A look behind the scenes at an index fund with Vanguard's Gerry O'Reilly
bloomberg.com
bloomberg.com
[0] https://www.bogleheads.org/wiki/Bogleheads%C2%AE_investment_...
EDIT: I'd still love for Vanguard to create Betterment's version of RetireGuide, which takes into account your income and various other financial parameters, and tells you how much to save in which accounts (tax deferred, 401k, etc). I'm a sucker for sexy UX.
I've always liked that idea as an alternative to doubling down with entrepreneurship. There's always the constant internal debate between both paths. It's nice to have a good guide to map out how to accomplish it, as the investment path comes with much more certainty if you have the right amount of self-control.
More focused on the hows of the saving than the goal of it. Although there's some overlap it isn't the same type of community as something like Mr. Mustache Man that emphasizes a specific life-plan (in that case very early retirement).
It's a helpful community. Sometimes a bit doctrinaire about "rules of thumb" that are very useful for many people, but aren't exactly laws of the universe, but that's hardly the worst thing in the world.
Don't be a trader, but don't have a religious acolyte of someone's investment philosophy. Most quarters my portfolios are growing 2-3x than my contributions. It's a great feeling.
Anyway, investing a portion of each paycheck turned out to be too much repetitive work, so I wrote a cron job to do it for me automatically (more work, but more interesting work!). Then, I added automatic tax-loss harvesting once Betterment added theirs. Then, some friends wanted to use it, so I built a UI.
I'm thinking about releasing it to the public. To do that, I would have to become a registered investment advisor, but that's not a big deal - just taking a test and filling out some paperwork. Would anyone be interested in paying $10/month for this service? I'm currently working on setting up a marketing site explaining what this thing does: https://zenve.st
I am personally not willing to pay monthly, and definitely not willing to share brokerage credentials, with a SaaS version.
It basically follows the Bogleheads approach: https://www.bogleheads.org/wiki/Getting_started
You get roughly your age as the percentage invested in bonds, with some adjustments up or down for risk tolerance. For taxable accounts, the bonds will be VTEB (tax-free munis). For nontaxable accounts, the bonds will be VCIT/VWOB (corporate bonds / emerging market bonds). The stock ETFs are VTI (US), VEA (foreign developed), and VWO (emerging markets). Nontaxable accounts also get VNQ (real estate), based on how much real estate you already own.
For example, my taxable account is: 60% VTI, 18% VEA, 12% VWO, 10% VTEB
Tax-loss harvesting is a bit tricky. In order to tax-loss harvest, you have to sell one ETF and buy another correlated ETF. This is usually done by purchasing another company's ETFs (ex: Schwab). Unfortunately, while Vanguard charges no fees for its own ETFs, it does charge fees for others' ETFs.
The algorithm takes this into account though, so it only initiates a harvest if the tax refund you'd get is significantly larger than the cost of buying the non-vanguard ETF. In order to make this cheaper, VTI is paired with VOO - even though the index tracked is different, they are highly correlated with each other (>99%).
VTI (Total US Stock market) / VOO (S&P 500)
VEA (FTSE Developed All Cap ex US, 3735 stocks) / SCHF (FTSE Developed ex-US, 1471 stocks)
VWO (FTSE Emerging Markets All Cap China) / SCHE (FTSE Emerging Index)
The Bogleheads forum is also often very useful for getting your personal finance questions answered, though of course the usual Internet stranger disclaimer applies.
The information you may find about tax-loss harvesting gains on the internet is usually incorrect if it comes from people trying to sell you something. For example, Betterment / Wealthfront claim that it adds an extra 1% of returns (only if you have a $100,000 portfolio and your marginal tax rate is 33%, which are the assumptions they use to get that number). On the other hand, human investment advisors are generating FUD about tax-loss harvesting[1][2], because they want to discourage people from requesting that service.
[1] http://www.cnbc.com/2014/10/24/weighing-the-pros-and-cons-of...
