Dollar Cost Averaging is a Myth?

[JAC]

I used to believe that constant, regular purchases of a security minimizes market risk. However, now I read that this is wrong, either as compared to the same number of buys at random intervals, or even lump sum investing.
One site says, "Here's the problem: It's bunk. Finance professors have known this for more than 20 years, since an article disparaging the concept appeared in the Journal of Financial & Quantitative Analysis in 1979. Since then, numerous studies have confirmed this analysis. And yet, "despite more than two decades of damning evidence, DCA remains as popular as ever amongst the rank and file of individual investors," wrote Moshe Arye Milevsky of the York University (Ontario, Canada) school of business in a journal article in 2001." (http://moneycentral.msn.com/content/P104966.asp)
Another says "As appealing as that theory is, its advantage looks like a myth, as this calculator shows. It uses market data to let you compare dollar cost averaging with lump sum investing for the start date you specify." (http://www.moneychimp.com/features/dollar_cost.htm)
I can see why random buys would perform similarly. One argument in favor of lump sum buying is said to be that markets go up about twice as often as they go down.
TSPers don't have a lot of choice, obviously, but I think it is interesting that such a widespread belief is said to be wrong (in the sense that DCA is not superior to other methods). What does anyone else think?
Jonathan

[rokid]

Do know the title/author of the 1979 article referenced by Mr. Middleton?

I have read articles comparing the advantages of investing a lump sum to DCA when you have a lump sum to invest. For example, if someone leaves you $1M. However, DCA investing over a long period of time into a 401K seems to be universally applauded.

[JAC]

Nope. Online indices only back to 1995. DCA is reasonable given computerized banking. It makes people invest regularly. Sales people like it because they get your $$ in a perpetual stream.

BUT, the reason usually given is that that it reduces market risk or the average longterm cost of the security better than lump sum or stochastic investing, which seems to be false. When you think about it, random would be just as good, and to beat the longterm average cost of the security (no matter when you invested), you'd have to time the market to avoid highs, and that requires prediction, which is exactly what DCA was supposed to finesse.

[rokid]

Nope. Online indices only back to 1995.
I'm not quite sure I understand your statement. I have TSP fund index data back to 1988. I have C Fund, I Fund, and REIT data back to 1970 (S&P 500/EAFE/NAREIT indexes). Most researchers use stock/bond market data starting in 1926.

I find Mr. Middleton's referencing of a study - no author, no title, suspect. In addition, it's easy to come up with scenarios in which DCA would out perform lump sum. For example, this year, lump summing the I Fund on 9 May would have been a disaster compared to starting a DCA program on the same date.

Mr. Middleton may very well be right. However, it's hard to verify his claim without reading the referenced study.

[JAC]

Sorry, I meant the online indices to the Journal of Financial & Quantitative Analysis only go back to 1995, so unless someone has access to that pub, we'll probably never know what the academic said.
Looking back in time one can certainly pick days when lump summing does better or worse than DCA. The hard part is looking forward. That's also why one of the articles said stocks move up about twice as often as they move down. If true, I guess it means bull markets last longer than bears. That might mean that lump sums are better than DCA because you get all your money into the rising market ASAP, as opposed to stretching it out. In a falling market, DCA would outperform single large buys.
DCA assumes that you can't time the market...in the long run, it's equivalent to many small random buys. If you can time the market, then lump sum might be better, unless missing out on lots of little bounces did you in.
I think the point is that DCA is no better that other ways. It's neutral.
J

[JAC]

And, what we do here is pretty much lump sum transfers, no? For example, did the market last week find a new bottom, or was it part of the much anticipated 10% correction? If the former, transfer a big amount to the C fund. If the latter, DCA into the C fund. But few here DCA in and out of funds, tho I guess we all DCA into TSP to maximize employer matches, etc.
My point. If DCA'ing was a sure bet, then I ought to move that 30% into the C fund in 15 weekly trades of 2%, not 30% tomorrow.
J

[rokid]

If DCA'ing was a sure bet, then I ought to move that 30% into the C fund in 15 weekly trades of 2%, not 30% tomorrow.
J
Nothing is a sure bet. However, if to DCA or not is important to you, I'd get a copy of the study. Alternatively, engage Birchtree in a conversation. He's a huge DCA advocate.

[JAC]

I'm not personally committed one way or the other. Just back into learning about all this stuff after a gap and while nosing around found these attacks on DCA, which made sense. This being the longterm strategy forum, thought I would see if anyone had an opinion. I suppose Birchtree will happen upon this if he's interested.

