Hacker News new | past | comments | ask | show | jobs | submit login
You May Be Better Off Picking Stocks at Random, Study Finds (studyfinds.org)
146 points by Vaslo 62 days ago | hide | past | web | favorite | 77 comments

This has been known for what, 40 years?


Try 156 years!

> The concept can be traced to French broker Jules Regnault who published a book in 1863 [0]

[0] https://en.wikipedia.org/wiki/Random_walk_hypothesis

The book is about much more than stock picking, though.

It is a good read. And still relevant.

If a random strategy is better than whatever the non-random strategy was then intuitively in most cases the inverse of the non-random strategy is better than the random strategy.

I am currently reading and would recommend "Bull!" by Maggie Mahar for a longer term view on the stock market than many people today seem to consider it on.

That sounds intuitive but does not follow. The strength of the random strategy is that it approximates the theoretical ideal - a maximally diversified market-cap weighted portfolio. The non-random strategy is too concentrated, and inverting that can leave you with another strategy that is also too concentrated compared to the ideal.

The problem isn't that the non-random strategy did poorly, it's that it has too much exposure to needless sources of uncertainty. The inverse strategy doesn't get rid of uncertainty, it just flips the sign of their exposure to that uncertainty.

> The strength of the random strategy is that it approximates the theoretical ideal - a maximally diversified market-cap weighted portfolio.

That was a very elegant way to make it click for me, thank you.

> The strength of the random strategy is that it approximates the theoretical ideal - a maximally diversified market-cap weighted portfolio.

How is that the theoretical ideal when a few good stocks far outperform a maximally diversified portfolio?

The short answer is that hedge funds etc hire teams of math PhDs, rent out supercomputers, acquire non-public information like satellite imagery, and do various other hard activities in order to accurately price securities. Unless you think you can beat these players at this game, there's no way to pick better-than-average stocks.

The longer explanation is that you can break down the price effect of various real-world events into two axes. There are things that transfer market value between companies, and things that change the value of the market as a whole. For example, people buying more Macs instead of PCs will transfer market value from Microsoft (among others) to Apple. On the other hand, a novel form of efficient energy production will do more to increase overall economic activity.

The maximally market-cap weighted portfolio has diversified away the first axis of risk. Suppose company A, worth $500M, loses $50M of market cap to company B, worth $100M. If you own $500 of company A and $100 of company B, the $50 loss from owning company A stock is completely offset by owning $10 of company B stock. You're purely exposed to changes in the value of the economy as a whole, and have eliminated diversifiable risk from your portfolio.

If you can't benefit from taking on diversifiable risk, you're better off getting the same return at a lower level of risk by simply removing it.

I feel like a significant fault in the model you propose is that you're exposed to the value of the public market, not the economy.

Market cap lost to non-public companies will disappear from your portfolio.

That's a fair critique, but one that applies equally to other strategies for picking stocks of public companies, so is kind of out of scope.

Because you only know which stocks in hindsight.

I think the logic here is flawed.

Take a really simple portfolio... I buy (or go long on) 1 share of GOOG and 1 share of MSFT. Over a decade I make an average of a 7% return per annum, while the random strategy makes, 8%.

What exactly would the inverse of my strategy be? To go short on GOOG and MSFT? That would do even worse and in that case I'd actually lose money.

Would it be long every stock except GOOG and MSFT?

Boglehead reporting in, you should long every stock.

Observation: I find the uptick in stock buying articles just when the market is at it's frothiest quite interesting.

Stay vigilant out there traders, don't eat the 30% haircut that will be visiting global financial markets shortly.

I'm confident that haircut will come.

I am not at all confident that anyone can predict when it will come well enough to avoid it.

Remember that the last 3 months of 2018 saw the market drop ~20%, only for it to recover over the next 3. Were you able to sell in September and buy at the bottom in December? Almost certainly not...

Yes, our prop group made quite a bit with out-of-the-money AAPL 135 puts (AAPL will hit this strike price again).

Alarms bells are ringing in this market... If your good at risk management and can initiate low cost perma-bearish positions, you will be rewarded heavily over the next year or so.

Alpha chasers that don't have risk management skills should just buy Bitcoin (dont sleep on this, the halvening is May 2020-ish). Dollar-Cost-Averaging will get you a decent entry price.

Warning: Bought more 120,130,140 AAPL puts in prep for next collapse, if your in my way, I will literally eat your lunch.

Good luck guys!

> What exactly would the inverse of my strategy be? To go short on GOOG and MSFT?