[2] https://www.kitces.com/blog/is-capital-loss-harvesting-overv...
https://support.wealthfront.com/hc/en-us/articles/209353226-...
The securities appeared in my Fidelity account, and the cost basis information populated a few days later. It also doesn't trigger a taxable event, and there were no fees for transferring out of Wealthfront.
0.25% a month would be really really high!
Disclosure: I work with numbers at WiseBanyan. Our business model charges a la carte for extra services (tax loss harvesting, for example), so you basically pay for what you actually want.
[1] You must stay invested for 30 days.
If you have enough money (typically $10k) to invest, then you can buy the Admiral Shares of the mutual fund, which has the same expense ratio as the ETF, but won't require you to pay a bid-ask spread to buy or sell.
If you start with the Investor Shares and end up with more than $10k in the fund eventually, you can always convert them up to Admiral Shares without tax implications. You can't convert ETF to Admiral Shares without selling though.
If you don't think you'll have $10k in the fund anytime soon, then the ETF is likely the cheaper option. However, the difference is practically really small - for a $9000 investment, you'll pay $4.50 per year in expenses for the VTI ETF vs. $14.40 for the VTSMX fund.
[1] https://advisors.vanguard.com/VGApp/iip/site/advisor/investm...
The conclusion most of my colleagues have come to is the roboadvisors don't really offer much beyond Vanguard target date funds. The roboadvisors seem to be failing to acquire significant assets.
Not much of a difference, but it was enough to prompt me to shift my auto deposits from one to the other (still have stuff in both).
The institutional funds that some 401ks carry can be cheaper than what's available to the general public. I'm pretty sure the Vanguard S&P 500 fund in my 401k is at 0.03%. The government employee Thrift Savings Plan's C fund charges 0.029% for an S&P 500 fund, managed by Blackrock, but I don't think you'll get that as a member of the general public. I'm sure you could find Fidelity and other funds in similar situations.
This is all somewhat relevant since those of us in the US can contribute many more tax-free dollars to our 401ks than our IRAs, sadly.
Some of Vanguard's funds have corresponding ETFs, which hold the exact same investments the funds do, and usually offer the low expense ratio of the Admiral fund shares with no minimum investment.
Perhaps that's why Schwab has a fee free robo advisor, and surely every other brokerage and bank is working on the same thing.
I'm guessing the dedicated robo-advisors will end up being the cheapest services, since they're online only and don't need to keep all those bank buildings and old guys in suits around.
They mention some reasons that human decision-makers are needed, but I'm actually not convinced. Why not just two algo indexes? One tries to get the best price, the other simply doesn't trade the difficult illiquid stocks?
I'm surprised an enterprising company hasn't come along that allows you to pick all your own stocks, and allow them to be rolled up in an index. That kind of service shouldn't be too hard to create from a technical point of view, so I'm assuming the issue is with regulation.
Moreover, how would you balance these two algorithms' holdings?
Regardless, there are a great many nuances, complications, and technicalities involved in trying to keep an index fund on-target. Not just the pricing and liquidity issues mentioned in this article. This is especially true for very large funds with hundreds of billions under management. Index funds are as close as one can get to full automation today, but taking the last step to full automation won't happen without fundamental changes in underlying systems like the exchanges.
It's not clear if it would even be advantageous to automate human decisions away completely here. Proposed automations often run in parallel simulations to a manager's work and are plugged in when they seem empirically to do better than a human manager can. Obviously that hasn't been the case across the board just yet.
Do you mean basket trading? https://www.interactivebrokers.co.jp/jp/?f=/en/trading/order...
The grandparent post seems to be calling for a service that lets me say, "I'd like to start my own custom-ETF, consisting of A% Facebook, B% AT&T, C% Walmart, D% McDonalds, etc." and then let me invest money that "fund" every so often. I agree this would be a neat service. Currently, if you want a portfolio of specific diversified stocks, you have to either 1. Find a fund that is similar enough to the basket of stocks that you would like or 2. Purchase each stock individually (getting eaten up in commission fees). Both are less than ideal.