[Birchtree]

JAC,

Your investing strategy may depend on the amout of energy you have available to stay market current. When a member asks me for a recommendation prior to deployment for example, I usually tell them to put their DCA on automatic pilot, suggesting 75% C fund and 25% I fund until they are in more control. Fate can be an excellent guide providing you have a trust in fate. It's a simple way to cycle ride and accumulate as many shares as possible with the least amount of energy expenditure. I would rather accumulate the shares than push for the dollars - the dollars will eventually arrive on their own accord. I want 40,000 shares backing me up when I'm ready to pistol shoot - and staying with TSPtalk will help educate me in the fine art of hitting the targets. When you play with that kind of money you don't want to make many mistakes. There will be more volatility going forward in the next few years and I'll be ready with my dual Sturm Rugers.

Dennis

[Dave M]

This is the question I can't answer, consistently. I once believed that number of shares was the key. Then I thought, the only reason to accumulate shares is in order to accumulate dollars, so why not count dollars instead?

Which would you rather have, lots of shares worth nothing or lots of dollars worth nothing?

Dave

[Birchtree]

I'll take the shares every time - because once they are yours they are yours. They may fluctuate in value over the years as you accumulate them, but they still remain yours. And at some point in the distant future the gift horse may come along at the right time - and bingo you do an IFT for safety. Wasn't that easy?

[GordonGecko]

Why look at DCA as an 'all-or-none' proposition? Why not use a combination of DCA and technical analysis to adjust the contribution rate?

Since markets, when viewed from a long-term perspective, have discernable trends, for example an intermediate trading-range within the long-term uptrend, adjustments in contribution allocation can be used to take advantage of those fluctuations.

It's not that hard to identify the upper and lower trading range, and adjust your contribution allocation accordingly, depending where the index value is within the trading range.

For example, in a 'long-term up' market, you calculate the 50-week or 60-week moving avg of the S&P 500 (this is assuming that the index value is higher than the MA), then adjust your contribution allocation upward as the index falls closer to the average, and decrease it as it moves up and away. You still maintain DCA throughout, but at a varying rate.

The same goes for declining markets. If the MA is declining and the index value is below the MA, then you decrease the contribution allocation the closer you get to the MA, and increase it as it moves farther away.

This is clearly an improvement over using DCA with a constant rate of allocation.

In summary, contribution adjustments ought to take into account that intermediate changes in price within the long-term trend can be measured, and should increase the number of shares over the 'constant rate' approach.

[nnuut]

EXACTLY GG!!!:D

[JAC]

GordonGekko's point assumes you can time the market successfully, whereas the usual rationale for DCA is exactly the opposite: it minimizes risk regardless of market-timing. But that turns out to be a myth. Of course, this whole site is dedicated to the notion that market-timing works, so one wouldn't expect subscribers to think otherwise.
There are psychological and efficiency reasons to DCA: think hard, make your choice, and set it on auto-pilot.
But the original query had to do with whether regular small buys reduced the long-term average price of the security compared to random buys or even lump-sum. Apparently it does not.

[GordonGecko]

Agreed. My slightly off-topic point was that both DCA and market timing approaches each have benefits. Why not use the best from each?

My opinion is that a combined approach is more likely to be more successful than either DCA or market timing alone over the long haul. Do I have proof? No.

I can say it takes into account that few can make accurate predictions about short-term market direction, while intermediate and long-term market direction are not difficult to determine.

[Pilgrim]

Try this one - conclusion: It's Bunk

"The costly myth of dollar-cost averaging"
http://moneycentral.msn.com/content/P104966.asp

[charmedboi82]

Try this one - conclusion: It's Bunk

"The costly myth of dollar-cost averaging"
http://moneycentral.msn.com/content/P104966.asp


The conclusion I've reached from all of this is that it really depends on what happens in the future. They're using the logic that markets rise twice as often as they fall, so they're using the past to 'predict' the future. Of course, you could set the example up to show a market drop. So, of course, it depends on what you're investing in and what happens in the future. The future's always up in the air.

Also, they're talking about dollar-cost averaging, but as in the article, contributing to a 401/TSP/IRA every pay period is NOT dollar-cost averaging. It states this clearly. It's the difference between having the lump sum to invest initially and not having it (spacing it out over time because you have to, not because you have a choice).

[VirginiaBob]

Of course, investing with a large lump sum is better, but if you had a large lump sum to begin with, it's likely to be family money, which most of us didn't have access to.

[Partyartie]

Having read the Middleton article and digested all that has been written here, I find that I cannot/will not write off the concept of dollar cost averaging. Since the statistics show that investing in common stocks has provided the best return (compared to other types of investments) since the 1920's, it isn't hard to understand that investing in a lump sum would provide a larger return, particularly over a long period of time.

