Debunking the Myth of Dollar Cost Averaging
alexmolas.com
alexmolas.com
If I give you this once in a lifetime trade: 100k for a 1 in a 1000 chance to win 500M, would you take it? There are very few people who would, even though it has 400k of expected returns, a whopping 400%.
Most of us simply don't make enough money in a life to take that trade enough times to cope with its volatility. Elon Musk should probably take that trade though.
Volatility is also a lot easier to read and pontificate about than to experience in real time. What people do during draw downs is often emotion driven and for people who have never been through it, it's hard to infer from textbooks.
And there have been periods where the market has been down for near a decade. If you put a lump sum at the 2000 peak, you wouldn't be back to even until the 2007 peak.. at which point it promptly crashed and didn't make back to even until 2013. (+/- a few years when you factor in dividends). Similar for 1973/1980/1982.
Slow grind downs like the GFC are harder to experience as you watch slow losses pile up day after day for near 18 months.
A good way to respond to your example would also be to bring in a discussion of the St. Petersburg paradox and expected utility theory.
The use case for lump sum and dollar cost averaging (or "Systematic investment" that he mentions in the article, which sounds very much like what I have always called DCA but whatever) are different. Most people don't have a chunk of cash to invest, most people have some small amount of excess cash to invest each month or whatever. So in that world are you better off investing each month, or saving up until you have a lump sum? Almost certainly investing[1] each month.
For that reason, most people in the investment world would use some variant on internal rate of return here, which you could do a risk-adjusted variant of if you wanted to. I expect if you used either of those metrics you would find that DCA pays a small amount of excess return in cases when you would luck out and invest big after a decline in the lump sum case. In return for this, DCA has far lower variance, and far less likelihood of "effective ruin"[2].
[1] Technically if your savings amount each month is very small the transaction costs will eat up your capital if you trade too often so you should save until you have chunkier amounts in that case to mitigate this.
[2] Strictly if you're just investing long in cash equities (not derivs) risk of ruin is practically zero, but for a real person you can be very significantly harmed if say you bought equities at some high water mark, there is a big sell-off, the market basically goes sideways (declines in real terms after inflation) for a long period of time and you have to retire (so you have no earning power) and have to sell gradually at a loss to sustain your standard of living in retirement. This is effectively the position a lot of Japanese investors found themselves in after the big slump there.
I'd say that the strategy you're referring to is the same as "buy immediately (whenever you can afford it)," but interpreted in a DCA light.
You're right that it matters if we're talking about whether or not you have an existing savings you want to invest. But I don't see that distinction to matter here, since many people believe DCA is powerful because it rejects the idea of market timing and reduces risk, not because they're trying to put their money to work as fast as possible.
1. the S&P 500 has never gone completely to zero. It is not the typical gamble where you lose the entirety of your bet if the random doesn't happen your way.
2. you're assuming that it is an event instead of an investment where time matters, where are you are making multiple bets and historically the random is in your favor.
It can take 20 years for the S&P 500 to regain its former position if you include inflation, but for most people learning about investing, that is about a half of their expected investment time.
The S&P 500 specifically, no. But markets have gone to zero when (e.g.) there were Communist revolutions and private property went away.
If you were invested in those then you'd lose the money (and if it was as a domestic investor you'd have other (political) problems as well).
It is certainly possible for the United States to dissolve in an asteroid impact, but those tail risk type events are not worth worrying about for the type of investor that is going to be thinking about dollar cost averaging.
When I first started investing and only had a little money, I remember thinking that if I had twice as much, my money worries would be over. Many years later, I have 50 times as much money as then, and part of me still thinks that if I had twice as much as I have now, my money worries would be over.
And the other part knows that I am just the type of person who constantly worries about money.
> How do you know?
Empirical evidence:
> Technically, the researchers found that life satisfaction rises with the log of income[1] (i.e. multiples of income), not linear changes in income. The reason why log income is more relevant here is because someone with $10,000 might be much happier than someone with $0. However, someone with $10,010,000 is probably no happier than someone with $10,000,000. In other words, $10,000 means a lot to someone with nothing, but nothing to someone with a lot. I’ve written on how this idea applies to wealth[2] in the past and I plan on expanding on it in the future.
[…]
> In 2021, a study by Matthew Killingsworth[3] titled “Experienced well-being rises with income, even above $75,000 per year” shook the foundations of happiness research by contradicting the findings from Kahneman and Deaton’s famed 2010 paper. In particular, Killingsworth’s study found that both life satisfaction and emotional well-being continued to increase with income (i.e. the log of income), even beyond the $75,000 threshold reported by Kahneman and Deaton.