Running an index fund is largely mechanical, but the main issue is dealing costs.
What Vanguard have to do is seemingly impossible, which is to deal at the mid-price every day, when they invest net inflows (or dis-invest the net outflow).
They also have to deal at the mid-point whenever the index changes (usually every quarter).
This is what Motif does, but it's super expensive, and picking stocks is a terrible idea. Vanguard/Betterment are cheap because everyone owns the same thing and you can trade between customers for free.
There was a paper about this a year or two ago which I'm unable to find now. I'd caution risk for anyone planning on using ETF's to ride out a storm.
EDIT: I'll have to see if I can find the paper again. The basic summary was that in situations like 2008 ETF's failed in ways that were materially worse than holding the same basket of assets.
But this is not investment advice. Make up your own mind and plan according to your risk tolerances.
I'd caution anyone from parking their money anywhere else but Vanguard index ETFs.
[1] https://en.wikipedia.org/wiki/Contango
EDIT
To add relative context from the link:
A crude oil contango occurred again in January 2009, with arbitrageurs storing millions of barrels in tankers to profit from the contango (see oil-storage trade). But by the summer, that price curve had flattened considerably. The contango exhibited in Crude Oil in 2009 explains the discrepancy between the headline spot price increase (bottoming at $35 and topping $80 in the year) and the various tradeable instruments for Crude Oil (such as rolled contracts or longer-dated futures contracts) showing a much lower price increase.[10] The USO ETF also failed to replicate Crude Oil's spot price performance.
[1] https://www.oaktreecapital.com/docs/default-source/memos/201...
Managing the S&P 500 fund is actually pretty tricky since everyone knows when the index changes and they know what trades you have to make so they try to take money from you. You have to employ a lot of tricks to avoid being victimized as an index fund.
Edit: I should point out that the index isn't really _just_ the largest 500 companies; they try to keep it properly weighted by sector and some other considerations as well.
I'd agree that managing an ETF is probably harder than most people think, but its very debatable as to if ETF's are being "victimized". Like the linked article says, if you want to buy large chunks of a company then you have to pay for it.
http://kiddynamitesworld.com/where-bloomberg-discovers-that-...
Which leads to my question--is that extra work relatively minimal or significant and hard to produce?
A simple example: lets say company X and Y are both on the index and they decide to merge one company X on such an such a date. In order to keep 500 stocks in the index S&P has to choose something to replacement. You don't have to be very smart to see that X and Y are in talks about merging months in advance. It isn't hard to figure that the stock S&P chooses will be one of a few: Buy a bunch of them each at today's prices. When the merge happens all those index funds now buy the replacement stock which drives the price up, you have shares at an inflated price to sell to those funds. (You have figure out when to sell the stocks that were not added - but S&P is unlikely to add a bad stock so as an investment these are likely to do okay). This is easy if the index funds try to hold exactly what the S&P 500 has in it.
Because of the above index funds promise to mirror the S&P500 results, but not the actual stocks. It is easy to say mirror results, it is much harder to pull that off, even if we ignore those above who are trying to cheat you, index funds tend to be the largest funds (because they do so well they are popular!) which means most of your trades will affect the price of the stock.
There is a lot more involved (much of it that I don't know), but the above is a simple example that will get you on the right track of thinking.
The former indicates that you must get your order filled at the closing price, the latter allows you to specify a limit at which you'll go to, in order to trade at the closing price.
How else would funds that need to trade at the closing price, actually trade at the closing price?
Now you may not like the closing price, but that's another story.
> The closing price is the just the price that marked the last trade of the day, and isn't otherwise special.
I guess i should also point out that many ETF's mark their value based on the closing prices of the NYSE or Nasdaq making their closing prices very important.
I'm pretty sure at this point I'm being trolled but just in case you are being serious and don't know how to google......