I feel it is important to recognize, however, that investing equal sums on a periodic basis should (at least theoretically) address short-term ups and downs in the market, and in some manner address the small investors qualms about "losing money".

If the DCA concept was indeed conceived to extract the funds from small investors that they were unwilling or unable to commit in a lump sum, I still feel it's indeed better that they invest on a DCA basis, rather than not invest at all. If it does address those short-term peaks and valleys, then so much the better.

Certainly, you true financial aficionados can speak derisively of such a basic (or possibly even wrong-headed) concept, but it may have some value for those who are not financially astute, or are just beginning to invest.:)

[Birchtree]

I'm a seasoned investor and DCA is my redeemer. I enjoy looking back over dividends and allocations that have been previously reinvested and noticing the low prices. I wish I could be so smart on my timing.

[Bullitt]

The key is that a DCA approach has zero emotion attached to it. On a given date, money goes to the account and it happens. Even using Technical Analysis has an emotional side to it. Since Tech Analysis must be interpreted, you could be making a premature or late decision based on speculation. Besides once you see a MACD Cross or whatever it is a trader looks for, it's already happened. In my case when prices drop significantly because of something stupid like a Chinese Crash, I have to take the opportunity and add more money to the DCA amount.

[JAC]

Confining ourselves just to the TSP system, I think the actual facts are more subtle.

The only way into the TSP is regular small deductions every paycheck, which looks like DCA. It is DCA if you pre-allocate that deduction to some mix of F, C, S, or I, and never or rarely re-allocate once you buy. But if your contributions are 100%G, you are essentially accumulating cash in anticipation of a lump sum trade into one of the funds whose price can actually go down. I guess you could do an even more extreme lump sum strategy by setting your contributions to 100% of your pay to G for 3 months, then 0% for 9, but I doubt if many do that.

So, 100%G is a lump sum strategy, and 0%G is DCA.

Personally, I now do the former, but did the latter for 20 years. For most of those years, you couldn't trade daily, and your choices were stocks, bonds or cash, so the ONLY way to allocate was to DCA upfront and fix things during "open season." I DCA'ed into the C fund.

Now we can day trade. Buying 100% G sets up lump sum investing, and things get interesting.

Once in, though, who DCAs inside TSP? Nobody. Of the people who are tracked on this site, everyone swings in big trades: Tom just went 100%G in one trade. Others might go from 50%I to 10% I or whatever, but they do it all in one large trade.

DCA'ing inside TSP would mean going from 50% to 10%I in a series of, say, 20 2% trades. In that sense, almost nobody DCA's. Single large trades.

Therefore, for TSPer's DCA means differential allocations at the contribution stage. If we wanted to compare strategies using real data, we'd need a tracker that, say, only moved between C and I, and a tracker that DCA'ed into the C and I initially but didn't trade thereafter. For a fair comparison, their long-term averages would have be the same, e.g. 50C:50I.

So the comparison would be: 100%G contributions followed by sporadic, random (or non-random, "timed") buys of 50C:50I, versus initial "DCA" 50C:50I buys. The financial gurus say none of it matters. In the long run, random and DCA don't differ. The only one with a chance to beat the averages is the non-random timed buy, but beating the averages is very, very difficult.

Jonathan

[Birchtree]

That is an excellent post. I did some DCAing from the C fund into the I fund in 2% increments (around 10K) from the $21.65 price, again at $22.04, then at $23.21, then at $23.55, and finally at $24.10 for a total 10% position. I went back to the C fund from an I fund price of $24.41. I may try and do that again on the next spill. Most of the time I prefer to DCA my allocation into one fund and let fate be my guide. It really works well in a bear market like 2000-2003. The lower the price the more shares that are accumulated - my portfolio redeemer. You just have to be willing to dump good money down a dark hole. But it always pays off in the end. Now I will have to pay over $17.00 for my next DCA into the C fund - ah you can't win them all.

[James48843]

Parable of the TSP Talents

14"For it is just like a man about to go on a long extended TDY journey, who called his own workers and entrusted his possessions to them.


15"To one he gave five talents, to another, two, and to another, one, each according to his own ability; and he went on his TDY.


16"Immediately the one who had received the five talents went and traded with them, keeping them in the “I” fund, and exchanging with the “S” fund, and, from time to time, retreating in the “G” fund. He watched TSP Talk, listened to Cramer, did his “due diligence”, found ebbs and flows, and over time, this one gained five more talents.


17"In the same manner the one who had received the two talents gained two more, although only alternating a portion of his portfolio occasionally between the “C” fund, and mixing it with bonds in “F” for diversification, while filling the balance with an age appropriate “L” fund- the L2020, and sitting on that portion quietly. And compounding over time led to the two additional talents, not bad for moderate risk.