No, it would be to take 1 share of every stock except GOOG and MSFT and it would work (well, under the given assumptions and with random = 1 stock of every pick).

Buy random, short non-random.

It is already being done by HFTs. https://seekingalpha.com/article/4205379-robinhood-making-mi...

The reason why this doesn't work for the average person is because they don't have the infrastructure for it or they don't trade enough to get the commissions down to make it profitable. Shorting comes with additional fees.

Concrete example why it doesn't necessarily work:

Lets say you always have two companies, A and B. Each year one of them triples in value, the other loses all its value, but the loser gets recreated so you can invest in it again. Each year you get to distribute your funds.

Your original strategy is to invest everything in A every year. But after doing some simulations, you see that your current strategy makes you lose all your money within a few years since A will sooner or later make you lose all your money. Should you invert your current portfolio and only buy B? Of course not, both stocks are equivalent in isolation!

The best strategy is to allocate your funds 50/50, that way you double every year. Allocating each part of your money randomly would get close to this, so is still way better than just picking one or the other. What is important is to find stocks whose risks doesn't correlate with each other, since correlated risks will cost you a huge amount in times of down-turns.

I'm not sure there implementation of inverse in stock strategies, but assuming the basis is returns proportional to market rate x, that logic doesn't really follow for all cases I think. If the random strategy has gains of 1x and the non-random strategy has a loss of -0.5x, then wouldn't the inverse just be a gain of 0.5x?

If the non-random strategy was “buy stocks A, B, and C”, the inverse might be “the broad market minus the stocks A, B, and C” rather than “short stocks A, B, and C”.

Hmm, let's call "buy nothing" a neutral strategy. Then a combination of “buy stocks A, B, and C” and “short stocks A, B, and C” has the same effect as the neutral strategy. But the combination of “buy stocks A, B, and C” and “the broad market minus the stocks A, B, and C” is equal to "buy everything" and has not the same effect as the neutral strategy.

You seem to assume "buy everything" to be the neutral strategy. Then everything plays out as you say.

I believe the strategy comparison here is based on how to best allocate $X in the stock market.

Choosing not to allocate it is kind of irrelevant, just like "play tic-tac-toe instead" is not a chess strategy.

Not every hand is a winner in poker. Best big or get out and hold but don't always do only one.

That's true but still doesn't map here. Or at least, to my interpretation of it.

Stock-picking is specifically referring to where you should allocate a given amount if you are investing. There's obviously more to stocks than simply picking them, but as for picking specifically, there's no analog in poker because you don't divide your bet that way. (Aside from raising on a bluff, but that's stretching it).

Whether, when, and how much you should invest is a separate strategy that's related to your confidence and expected return of the picking strategy—but not the same thing.

Just like whether and when you agree to a game of chess vs tic-tac-toe might depend on your confidence in your skills; but that decision is _not_ relevant to chess strategy, which assumes you're already playing the game.

Why not call buy everything in equal amounts a neutral strategy?

Buy nothing is not even a strategy. It's like calling empty set a neutral real number

This still doesn't make sense to me. If the return of inverse strategy S' were greater the loss of strategy S, couldn't you then guarantee a positive return by using both strategies simultaneously?

You have a set of stocks either correlated with the U.S. economy or China's economy.

Amateurs tend to pick uptrending correlated stocks (all stocks trending up in the U.S.). When U.S. economy crashes, all their eggs are in the same basket.

If you'd tell the amateurs to pick the inverse, they'd go for downtrending Chinese stocks. When China's economy crashes, all their eggs are still in the same basket.

I think what you are looking for instead, is contrarian trading strategies. Here you follow counterstrategies to what the large herd is doing. A good contrarian strategy for buying bitcoin, may be to gauge crypto sentiment on HackerNews. If the majority is gleeful or pessimistic, the price is too low for future value, if articles get posted on how to build your own blockchain in Python, then you should be ready to start converting to money, because one month later, every smart nephew's uncle fomo-bought and panic-sold the hype and caused a crash or depression. Similarly, if the U.S. president is glowing about the heated economy, and dismissive of China, this opens up new profitable options for contrarian traders (which are less risky/more informed than completely random or following the herd).

Not true because of trading costs. Both the non-random strategy and its inverse would lose money.

Given non-random strategy S, what does it mean for strategy S' to be "the inverse of" S?

Couldn't you phrase a strategy as a time series of binary decisions, e.g., of the type "buy/sell stock X at current time: yes or no"?

Then the inverse would be obvious. However, it's not immediately clear to me whether the claim holds that an inverse thus defined would perform better than the original strategy.