> No you can't. You can only buy a stock based on the set of offers to sell that stock
First you said that you can't buy/sell at the closing price and I showed you that this is wrong. Infact there are a whole lot of people who are obligated to trade at the closing price
Then you said the closing price wasn't special. This as well was shown to be false as many ETF's are marked by the closing prices of certain markets, notably the NYSE.
Then you said you aren't guaranteed to get the closing price. This was an even stranger error as the message before I showed you the MOC order which indicates that you will trade at the closing price.
Finally you tripled down on this mistake by changing your argument that technically there might not be any trading partner at the close.
This is technically true that in theory it might happen, but in practice I'll sit and wait for you to find me a case where this happened.
I mean think about it. You put out an MOC order that indicates you need to get a buy order filled. Unless you are trying to buy some crazy amount of shares, you will get filled. depending on the price the closing price can move up to 10% or 20%. This means that an arbitrager can pick up shares on the cheap or sell them for an artificially high price.
The system just works.
I get that you don't work in finance but you continue to keep making the same false statements over and over again:(
> kgwgk: But you can trade at the closing price, which is the one used by S&P to do their numbers, can't you?
> readams: No you can't. You can only buy a stock based on the set of offers to sell that stock
readams is unequivocally correct here. You can attempt to trade at the closing price, but doing so is entirely at the mercy of whether or not enough open offers exist to sell that stock.
> chollida1: Then you said the closing price wasn't special. This as well was shown to be false as many ETF's are marked by the closing prices of certain markets, notably the NYSE.
This point by readams is also true. The closing price of a stock isn't special — it's, as (s)he said, simply the price of the last trade of the day. The fact that some ETFs are marked by the price of market close makes their trading price "special", but that doesn't imply anything particular about the closing price of the asset(s) they're based on.
> chollida1: Then you said you aren't guaranteed to get the closing price. This was an even stranger error as the message before I showed you the MOC order which indicates that you will trade at the closing price.
Again, readams is correct. You're guaranteed to get the closing price if and only if there are open orders at that price. Which is a big if and only if, and not actually a guarantee.
> chollida1: Finally you tripled down on this mistake by changing your argument that technically there might not be any trading partner at the close. This is technically true that in theory it might happen, but in practice I'll sit and wait for you to find me a case where this happened.
Consider the context of this entire conversation: index funds that manage billions in assets. If you don't think they can utterly exhaust open buy/sell orders at market close for any trade they need to execute to track their index, you've lost your mind. This whole thread was about how these funds have to strategically place their orders to track the underlying index as faithfully as possible, without losing their shirts to vultures who know that large funds have to execute certain orders to stay on track.
The closing price is special -- there's a special procedure to set the price and determine which orders execute (quite similar to the opening procedure), NYSE has a closing auction, NASDAQ has a closing cross, I'm sure most other exchanges have similar.
If there's no market/limit on close orders for a given stock, then the closing price would be the last trade; presumably the same for a stock which had its trading halted earlier in the day.
Wikipedia says:
Limitations on types of securities:
> Securities that are ineligible for inclusion in the index are limited partnerships, master limited partnerships, OTC bulletin board issues, closed-end funds, ETFs, ETNs, royalty trusts, tracking stocks, preferred stock, unit trusts, equity warrants, convertible bonds, investment trusts, ADRs, ADSs and MLP IT units
Limitations on exchanges:
> The securities must be publicly listed on either the NYSE or NASDAQ.
And of course the actual selection criteria is even more elaborate.
On a slightly related note, I am working on the trading part mentioned in the article - i.e. APIs for advanced methods of getting the trades done. Basically, these trading methods are use machine learning to predict near future movement and use that to optimize the orders. (Some more details here - https://www.qleap.co/ - in case, someone is interested.)
http://www.mrmoneymustache.com/2012/01/13/the-shockingly-sim...
Oh, is that all?! ;)
Say I have $100 mn under management at an index fund.