Wouldn't we expect them to be happier with their wealth doubled more than twice?
Anyway, most family wealth doesn’t turn into a dynasty. Usually the first generation that didn’t have to work for it ends up pissing it all away.
Your great-grandkids would find utility, as would their children's children. It's their perspective that matters for utility, not yours.
The fun thing is people talk log utility but forget there's an exponential process (children) that can follow until the lineage dies out.
Agreed it gets pissed away. Which means the children process doesn't have to be particularly exponential to make log utility become linear looking for all practical values when one includes descendants.
Not sure I agree. What I said at the beginning was that *I’d* be just as happy with 100 million vs 500 million. Their utility only matters in that calculation to the extent it affects my utility.
DCA is just a way to choose your security/cash split at different points in time though. Their relative performance at different points in time greatly impacts the results. Any time you have additional (probabilistic) information to bear on the problem, DCA can start to look a lot more attractive. For example, if you expect that in some time horizon there will be a dip of unknown magnitude and unknown bottom timing, then DCA outperforms many alternatives including buying up-front.
Anywho, just take care to blend what you know about the market (probably approximately nothing) with what your goals are (0% chance of hitting $0 vs a low chance of hitting $10M+ vs highest average yield no matter what vs highest yield predicated on being able to afford an upcoming purchase vs whatever else floats your boat). The right strategy for somebody else is not necessarily the right strategy for you.
I do think a lot of advice is about reducing volatility because amateur investors freak out over sudden negative drops and then make mistakes out of panic. The advice is not about maximizing returns.
I agree with this, but personally, I don't like DCA because you aren't managing risk. I found that it's way easier to manage fear and greed in markets if you have proper risk management, and most people simply don't want to learn this skill for whatever reason.
"Every battle is won or lost before it is ever fought", so you should have a plan where to enter and exit a trade.
> Time in market beats timing the market so to speak?
This works for past performance of S&P500, but if you look at other indices like Nikkei225 (Japan's lost decades of economic stagnation) this strategy doesn't work too well.
* https://ofdollarsanddata.com/dollar-cost-averaging-vs-lump-s...
Most of us do not have a pile of cash though, so it's better to get in the market a little every month:
* https://ofdollarsanddata.com/just-keep-buying/
* https://www.goodreads.com/en/book/show/59046778
And waiting to buy the dip doesn't work any better:
* https://ofdollarsanddata.com/even-god-couldnt-beat-dollar-co...
Yes the average person randomly assigned a lump sum on a random day will do much better if they deploy it immediately. But the tail person who gets the bad roll of the dice and receives it on a bad day will do much worse and may not have the money they needed to rely on.
You're not the average person, you're just one person from the population, and you don't know if you'll be the one with the bad roll of the dice or not. That's why people DCA.
If receiving a large lump sum like this you need to live off, either DCA over 12-24 months, or make sure you have some really good tactical allocation system that moves you to safety in the first few years.
The DJIA was 381.17 on September 3rd, 1929. It then went down and did not return to that level until over 25 years later - November 23rd, 1954. Factoring in inflation it would be losing money over a period of 25 years.
The DJIA was 1,051.70 on January 11th, 1973. It then went down during a period of enormous inflation, until it hit that level again on November 3rd, 1982.
We can look at the dot-com collapse in 2000, followed by the sub-prime collapse in 2008, which really roiled the economy, followed by the more recent period of stocks sinking last summer, inflation, FAANG layoffs, relatively tight VC money etc.
As Keynes said, in the long run we're all dead. You can put your money in the market and be underwater for over 9 years, or even over 25 years. Of course, on the other hand you can be completely out and miss out during one of the go-go periods.
And in many cases, if you didn't cash out and crystallize your losses, you could still be fine:
* https://awealthofcommonsense.com/2014/02/worlds-worst-market...
Also why diversification and asset allocation is important: having 20% in bonds in addition to being in the S&P 500 during the 2000s (post-dotcom, post-GFC) still allowed you to have an positive return:
* https://www.forbes.com/sites/advisor/2010/09/13/its-not-real...
Some Bogleheads did a similar analysis for 1980s Japan bubble: having some bonds and foreign (non-JP) stocks, limiting your JP equities to (say) <60%, and rebalancing ~annually gave you really good numbers, even after the crash.
> This is because when one decides to follow DCA is implicitly expecting that the market to fall in the recent future. However, our experience - at least as far S&P 500 is concerned - tells us that this is not what happens.
