Random investing?(!)

Ask your investment, budget, and other money related questions here
Post Reply
shadow
Posts: 13
Joined: Sun Aug 19, 2018 1:18 pm

Random investing?(!)

Post by shadow » Fri Nov 02, 2018 9:10 am

https://arxiv.org/abs/1303.4351

This is pretty far over my head but maybe someone here can start a discussion I can hope to understand :lol:

firerufus
Posts: 4
Joined: Thu Oct 11, 2018 3:35 am

Re: Random investing?(!)

Post by firerufus » Wed Nov 07, 2018 10:09 am

Three things I've "learned" recently might contribute or explain this theory:

- Efficient Market Theory: Every information is already priced in. Future prediction are just guesses. A stock today is already priced as it should be.
https://www.investopedia.com/terms/e/ef ... thesis.asp

- Random Walk: Stocks don't depend on yesterdays movement and act completely random over a long period of time
https://www.investopedia.com/terms/r/ra ... theory.asp

- Regression to the mean: Over a long period of time, "everything will balance itself out". All outperformer are statistically underperfomer in the future and vise-versa. https://en.wikipedia.org/wiki/Regressio ... d_the_mean


You can do some research around those topics. All of this basically means that stock picking doesnt work over a long period of time and ETFs (and new: Factored ETFs) are performing better.

I am not biased just reading tons about it lately. I have yet to make up my own mind about it.

ThisDinosaur
Posts: 964
Joined: Fri Jul 17, 2015 9:31 am

Re: Random investing?(!)

Post by ThisDinosaur » Wed Nov 07, 2018 10:45 am

I cant seem to download the paper, but I would not be surprised if random investing outperformed technical analysis.

My opinion is that large markets (many participants, many options of what to buy, lots of total capital, frequent trading) are basically efficient most of the time. All publically available information is priced in.

But people are also herding animals and we sometimes follow each other off a cliff like lemmings. The history of speculative bubbles proves markets can be stupid.

User avatar
daylen
Posts: 621
Joined: Wed Dec 16, 2015 4:17 am

Re: Random investing?(!)

Post by daylen » Wed Nov 07, 2018 11:01 am

No need to "randomize". Just increase diversification to decrease volatility; easy with indexes. Generally, the diversity of your portfolio should be correlated with your ignorance.

Technical analysis is ignorant of the random walk hypothesis. Assuming structure that does not exist often hurts prediction.

Also look at maximum entropy principle. It is best to choose a prior distribution that maximizes entropy (fits your ignorance) when predicting

jacob
Site Admin
Posts: 10442
Joined: Fri Jun 28, 2013 8:38 pm
Location: USA, Zone 5b, Koppen Dfa, Elev. 620ft, Walkscore 73
Contact:

Re: Random investing?(!)

Post by jacob » Wed Nov 07, 2018 11:04 am

The OP result is not unexpected. https://www.amazon.com/gp/product/0470008741/ is a classic. He did about the same thing and found nothing.

Trading models that actually work in reality are never made up of just one signal. The general observation (from having worked in the field) is that it's a particular combination of 5-7 different ideas. The secret sauce is in the combination (a structure) and not in the ingredients themselves (no structure).

Because academics and retail amateurs only have access to a very limited set of data and/or analysis tools, they're prone to committing type II errors. Basically, the inability to find a signal by mashing together some R-code and writing a paper does not prove there's nothing there.

If you look a lot deeper, you begin to find things: https://www.amazon.com/Non-Random-Walk- ... 0691092567

ThisDinosaur
Posts: 964
Joined: Fri Jul 17, 2015 9:31 am

Re: Random investing?(!)

Post by ThisDinosaur » Wed Nov 07, 2018 11:33 am

daylen wrote:
Wed Nov 07, 2018 11:01 am
Generally, the diversity of your portfolio should be correlated with your ignorance.

Technical analysis is ignorant of the random walk hypothesis.
Then my ignorance has been increasing over my investment career.

I dont think tech analysts are ignorant of random walk. I think they have counterarguments to it. Like that their is signal within the noise. Or that the "random" part is in how the news/information comes out, not in how it propagates through the market's prices.

I have no doubt their are trading strategies that "work." My position is that they typically dont increase return enough to justify the effort. S-curves, etc.

black_son_of_gray
Posts: 325
Joined: Fri Jan 02, 2015 7:39 pm

Re: Random investing?(!)

Post by black_son_of_gray » Wed Nov 07, 2018 11:44 am

Relevant read with respect to math- and algorithms-based strategies: https://www.institutionalinvestor.com/a ... ath-Hustle

User avatar
daylen
Posts: 621
Joined: Wed Dec 16, 2015 4:17 am

Re: Random investing?(!)

Post by daylen » Wed Nov 07, 2018 11:46 am

Well human behavior is often irrational. :P ..or there are more factors at play in your case (such as risk aversion, laziness, change in utility/loss function, etc).

The part about technical analysis was not an ad hominem, since I was referring to the technique being ignorant. Good point though.
Last edited by daylen on Wed Nov 07, 2018 5:30 pm, edited 3 times in total.

ThisDinosaur
Posts: 964
Joined: Fri Jul 17, 2015 9:31 am

Re: Random investing?(!)

Post by ThisDinosaur » Wed Nov 07, 2018 1:50 pm

black_son_of_gray wrote:
Wed Nov 07, 2018 11:44 am
Relevant read with respect to math- and algorithms-based strategies: https://www.institutionalinvestor.com/a ... ath-Hustle
If you're gonna invest, you have to have some way to choose an approach. If your mind thinks in math, algorithms make sense. But it's not the only way. Buffett and Munger have said they never use anything more complex than high school algebra.

IlliniDave
Posts: 2204
Joined: Wed Apr 02, 2014 7:46 pm

Re: Random investing?(!)

Post by IlliniDave » Wed Nov 07, 2018 4:49 pm

Flipping a coin is not truly random--with perfect knowledge of all pertinent information the coin's movement can be anticipated perfectly including which side lands up. But that doesn't do a football player much good when he's asked to "call it in the air".

Attributing randomness is often just a hack when it is too difficult to juggle all the possible variables. Processes when viewed from the outside can appear random when they aren't. Making assumptions of randomness is often the most pragmatic approach to a situation.

It's not a surprise that there is perhaps some signal in the noise being exploited by secretive folks deep in underground bunkers below large financial buildings. And if the signal is persistently valuable for exploiting via trades those people will soon have all the wealth in the market. If not, well, they'll at least collect nice salaries.

I can't comment on the OP because I don't know what is meant by random trading. I'm not a trader, more into the buy/hold strategy. It's fun though, to daydream about finding the magic money faucet algorithm!

jacob
Site Admin
Posts: 10442
Joined: Fri Jun 28, 2013 8:38 pm
Location: USA, Zone 5b, Koppen Dfa, Elev. 620ft, Walkscore 73
Contact:

Re: Random investing?(!)

Post by jacob » Wed Nov 07, 2018 5:08 pm

IlliniDave wrote:
Wed Nov 07, 2018 4:49 pm
And if the signal is persistently valuable for exploiting via trades those people will soon have all the wealth in the market. If not, well, they'll at least collect nice salaries.
Inefficiencies are limited in size (e.g. $300k or $25M) and time as more people learn to exploit or avoid [creating] it. The secret bunker people operate this as any other business. Lets say we're dealing with fairly complex inefficiencies that requires a research team of three people (a trader, a coder, and a quant) 12 months to research and develop a trading strategy. Then you have 3 people at 50-100k each plus 200% in overhead. Trading capital is cheap. The basic economics indicates that they should only look at inefficiencies that are at least ~1M in size. If it's smaller, then it doesn't pay and the inefficiency is left alone.

Suppose it's larger. It might be that it's created by a pension fund or a bank that need to move $100M per year. Of course they don't like to lose $1M to traders. OTOH, they don't want to hire traders on their own and avoid creating the inefficiency themselves. However, if they were consistently losing enough to make it worthwhile to remove it, they would. A company like Vanguard would have an entire trading team devoted to ensuring that they're not picked off when they have to buy or sell the individual stocks when the pension or DCA funds roll in or out. For boring details on how this works, see e.g. https://www.amazon.com/Algorithmic-Trad ... 956399207/

It's basic economics.

Normally, the size of inefficiencies are not known in advance. This goes to the point of what I quoted:

Inefficiences are NOT persistent in the market and they do not scale infinitely. This is why those [bunker] people will never have all the wealth in the market even if they will have a significant fraction of it.

What happens in practice is that one starts a strategy with "small" amounts and then trades for a while (lets say a week or 5-10 times). The profit is noted. Now the trading amount is increased by 50%. Again the profit is noted. If the profit rises 50% as well, the bet is increased again. This time the profit might only rise 35%.

This way you develop a bid vs profit curve. It will be linear at the bottom (bets ~ 0) then begin to bend off and actually begin to decrease. You want to find the peak here and trade with that sized bid. You can't bid more because you're essentially quenching the inefficiency.

Once you've maxed out the ineffieincy for a while, the inefficiency will start changing because other people get involved. This will change the curve. You either have to figure out a new curve ... or tweak your model.

...

By random trading the OP authors simple changed the long or short orientation randomly and compared it to different technical signals. You can hack this out in about 20 lines of R-code. Real algorithms are MUCH more complex than this, but the OP paper is good for pedagogical purposes. It's something I would have an intern fiddle with. It's not hard to make but not really useful either.

User avatar
7Wannabe5
Posts: 4057
Joined: Fri Oct 18, 2013 9:03 am

Re: Random investing?(!)

Post by 7Wannabe5 » Thu Nov 08, 2018 6:09 am

Yeah, like how you can only make a profit by moving end lot Ralph Lauren sheet sets from the clearance area at TJ Maxx to the Amazon Marketplace for about 9 months until somebody else catches on.

jacob
Site Admin
Posts: 10442
Joined: Fri Jun 28, 2013 8:38 pm
Location: USA, Zone 5b, Koppen Dfa, Elev. 620ft, Walkscore 73
Contact:

Re: Random investing?(!)

Post by jacob » Thu Nov 08, 2018 7:55 am

Yes, or how the best way to eat in the long run is to buy food according to the shelf space allocation at Walmart (it has the lowest fees) because all the consumers have determined both the correct prices and the correct size of the shelves. I don't know what good food is, [therefore] nobody knows, and so it's basically a waste of time trying to pick your own food. In fact eating at restaurants will cost you a fortune in fees and wages. There are also indications that simply pulling food down randomly from the shelves will beat a trend-based diet. Eating a well-diversified diet will assure you get all the nutrients you need.

Yes, exactly like that.

Dream of Freedom
Posts: 219
Joined: Wed Aug 29, 2012 5:58 pm
Location: Nebraska, US

Re: Random investing?(!)

Post by Dream of Freedom » Thu Nov 08, 2018 12:30 pm

Market action is produced by man and machine neither of which seem to act completely randomly.

Mankind's behavior is dictated by instinct, emotion and to a lesser extent thought. Our instinct is relatively static with epigenetic changes only happening with changes to one's environment. Emotions too change little over time. Studies of both amputees and lottery winners show that life satisfaction returns to what it was within a year for instance. Thought is the most random of the three but also the least impactful.

Machines do what they are programmed to do. Even when some part of the code is random they still choose within a range say 0-50 or from a list. So even the randomness isn't completely random.

So if the market participants are not acting randomly why would the market be random? And if participants act consistently that would create behavior patterns. If you were to graph that behavior into a chart you may well see those behavior patterns in the chart.
Last edited by Dream of Freedom on Thu Nov 08, 2018 12:51 pm, edited 1 time in total.

User avatar
daylen
Posts: 621
Joined: Wed Dec 16, 2015 4:17 am

Re: Random investing?(!)

Post by daylen » Thu Nov 08, 2018 12:48 pm

Dream of Freedom wrote:
Thu Nov 08, 2018 12:30 pm
if the market participants are not acting randomly why would the market be random?
Not sure if this is a serious question or not, but IlliniDave gave a good answer. True randomness has never been proved and can never be proved in any system. Randomness is an assumption that makes prediction computationally feasible.

Peanut
Posts: 327
Joined: Sat Feb 14, 2015 2:18 pm

Re: Random investing?(!)

Post by Peanut » Thu Nov 08, 2018 1:38 pm

https://www.nytimes.com/2018/02/02/your ... ading.html

Thought this was somewhat relevant. Confirms my own experiences in retrospect. My own answer to the “mystery” of overnight vs daytime gains is that many companies like Amzn tend to report earnings after hours, and those reports tend to effect big movements in stock price.

The Old Man
Posts: 317
Joined: Sat Jun 30, 2012 5:55 pm

Re: Random investing?(!)

Post by The Old Man » Thu Nov 08, 2018 4:30 pm

daylen wrote:
Thu Nov 08, 2018 12:48 pm
Not sure if this is a serious question or not, but IlliniDave gave a good answer. True randomness has never been proved and can never be proved in any system. Randomness is an assumption that makes prediction computationally feasible.
Remember, Random Number Generators are not random. They have an algorithm. Ideally, the results are indistinguishable from a random distribution, but know the seed and algorithm and it becomes very predictable.

User avatar
daylen
Posts: 621
Joined: Wed Dec 16, 2015 4:17 am

Re: Random investing?(!)

Post by daylen » Thu Nov 08, 2018 4:44 pm

That was partly the point I was trying to get across in a less direct way.

Strictly speaking that is not knowable, because there is no way to know if reality itself is random. I cannot fathom a random existence, but I remain agnostic. Then again, bringing this up renders my argument meaningless.

shadow
Posts: 13
Joined: Sun Aug 19, 2018 1:18 pm

Re: Random investing?(!)

Post by shadow » Thu Nov 08, 2018 8:35 pm

jacob wrote:
Thu Nov 08, 2018 7:55 am
Yes, or how the best way to eat in the long run is to buy food according to the shelf space allocation at Walmart (it has the lowest fees) because all the consumers have determined both the correct prices and the correct size of the shelves. I don't know what good food is, [therefore] nobody knows, and so it's basically a waste of time trying to pick your own food. In fact eating at restaurants will cost you a fortune in fees and wages. There are also indications that simply pulling food down randomly from the shelves will beat a trend-based diet. Eating a well-diversified diet will assure you get all the nutrients you need.

Yes, exactly like that.
So, carrying on the metaphor, you are saying the best option is to learn how to eat?

But as daylen said, given that an average investor doesn't have access to huge projects exploiting massive inefficiencies, doesn't it make sense to diversify to make up for lack of knowledge?

Or is the solution to go for smaller inefficiencies that would not be worth a large investment for developing a strategy?

Post Reply