When the markets close on Thursday, I have a 5 mn weighting in Apple and its market capitalisation is equal to 5% of the S & P 500 index.
On Friday morning, Tim Cooke makes an inspiring speech about the iCar and Apple's market cap rises to 8% of of the S & P index.
I want to sell some of my other positions and buy Apple so that I still have a balanced weighting of S & P 500 shares. However, I face a tradeoff.
If I do this in small bursts over the course of the day, I have a higher administrative burden. I also risk running up bigger transaction fees if Elon Musk makes another speech in the afternoon unveiling the Tesla-phone, forcing me to sell again after Apple falls back down to 5%.
If I wait until 3.59 to place a massive sell order for all my other stocks and buy order for Apple, (a) market participants might think I know something about Apple, causing the price to rise yet further as I struggle to complete my order (b) sneaky traders could buy up Apple Stock at 3.55 and then sell to me at 3.59 for a higher price.
This is the balancing act a passive fund manager faces.
Edit:
As comments point out, this example is incorrect. First example should say something like Apple splitting into Apple and iPhone and then the relative prices changing after the split. The second half still makes sense.
O’Reilly [..] came to the U.S. in 1983 to attend Villanova University
on a track scholarship. As a junior, he ran the mile in less than four minutes
—a feat he repeated six times. He led Villanova to a conference championship
and represented Ireland at the 1988 Summer Olympics in Seoul.
A few years later, O’Reilly joined Vanguard.
And when indexing with a bond fund, do you really want to be buying negative yield bonds?
The majority of the "income" from government bonds has come from capital gains, not the coupon.
Then the yield decreases it is because the bond itself has become more expensive. Sell the bond at the higher price.
Negative yield is an annual rate.
Taking a short position is more expensive.
Holding a CDS is more expensive.
If you expect negative yields to continue or go deeper, then bonds are a good investment.
Now, if you are in more illiquid corporate bonds, then you may be forced to hold them till maturity.
You're better off putting your bond allocation in high yield savings accounts.
That's the thing about index/robo people that I don't get. The magic formula can do dumb things.
Take a look at Vanguard's bond fund (BND). It went from $80.5 to $84 since January 2016. That's a 4.3% return over 8 months and it wasn't because of the interest the bonds pay. It was also at ~$84 back in 2013. As the economy improved the price dropped then went back up again.
Basically, bonds provide a way to diversify to an asset that is not correlated with equities. Investing in bonds reduces your overall return (vs. just equities), but it also reduces the risk level. If you find that sweet spot you can optimize for slightly reduced returns at a much lower risk. This is standard Boglehead theory.
Folks who know more than me can correct my explanation.
There are other answers to your question, of course, depending on the situation. Munis or treasuries will have different tax advantages and so on, while the money your high (ahem) yield savings account pays will probably be taxable income.
You probably know all this, just thought it worth pointing out that all of this stuff can be used sometimes for a good reason...
[0] https://www.theice.com/publicdocs/endex/ICE_Endex_Trade_Canc...
(Edit: Ok, I'll accept the downvotes. I mean not as much work as other fund managers. eg don't have to select companies to invest in.)
Vanguard has invested heavily in technology. Algorithms help managers figure out where to buy and sell while minimizing market impact. Risk software makes sure the portfolio stays close to the index. Because of these technologies, Brennan says, “we probably have the most assets per head of any asset manager in the world.” (The key determinant of staffing, he says, is the number of portfolios, not the assets in them.)
But you can automate only so much, and the tough decisions still fall to the portfolio managers. When is the best time to buy an illiquid stock that trades only once or twice a day? How do you handle a corporate deal structured in an unusual way or the issuing of a new class of stock?
“People think a computer could run index funds—and they’re so wrong,” says Brian Bruce, a former index fund manager who’s now chief executive officer of Hillcrest Asset Management in Plano, Texas, and editor-in-chief of the Journal of Index Investing.