Does that work for the "no"s? I didn't buy or sell any stock yesterday; with that definition of "inverse", wouldn't I have to perform all possible actions? For any time, stock and amount, the inverse of inaction would be "yes, buy/sell X shares of stock Y."

It wouldn't work because you don't have an unlimited amount of money. So it's impossible to turn all "no's" into "yes" for the inverse strategy.

In addition, buying/selling is not a binary action per se for stocks because you need to specify an amount.

Yeah there's just way too many possible ways to consider an inverse strategy. Even with your example, if my strategy is to buy GOOG at noon on Monday... Is my inverse strategy to buy 1 of every stock that isn't GOOG at Noon on Monday? Is the inverse to short GOOG? Is it to have already owned GOOG and instead sell it at Noon on Monday?

If your non-random strategy is buy GOOG and your 'random' strategy is buy shares of 5 companies chosen at random then inverse of your non-random strategy, buy everything except GOOG is better than 'random' because it's even more random.

One important caveat is that the non-random strategy would need to be invertible. For e.g. a ml model doing binary classification this is easy (switch the classification), for most other things it's not clear what invert means.

> the inverse of the non-random strategy is better than the random strategy

There is not one single non-random strategy. There are millions active at any moment. Which specific non-random strategy should we pick to assure success?

insider trading offers the best rewards/risk ratio.

I’d like you to expand on your first point further as it’s not trivial.

How do you define the "inverse" of a stock-picking strategy?

Isn't the audience for this individuals unqualified to be investors?

The UBC press release isn't as advertorial bad as the studyfinds.org piece, but they both seem to implicitly endorse lottery-playing by unqualified individual investors, IMHO.

The UBC release could've segued from "diversify", to pushing unqualified to people to low-ER total-market index investing, or to balanced target funds based on same. Instead, they're just helping pump unqualified people to gamble and be fleeced.

indexing is not random but a bet on the market as a whole.

Random is random. In a way picking more varieties may offer protection in a downcycle.

Most stocks go up but bigger stocks push indexes. Betting on bigger players is different than betting on any player to go up.

Another recommendation would be to take an approach outside of price-based time series models. Elastic has an article on this 'Generating and visualizing alpha...' https://www.elastic.co/blog/generating-and-visualizing-alpha...

Isn't this the whole premise of A Random Walk Down Wall Street.

How is this not the same conclusion as other studies that find that index funds out perform others?

In addition to diversification that reduces risk of overexposure to down sectors or typically over-performing assets, index funds have survivorship bias: underperforming assets are replaced by assets that meet the fund's criteria.

Yes its sad that every one jumps on index funds which can be a good idea but its not the right choice 100% of the time.

Isn't that mostly down to the low fees?

2 years after the ~2008 crash, I bought stocks in 4 companies that didn't yet recover yet from the crash... 2 of them doubled in value in just a few months and 2 of them went to basically zero... so I just about broke even.

A portfolio of four stocks is as good as a single random stock. Diversify more, there's a reason why the S&P500 is 500.

i think the utility of further diversification becomes pretty low after you have around 30 stocks in a portfolio. assuming those 30 stocks are actually diversified and are not e.g. 30 tech companies, or 30 companies in the UK.

I wonder if there's any advantage to picking 30 stocks versus just investing in some big index fund like VT or whatever.

There's a lot to be said in favour of investing into a low cost ETF that attempts to passively approximate the distribution of stocks as they appear in the market.

I think there are potential advantages to selecting stocks or weighting stocks according to criteria other than market capitalisation (unlike passive investment approach) but most people who try to do that do worse than a passive approach.

A passive investment approach like putting money into VT ETF depends upon some fraction of market participants doing the work to actively set market prices. If enough of the market goes nuts and bids up the price of some security far above its actual value (e.g. IPOs for businesses that run at a loss with no obvious path to making a profit) then the downside of a passive approach using market capitalisation as weights is that you will invest some tiny fraction in these companies with irrationally inflated market caps.

Part of it is also a question of how much time and mental energy you want to put into learning about how to value companies and investigating options for investment. Investing in 30 individual stocks vs 1 ETF also creates about 30x as much work at tax time...

Yes, variance is reduced by averaging over many uncorrelated variables. This sums up about 75% of investing articles.

May be? We have known for years (decades?) that beating the market isn’t exactly easy even for professionals. So why would amateurs? The only way to beat the market is to know more than the market.

Knowing more than the market could work - that seems like winning through skill. Perhaps a second (perhaps even a more likely) mechanism would be winning through luck?

