DCA makes psychological sense if the concern is a FOMO on the best possible entry price. It takes the psychological pressure of finding the absolute best off the table and replaces it with a smoothed average entry price.
I think it's a more powerful psychological device (and I do tend to invest pretty much all at once when I reach a decision), but it's not entirely pointless mathematically.
This is a reason I haven't made active use of DCA. But DCA doesn't give up much, either, so I'm fine with "passively" engaging in DCA-like behavior by haphazardly dumping spare money into index funds after min("significant" drop, time limit) instead of purely maximizing time in market.
The time-weighted average price was $37/share.
Yes, that exactly. It's a psychological trick (if you have all the money upfront) and not a financially sound one.
Math citation needed - the noise function you're hypothesizing could just as easily drive the average up as down, unless you have a reason to think it should go down, in which case you're timing the market and there's a better future time to go 100%.
If you commit to the strategy (buy monthly for 10 months) then you're just choosing an unknown price over a known price.
And if you don't commit to the strategy, then you're actively trading and not DCA'ing.
Another thought: if the thesis is that the price will go up, then whatever sum you have and want to invest, should you invest it all now - given that the price is supposed to go up?
A financial hedge is, generally, something which reduces risk or improves on the risk:reward ratio in a mathematically demonstrable way. Generally this means bundling uncorrelated (or anticorrelated) assets. This concept won a Nobel for Econ in 1990(?), and is the foundation of modern portfolio theory.
Using "hedge" in a more colloquial way, it can mean "a way to protect yourself from an unexpected poor outcome". But by that definition this also fails, since you're just as vulnerable to outlier events, but the nature of those outlier events is different (and perhaps less intuitive).
Sure it does. Just like you say further down, people are "willing to accept lesser theoretical returns", except in this case it's to reduce e.g. tail risk. They could also buy up front and hedge, but that's already more technical than most people want to deal with.
If you're expecting a non-continuous return function, such as a pending drug testing result or a major governments deal, then you're actively trading and a DCA strategy is almost certainly suboptimal.
It's very hard to construct a situation which is both plausible and favorable for DCA _if you have all the funds upfront_.
DCAing is worse on average for simple objective functions, absolutely. Markets rise more consistently than they fall. But markets also fall faster than they rise. This may not matter to the hypothetical average person, but it matters to real people, and some of them make the perfectly rational decision, based on real-life economic factors that aren't captured by simple models, to trade returns for stability.
You can formulate this as e.g. an MINLP model and it isn't at all hard to construct situations that are favorable to DCA when lump-sum is an option. All it takes is adding constraints to reflect a real person's life circumstances. Other strategies are still better than DCA, but they're also more complicated to execute, and we aren't talking about sophisticated investors here.
Right, but what if the prices went down during that 10 month period, even if you were bullish? You invested at the top in one chunk, when you could have smoothed it out in increments instead.
That smoothing function is the entire point of DCA and takes the guesswork out of reading the market, like trying to find entry points.
The smoothing function of DCA will help mitigate FOMO and Buyer's Regret, but it won't (statistically) deliver higher returns.
Would you take a coin-flip where heads you double your net worth and tails you lose all your money and assets?
(Statistically) the expected outcome is zero, so it really shouldn't bother you either way, you could take the bet or not, it's the same if you go for it or if you don't.
In the real world, though, lowering your risk means reducing the chance of both winning the bet and losing it. For many people it's worth more to have less risk of randomly timing the market poorly, even though they have less chance of randomly timing it well.
Yes, over all investors the two decisions cancel out, but to the individual DCA can be worth lowering their risk.
The only risk it mitigates is the risk of a sudden bankruptcy, or some other massive exogenous event which disrupts your future plans.
However even accounting for those events the expected value of DCA is still lower.
So assume there's a normal distribution of price possibilities at each purchase point. You're going to end up with the _sum_ of those distributions, not the correlation or combination of them.
Presumably the deviations of those are growing more larger over time, but the are still the same underlying distribution curve as they sum up. The mean will trend steadily upwards.
The resulting price will be a normal distribution with a deviation of somewhere around the midpoint, and a mean of somewhere around the midpoint.
Now, with the mean increasing over time, your expected return will be decreasing the longer you wait. But what about the risk:return?
Unless you have outside knowledge about risk growing or lessening over time, it will remain constant over the DCA period, which means the risk factor of your investment (the standard deviation of the resulting DCA price) will be at exactly the midpoint of your purchase series.
So your expected return will be slightly lower, and your risk profile will be substantially larger, than just buying immediately.
DCA: math does not check out, not even close, _if you have all the money upfront_.
DCA versus lump sum has been debated for a long time and there is really not a clear answer or expert consensus on the subject.
DCA cannot save us from a crash, though it can reduce the fear of a crash to the point where we actually make an investment rather than sit on the sidelines.
If I think the stock will go up over the course of years but have really no idea what it's going to do over the course of hours/days/weeks, then approximating a buy-in at the average price over a longer period would seem to be a de-risking approach.
Your point that "If you think it will go down over several months, you shouldn't be buying in" is fine, but I'm not sure it's equivalent to "I don't know what it's going to do anytime soon I'm buying this for long-term prospects".
Imagine a stock with a sinusoidal price fluctuation at a frequency of just under a week, but which you expect to rise significantly over a 5 year period. Investing in that weekly over 5 weeks will net you a more locally average price than going all in today, particularly if you don't have good information about that sine pattern. Going all-in right now might net you a low price, but you may also be buying at a local maximum.