18"But he who received the one talent went away, and dug a hole in the ground and hid his master's money. Kept it in the “G” fund always. Never took a risk, afraid that the markets may repeat the 1929, or 1987 downsides, and living a life of fear. He bothered not to learn about investing. He simply made a single one-time deposit, and left it there.


19"Now after a long time the master of those workers came and settled accounts with them.


20"The one who had received the five talents came up and brought five more talents, saying, 'Master, you entrusted five talents to me. See, I followed the markets closely, learned much, exercised careful swing trading based on the Ebb trader, and TSPTalk, and other appropriate learning tools, , and I have gained five more talents.'


21"His master said to him, 'Well done, good and faithful servant You were faithful with a few things, I will put you in charge of many things; enter into the joy of your master.'


22"Also the one who had received the two talents came up and said, 'Master, you entrusted two talents to me. See, I have exercised age appropriate risk models, using the “L funds” for part of the time, and diversified with mixed stocks and bonds, and gained two more talents.'


23"His master said to him, 'Well done, good and faithful slave. You were faithful with a few things, I will put you in charge of many things; enter into the joy of your master.'


24"And the one also who had received the one talent came up and said, 'Master, I knew you to be a hard man, reaping where you did not sow and gathering where you scattered no seed.


25'And I was afraid, and went away and hid your talent in the ground. The G in G fund stands for ground. See, you have what is yours.'


26"But his master answered and said to him, 'You wicked, lazy slave, you knew that I reap where I did not sow and gather where I scattered no seed.


27'Then you ought to have put my money in the L INCOME, and on my arrival I would have received my money back with interest.


28'Therefore take away the talent from him, and give it to the one who has the ten talents.'


29"For to everyone who has, more shall be given, and he will have an abundance; but from the one who does not have, even what he does have shall be taken away.


30"Throw out the worthless slave into the outer darkness; in that place there will be weeping and gnashing of teeth.

[SkyPilot]

Excellent use of the literary/Biblical form... :)

[malyla]

Confining ourselves just to the TSP system, I think the actual facts are more subtle.

The only way into the TSP is regular small deductions every paycheck, which looks like DCA. It is DCA if you pre-allocate that deduction to some mix of F, C, S, or I, and never or rarely re-allocate once you buy. But if your contributions are 100%G, you are essentially accumulating cash in anticipation of a lump sum trade into one of the funds whose price can actually go down. I guess you could do an even more extreme lump sum strategy by setting your contributions to 100% of your pay to G for 3 months, then 0% for 9, but I doubt if many do that.

So, 100%G is a lump sum strategy, and 0%G is DCA.

Personally, I now do the former, but did the latter for 20 years. For most of those years, you couldn't trade daily, and your choices were stocks, bonds or cash, so the ONLY way to allocate was to DCA upfront and fix things during "open season." I DCA'ed into the C fund.

Now we can day trade. Buying 100% G sets up lump sum investing, and things get interesting.

Once in, though, who DCAs inside TSP? Nobody. Of the people who are tracked on this site, everyone swings in big trades: Tom just went 100%G in one trade. Others might go from 50%I to 10% I or whatever, but they do it all in one large trade.

DCA'ing inside TSP would mean going from 50% to 10%I in a series of, say, 20 2% trades. In that sense, almost nobody DCA's. Single large trades.

Therefore, for TSPer's DCA means differential allocations at the contribution stage. If we wanted to compare strategies using real data, we'd need a tracker that, say, only moved between C and I, and a tracker that DCA'ed into the C and I initially but didn't trade thereafter. For a fair comparison, their long-term averages would have be the same, e.g. 50C:50I.

So the comparison would be: 100%G contributions followed by sporadic, random (or non-random, "timed") buys of 50C:50I, versus initial "DCA" 50C:50I buys. The financial gurus say none of it matters. In the long run, random and DCA don't differ. The only one with a chance to beat the averages is the non-random timed buy, but beating the averages is very, very difficult.

Jonathan

Just found this. It answered much of the question posed below. Great website. - Malyla

This has been on my mind for a while. I know this thread is old, but I think someone may know what the answer to this contribution question.