Which is basically saying that the S&P 500 always goes up in the long run.
With DCA you won't put all your money in one day before a crash.
Of course you are measuring that against the fact that the market generally goes up, so any dollars held outside the market are missing gains over time.
Randomly getting a huge lump sum of money would only happen very rarely. In reality, investing your paycheck each week is a much better method of investing than either trying to play the market or saving up in a low yield savings account.
This is horrible advice for the vast majority of investors who are getting paychecks each week. Just because it's on the front page of Hacker News does not mean it's true.
> The difference between DCA and SI is that in DCA you have all the money available since the beginning, but in SI you have to wait until the next month to have access to the quantity to be invested.
Pedantically, many believe that this is not actually DCA.
DCA is a strategy that is a counterpoint to lump sum investing. From that perspective, the common factor required by both strategies is that you actually have a lump sum to invest.
Investing every month as you earn money, assuming you never had enough to do a lump sum isn’t considered DCA from this perspective because there was no alternative to lump sum invest.
I do adhere to this thinking, and I think it would be useful to have a term for each of these three options. Unfortunately nobody’s proposed one that’s stuck for what you’re describing, so an endless debate rages on financial subreddits whenever this comes up.
I only bring this up because whenever people talk about DCA, there is an implicit assumption that they’re discussing from the perspective of their definition of the term. The article is—I assume—using the definition I described. Because yes, saving up your money to invest in one go at the end is categorically the worst option, and silly enough a strategy that it doesn’t often warrant mention or discussion.
Saving with each paycheck isn’t dollar cost averaging. It’s lump sum. DCA would be something like putting 1/4 of your total savings every week.
Anyway, in my experience this situation comes up a lot due to the above
Let's assume you're just buying a broad-based index fund. The same amount of money buys fewer shares when the index is up, and more when it's down. That is the very definition of DCA.
If I suddenly inherited $1m in cash and I planned out my desired portfolio allocation, I would want to enter into it over the course of at least a couple months because:
1) Interest rates are so high, you'd still be doing well earning 5% interest with the money parked in money market / bond / treasury ETFs. This significantly reduces the opportunity cost of waiting to invest the money or entering into a position slowly over time.
2) Markets fluctuate on a day to day basis often in response to things like fed meetings, jobs reports, earnings announcements, etc. You can average out the volatility by entering into the position over a few months (which, again, due to high interest on extremely low-risk cash/treasuries/bonds returning ~5% currently, the opportunity cost is minimal).
There is a fairly significant body of work around this idea by professionals in the field and a pretty well established set of trade offs between lump sum and DCA.
This isn't really a "debunking" so much as it's a fairly naïve and uncharitable comparison of DCA vs LS.
That allows returns and variance to be combined into one number, allowing more intuitive comparison and evaluation of the trade-off.
Clearly investing $400 of saving each month is not the person this article is talking about.
People that get windfalls don't generally park the cash in a savings account, they'll put some in a CD, money market, or bonds as they decide what to invest in otherwise. The money isn't sitting around doing nothing. The premise of this relies on an idea that the money is doing nothing otherwise (though it's available). That is not the case in most circumstances, not just because the money is usually in a safer investment, but because most people don't have spare uninvested money laying around, as I said.
Instead of “oh, I should wait for the dip” you can view your automatic purchases as “buying fewer shares because they’re expensive”.
Instead of “oh, the market’s crashing. I should wait for the bottom,” you can think, “I’ll just be buying more shares the cheaper they become”
It’s not a strategy per se, but more of a reasoning to stay the course and make your static monthly contribution.
And if it gets people into a saving monthly then perhaps it is not the worst of things
Wow, he talked to some "experts", ran a computer analysis, and now he's on the front page of Hacker News.
> The main conclusion of this post is then [sic] invest all you have as soon as you can
that's exactly the point of dollar cost averaging: if you're able to save, say, $250 a month, every month you buy $250 worth of something. Especially now that commissions are zero or nearly zero, it's a good thing to do.
You can have your own definitions but this is clearly explained in the post.
The author is free to add new qualifiers to the word, but putting your regular salary deductions into a stock or index, e.g. your 401K, is DCA.
https://www.investopedia.com/terms/d/dollarcostaveraging.asp
https://www.forbes.com/advisor/investing/dollar-cost-averagi...
DCA is what regular employees do with their 401K, so the title implies that they're doing investing wrong. They're not. The "LS" hypothesis corresponds to virtually no behavior in the real world.