy.

Can't a counterparty anticipate your orders and make a risk-free return?

Reply on news.ycombinator.com