If I allocate my biweekly contribution to equal parts in C,S,I (say split $337.5 into buying shares in each fund), what happens when I rebalance my account with an IFT from G to stocks? Are the small amount in the stock funds (from the bi-weekly allocation) sold at the COB share cost and then shares are bought as defined in the IFT (say 100%C)? If that is the case, then doesn't that make DCAing mute as you do not really get to keep those C stocks that you bought with your contribution (possibly at a low stock price) at that price you paid because it is all reshuffled to the new stock price at the time of the IFT?

bought $112.5 C at $10/share = 10 shares or 0.15% of your account
bought $112.5 S at $15/share = 7.5 shares or 0.15% of your account
bought $112.5 I at $16/share = 7.03125 shares or 0.15% of your account
have $74,662.5 in G at $6/share (12,443.75 shares)

Perform an IFT 100% C at $11/share

Does IFT cause all the shares to be sold to buy the C fund at the $11/share price? Even if you could specify less than 1% increments in the IFT, everything would still be sold at the COB share price and bought in the distribution specified in the IFT.

Does this mean that DCAing with our TSP fund is impossible? The DCA method is about gathering more shares in a fund. The only way I can see how to do this is to go to G when C is high (say $13/share), then buy back into C when it is low (say $10/share) etc.... Is the DCA method just a sell high, buy low method and am I correct on what happens when we do an IFT? If so, then it’s an all or nothing DCA method as we can not keep shares at previous cost once we make an IFT.

Thanks.

[alevin]

Maylyla, if you look at your TSP quarterly reports of account activity, it shows all the account activity that has occurred that quarter, IFTs as well as biweekly contributions. I think you can figure out from those records what is really true re selling existing shares vs. buying new shares to ADD to the existing shares.

[malyla]

Maylyla, if you look at your TSP quarterly reports of account activity, it shows all the account activity that has occurred that quarter, IFTs as well as biweekly contributions. I think you can figure out from those records what is really true re selling existing shares vs. buying new shares to ADD to the existing shares.


That's just it. It doesn't seem to be an ADDing to existing shares when you do an IFT. Contributions look like an ADDing to existing shares, but DCA does not seem to work if you make an IFT. DCA is a B&H strategy with contributions to stocks every two weeks which means you ride down with the Bear market ( no capital perservation).:sick:

[alevin]

We agree about DCAing into falling market, that's why I'm completely out, biweekliesand IFT, and plan to stay that way until TA tells me time to catch the train as it goes by for an intermediate rally (trend followers unite!).

[malyla]

We agree about DCAing into falling market, that's why I'm completely out, biweekliesand IFT, and plan to stay that way until TA tells me time to catch the train as it goes by for an intermediate rally (trend followers unite!).

TA = Technical Analysis ?

What will tell you that a rally is starting?

Thanks.

[alevin]

I'm still pretty new at this, I read a book called Trend Following earlier this spring, which kind of tracked with what I was already zeroing in on with intermediate trend indicators that already interested me, which means waiting til I see the trend started, using indicators that will put me in reactive mode. I tend to study conditions a bit too thoroughly and slowly at times, so I'm not good at the daily in-out jumps, am more comfortable seeing things build and waiting for confirmation that an intermediate trend started (my intermediate might be some other peoples "short-term"-Market Gauge defines short term that way anyway).

I don't know when a rally will start precisely, but I'll know it when I see it's started. I've been talking about this sporadically and somewhat cryptically in my thread, the indicators I've been learning about and watching and studying for entry signals include MACD, SAR, Keltner channels (which I haven't mentioned previously), ADX signals (that one is pretty complex and only really good for entries, not exits), and when I see the entire daily price range make it over the 5-day moving average coming up from below (HLC symbols or candles). I have occasionally jumped the gun and gone with a strong candle reversal signal, and had mixed success so far-its too quick for our 2x/month IFT situation.

One fairly good but imperfect intermediate exit signal seems to be when the HLC/candle drops more than halfway below the 5 day MA. I'm still working out other exit signals that will be more helpful in this upcoming bear rally environment (I'm not sure when the rally will actually start-it came close earlier this week in C, and was starting to move that way in I but no confirmation happened. I chose to miss the current/recent F rally crossover in favor of waiting for the "big one in stocks".

I don't rely on one signal alone, I'm always looking for 2 signals or more in combination to confirm these "intermediate" rallies have started, and for the other indicators to be moving that direction or close to getting there. For the return of the "bull" longer term, I'm waiting on the cross-up of the 20week over the 50 week (Karl Denninger's writings put me onto that one over at Market Ticker-we came close awhile back but never got there). I'm using my tracker account here to try different ideas, not always doing precisely the same thing in real account, but fairly close, timing a little different sometimes. I'm up over 2% in real account right now vs. tracker (-2%) due to slight timing diffs. If you'll notice I stayed out of the game this whole past week-none of my signals had hit yet, so I'm still waiting til 2 or more do in 1 of the funds-who knows when it will happen? I don't.