(Think of discrete waveform sampling)
I don't know what the conjecture is that would make it not de-risking that, or what a proof would be, maybe GP will clarify.
It should be noted that many analyses of DCA versus lump sum are around the S&P 500 overall. In the case of highly volatile growth stocks or just single stock investing like Fred is discussing in the article, market timing risk is more acute, since the drawdowns are more severe (it is quite rare for the S&P 500 to drop 70-80%).
Shameless plus, I am the owner of a DCA investing app + a simulator tool to backtest DCA with different stocks, over different time periods
If you invest at a few different times, whether fixed interval or random, is this not increasing the probability of approach an average value over the period you're investing?
I'm not asserting it's a good idea, I'm not using it as a strategy myself (or in fact making investment choices myself at all really), I'm just interested from the technical perspective.
It does indeed increase the probability that you will approach the average value over that period.
But there's no reason to believe that average will be lower than the current value.
100%, not disagreeing. But it does reduce the risk that you're buying at an outlier price.
Presumably you are choosing to invest in an asset because over the long term you believe it will go up. But you know nothing about what it will do in any defined timeframe.
Choosing to DCA with a fixed strategy is choosing x prices at which to enter, of which (x-1) are unknown. But they are fixed, and based on your above assumptions you know nothing about them except that they're generally trending upwards (and thus your entry price is averaging up by DCAing). Knowing nothing means knowing nothing, it's as likely to vary upwards as down.
You can express the above as a summation with a positive (but unknown) dv/dt and a canceling-out e if you think better in notation, but I can't figure out how to make that legible in a HN comment.
If you have any reason to believe that there is more downside risk than upside risk over the DCA period, then you should delay investing until that is no longer true, and go 100% at that time.
DCA is essentially the same intuitive leap as the Monty Hall problem, but backwards.
I have no reason to believe that over the relatively short period of DCA that there is specific upside or downside coming along immediately, but I know that there is noise in the pricing. The rising trend I’m betting on is further down the line and very slow.
You’re adding assumptions that support your view, rather than addressing the problem as stated.
The only thing we know is that we're choosing an asset which we believe will trend upwards.
Over the relatively short period of DCA, it will trend upwards.
If you're saying you know that the rising trend won't begin until a future date, then that's an assumption you're adding, and it probably indicates you should wait until the rising trend begins until increasing.
But this is an assumption!
> If you're saying you know that the rising trend won't begin until a future date, then that's an assumption you're adding
I'm saying I have no idea what's going to happen during the DCA period.
I think at this point its safe to say your view that DCA doesn't work to denoise a signal is predicated on assumptions about what the price trend is doing over the DCA timescale. If the stock is trending up over that period, then you're likely better off buying now regardless of local fluctuation. OK!
Depending on the equity, you are essentially sampling from a skewnorm & summing over 10 variates. eg. if you invest in the broad market, ^GSPC (your s&p 500) has a skew of +0.4 and a variance of 2.8, so your five number summary works out to
summary(mapply(function(x){sum(rsn(10, dp = cp2dp(c(0,1.67,0.4), "SN"))[1:10])},1:10000))
[-18.3, -3.7, -0.2,3.4,18.8]. While the median scenario is net neutral, you do end up underpaying 3-4 times if you fall below the first quartile.
otoh, if you are feeling frisky and want to dump 100k into robinhood over the next 10 months, $hood has a skew of -0.88 & a variance of 26.3, so 10k realizations of this scenario -
summary(res<-mapply(function(x){sum(rsn(10, dp = cp2dp(c(0,5.1,-0.88), "SN"))[1:10])},1:10000))
yields [-70.8, -10.6, 0.7,11.4,47.5] - much more interesting set of circumstances!
I guess he's betting on prices generally headed down over the next 10 months, so he significantly underpays if things go per plan.
Yes, exactly that! And if one is betting on prices generally going down over the next 10 months one should just not enter the market at all.
A good example would be an upcoming test result report for a pharmaceutical - it's either going strongly and gradually up (long) or suddenly and significantly down (put), but not gradually down (in which case you'd be out on both puts and longs).
But yeah, at this point you're a super-sophisticated investor and all of this discussion is meaningless. :)
https://www.financialplanningassociation.org/article/journal...
> If you decide to invest $100,000 into the stock market because you agree with my reasoning in this post, then it would make a lot more sense to invest $10,000 a month over the next ten months than invest all of the $100,000 this month. (Quote from article.)
That doesn't actually make a lot more sense, and that is not the argument for dollar-cost averaging.
Assuming you're investing in a single equity/portfolio, then at the end of that ten month period you will have an effective entry price for the entire lot either way. Either that price is the value it's at _now_, or that price is the weighted average of ten entry prices.
One thing we know, or at least which we're assuming when we choose to invest, is that over the long term prices go up. So unless you have outside knowledge or asymmetric insight suggesting prices will in fact go down, it makes no sense to DCA into the market. And if you do think prices will go down before they go up, then you should hold the entire amount back until you no longer believe the asset is likely to decline.
Dollar-cost averaging is not an argument to buy slowly when you already have money, it is a reassurance to keep investing when process go down and not to hold any money back when investing money you're receiving on a regular schedule (like a paycheque).
Ironically (and amusingly) DCA is also an entry method into a position when you're willing to accept lesser theoretical returns to avoid FOMO but are feeling timid, which is definitely on-brand for a VC.