Firms like renaissance have and continue to dramatically beat the market.

However, renaissance also has a collection of people that could probably produce a Nobel in physics if that's what they worked on.

"Why look for the needle in a haystack, when you can just buy the haystack?"

- Jack Bogle

The secret to wealth building can be found at bogleheads.org

This is a non result. Of course picking singular stocks yourself with little understanding of the financial market or the companies in question is a terrible idea. The real question is: Does this approach perform better than a maximally diversified index fund?

I read the article as a "just buy a low cost whole market index" - which seems to be the most common advice out there.

Assuming your randomness generator was fair and you bought enough stocks, your 'buy stocks at random' strategy, if you bought enough stocks, would eventually get you that maximally diversified index fund, I think?

I mean, I think the difference at scale is all in the weighting; if you bought one share of a random ticker on every throw of your random generator, you'd be overweight stocks with high per-share values vs. just buying, say, a basket of VTX, VXUS and VWO, which are weighted in more reasonable ways.

Not really, because they're only talking about amateur investors.

You’d need international exposure on your national exchange. Also by this logic multinationals should be solid investments but maybe they lack growth potential.

Isn’t it the same as an index fund?

Just not in altcoins. Just don't.

In public stocks there are always underappreciated future growth rockets. You can try to find them by analysis, at random, or by buying the entire index.

Altcoins don’t have growth fundamentals because nobody uses them, so at best you’re betting on holding a coin and selling it to a bigger sucker.

Considering that Bitcoin has very high and unreliable fees it's not good to buy things with. Instead Bitcoin Cash has gained a lot of adoption recently, for example all stores using BitPay also accept Bitcoin Cash. Also Monero is far superior in terms of privacy (and fungibility).

These are two examples with much better growth potential than Bitcoin, while being underappreciated by the masses.

I read the article and tried to read the linked study (paywall). But I see no mention of the 'random' part of this outside the introductory paragraph. In order for the headline to be true, wouldn't the study have had do asses some performance of 'random' picks?

Rather it seems that people were encouraged to take more 'risky' stock picks, and that led to some improved success, not anything to do with random.

I think the title and introduction are wrong, having extrapolated incorrectly from the study itself. I don't really know since I can't read the study.

But I think I strongly disagree with the premise. I think picking stocks at random is likely to lead to uninterested investors who may lose interest in frequent contributions to their stock accounts as they have no personal connection or awareness or interest in the location or use of their funds.

Yes, diversification is important, but so is awareness. Is it really better for the individual who may unknowingly invest in weapons manufacturing or a declining coal business despite their own morals and indicators of poor picks? I don't even think this was tested for in the first place.

If you're interested in the actual study, I found it here: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2956122&...

(If a signup form appears, just scroll down)

This is absolute malarky made to make people feel better about sub-optimal returns. There are definitely strategies which provide greater than 8-10% YoY returns, or random walk, or indexes. People that know them, rarely will share the full - but will occasionally share glimpses. It can be hard to do for very large portfolios, say $10B+, but not for smaller portfolios.

I'm not sure Warren Buffett would agree with you. In a 1975 letter to Katherine Graham, he wrote "If above-average performance is to be their yard stick, the vast majority of investment managers must fail. Will a few succeed — due to either to chance or skill? Of course. For some intermediate period of years a few are bound to look better than average due to chance — just as would be the case if 1,000 ‘coin managers’ engaged in a coin-flipping contest. There would be some ‘winners’ over a five or 10-flip measurement cycle. (After five flips, you would expect to have 31 with uniformly ‘successful’ records — who, with their oracular abilities confirmed in the crucible of the marketplace, would author pedantic essays on subjects such as pensions.)”

The whole letter is here:


Notice how whenever someone claims there are ways to consistently beat the market by some incredible amount, they never seem to be able to point to anything specific. It’s always “Ohhh, there are these hidden wizards who do it, and they (conveniently) don’t tell anyone the ways of their sorcery, but trust me, they exist in dark smoky rooms somewhere!! You’re just not invited.”

It’s like “I assure you, extraterrestrials exist, there’s just no visible evidence of them and they never show themselves. But they’re out there I swear!”

Not a good analogy because the people who hypothetically can best the market have incredible incentive to not share the details.

Sometimes they share how they beat the market in the past after their advantage faded, i.e. http://www.edwardothorp.com/books/a-man-for-all-markets/

Most recent one I heard was to day trade mean reversion for instruments within thirty minutes. No data or proof to back it up. Everyone's got an idea. What kind of glimpse are you referring to?

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact