Hacker News Reader

Comment by Bostonian | Hacker News Reader

The article mocks technical analysis by calling it "astrology", but there is much academic research supporting the use of cross-sectional and time series momentum. A simple way to decide whether an asset is an uptrend is to compare its latest price to its N-day moving average.

A site that lets you test moving average strategies is Portfolio Visualizer -- a backtest of the 200-day moving average crossover on an S&P 500 index fund is at https://www.portfoliovisualizer.com/test-market-timing-model... . The result is that a trend-following system, compared to buy-and-hold, has had slightly lower returns, 10.57% vs. 11.10%, but substantially lower volatility, 11.48% vs. 15.22%, and thus a higher Sharpe ratio, over the period Jan 1985 - Aug 2022.

View on HN

But there are tons of trading firms using very sophisticated mathematical modelling. Surely a signal as simple as this is already widely used and therefore, the Alpha has been largely arbitraged out already?

The high frequency trading firms are interested in strategies with Sharpe ratio of 3 and above, not the 0.10 incremental Sharpe ratio for the strategy I linked to.

Bostonian, portfoliovisualizer and approach you shared is very interesting. I am a new comer in this domain, can you share some good resources to get started? Courses / Books that share recipes can help me.

This. If there would have been any Alpha to gain by dead simple, well known pattern recognition, it would have been exploited very long ago already.

By now Technical Analysis is all about reading meaning into white noise (that’s overlaying fundamentals and a general light upwards trend).

Well there is historically quite solid Alpha to be gained from the dead simple strategy of buy and hold. That's just another dead simple strategy after all, yet plenty of people recommend it.

Because the long-term trend is always upwards. Trying to take advantage of volatility in the short term is completely different, it’s a zero-sum game.

> Because the long-term trend is always upwards

That depends on your model. (Because it is your model that defines "long term")

It is possible for markets to trend downward for decades at a time.

That's a good strategy, but it's not alpha, since alpha is defined as outperformance relative to a passive strategy.

Buying and holding the market gets you the market return, which by definition has zero alpha.

Well the voo has consistently outperformed vtsax when bought and held so I feel my point stands. A strategy is a strategy. You have to choose one when you invest no matter how simple or complex.

This is frankly a reductionist, borderline demoralization post.

"Alpha" is not just some blind thing that happens when you pick <popular_stock> and use it. Yes, in most cases "easy" strategies have been arbitraged out in well covered highly liquid issues.

However, there is significant alpha even in simple trend following strategies in certain time frames, markets, and liquidity. If it wasn't the case we wouldn't have independently verified traders bringing in excess alpha using such simple systems. Dramatic changes in the volatility regime of even a highly liquid issue can make a simple strategy profitable. Timing is often more important than being exactly right. In fact relative to position size a highly active simpler system will tend to produce excessive Sharpe.

The market you trade is more important than the system you use in a lot of cases. Retail traders can generate significant alpha by simply trading where the hedge funds can't see. In fact I've traded a simple seasonal strategy with significant upside profit in several markets using statistical arbitrage. This type of arbitrage is completely captured by variations of the Bollinger band.

Strong EMH is empirically disproven. My post is not a cope post. I would suggest doing more research rather than grouping all methods of technical analysis into the astrology category. Often times these expensive "top tier" quantitative models are nothing more than more complicated variations on moving averages (ARCH/GARCH, etc).

> In fact I've traded a simple seasonal strategy with significant upside profit in several markets using statistical arbitrage.

Statistical arbitrage worked well when it was first discovered then faded away.

It is true that if you discover a novel statistical technique, and keep it secret, you can do very well, at enormous risk, for a while.

Look at the history of "pairs trading"

But not knowing you, when you traded or the actual algorithms yu used etcetera I suspect you were trading in a rising market. Beating a rising market is easy, as is beating a falling market. Problem is, you do not know what you are in today until tomorrow.

Here is some advice for free: Buy and hold. Indexed funds. Get rich by working hard.

> Buy and hold. Indexed funds. Get rich by working hard.

I don't trade for the money I trade for the challenge. The money is just a point system I use to see if I am winning. The tone of your reply makes me think I am considering myself an expert - which I am not. I am pretty well read on various trading methodologies, the papers being produced about trading, and have quite a lot of time in seat across varying markets including most recently the futures market during the onset of COVID. After paying my tuition to Mr. Market several times I've gained some experience that makes me money.

As for pairs trading that is exactly what I did. I've done it every season for years in several markets both equities and futures, and it has absolutely no correlation with a rising or falling general market. Though I have occasionally lost money when the legs go out of sync and, being a retail trader, do not have prime execution. But alpha is alpha, and one example is enough to disprove OPs conjecture that all technical analysis is bunk.

I can't tell if you're talking down at me or not so I'll be charitable and say I think B&H is great and I do it all the time. In fact, 99% of my money is tied up in tax advantaged B&H. However, I enjoy trading like I enjoy poker. So I spend some of my money on it. It's one of the few places you can get a challenge that tasks you in several different areas and it's better than donating to yet another 501 (c) 3 with a CEO that takes down a questionably large salary.

> tax advantaged B&H

What is a tax advantaged Buy and Hold?

Shoving stuff in a 401k, IRA, SEP-IRA, etc and leaving it. As opposed to a traditional brokerage account where capital gains are charged yearly. I called it that so I didn't have to list out all the different vehicles :).

Swap out your S&P 500 fund for a random large cap fund (say VLCAX) and, whoops, worse CAGR and worse Sharpe ratio over the same period for your strategy.

You can prove you would've made money by backtesting anything.

> there is much academic research supporting the use of cross-sectional and time series momentum

Yes. It gets you published. and...

No. It is nonsense. I have researched TA. My conclusion is that the efficient market hypothesis, in its weak form, is true.

There are a lot of cognitive biases at work and a lot of wishful thinking.

All information has a cost, that cost is factored into pricing, but it is cheaper for some than for others. That explains some of the effects you see.

Another is trading costs. Transaction costs for most actors in the market are as high as they have ever been (it is a deliberate policy of financial firms, and many costs are hidden, but those gleaming skyscrapers on Wall St? Transaction costs). When yuo see a simulation of TA my advice is to calculate the minimum transaction cost for profit, and then compare that to what is actually available (Transaction costs often reach 200 basis points, and the cheapest I found were in German markets in the nineteenth century. A damning indictment on modern finance)

The efficient markets hypothesis does not, and has not, stood up to empirical research since it was first invented. The reason that it's a hypothesis, and not a theory, is that it's impossible to disprove because all of evidence that disagrees with it is dismissed as spurious.

> he efficient markets hypothesis does not, and has not, stood up to empirical research since it was first invented

Yes it has. In its weak form.

More pointedly all the supposed counter examples turn out to be snake oil. People really want to believe there is some magic that can predict the future....

> the reason that it's a hypothesis, and not a theory, is that it's impossible to disprove

It is an hypothesis. It is impossible to prove it, only disprove it. And....

> because all of evidence that disagrees with it is dismissed as spurious.

All the counter examples turn out to be snake oil. I spend years looking, repeating simulations, back testing algorithms.... Snake oil each and every time

People really want to believe....

Momentum, value, and size are all predictive of future excess returns. If the weak form were true this would not be the case. And those are just dead simple buy-and-hold strategies that anyone can implement.

Transaction costs have come down a lot in the past 20 years. 10-20 basis points is probably reasonable for a retail trader with a good broker. Double that for a round trip trade.

Any other working ways except of moving average? As a beginner trader, for me this seems the only working part of "astrology".

As another commenter said, the moving average has already been arbitraged out. You can now study the moving variance, but it will have also been… Then you can look at the kurtosis of covariances… At some point something will be new. That is astrology.

Reply on news.ycombinator.com