Comment by meeper16 | Hacker News Reader

Comment by meeper16 | Hacker News Reader

Parent Full thread

Let me correct you.

Indicators based on statistical analysis can't be argued with in terms of their value in providing additional insight into the present condition of a stock or index. Think in terms of moving averages, a common application in TA. http://mathworld.wolfram.com/MovingAverage.html

(BTW: Wolfram and mathematica do not deal in astrology)

For example, indicators involved with technical analysis include, lagging indicators or those that show overbought or oversold conditions for stocks or entire indicies such as the Moving Average Convergence/Divergence ratios or MACDs.

https://en.wikipedia.org/wiki/MACD

If you are a financial advisor, you should be using these to make sure that when you advise on taking a postion in a stock for example, the stock has not spiked up 10% on the previous day. If the stock did spike the previous day or week and as a financial advisor, if you advised your client to take a position in that stock, you'd effectively be placing them in a position of risk as the stock is likely to retrace. TA indicators are very practical in this case. This is just one of many examples.

You're getting your robust methods of statistical analysis confused with astrology.

Statistician here with experience in the financial world: you're wrong. Don't confuse the existence of Wikipedia pages and books written by people wanting to make money with actual science.

Moving averages have a function in statistics. But there is no evidence that they can be used to time the markets. If they could, it would be done algorithmically by hedge funds, closing the possibility.

Please show actual analysis of market data (i.e. recent published scientific papers) supporting your positions. Without those, it's quackery.

Hedge fund algorithm developer here. You misread:

1. The existence of mathematica was quoted first.

2. Moving averages are a "lagging indicator", which I stated, not a predictive indicator which I did not state.

3. I develop hedge fund algorithms.

References to papers include this most important one: "Contagious Speculation and a Cure for Cancer: A Non-Event that Made Stock Prices Soar," https://www0.gsb.columbia.edu/faculty/ghuberman/research.htm...

> 3. I develop hedge fund algorithms.

Do the hedge funds know? ;P

Our bank accounts do.

Until they don't. As a group, hedge funds far underperform index funds.

Algorithms degrade, we optimize. We're in the top 25. They last longer than more full time jobs.

I've been working in hedge funds for over a decade, and it's interesting what the variety of opinions is. There's a great deal of secrecy, and good reasons for it, so getting to know how other quants do things has been quite informative.

On the one hand, your arbitrage argument makes sense. If there existed a simple strategy that made money, it would be traded out of existence.

On the other hand, I happen to know how the major quantitative funds allocate their money. It is a simple formula, with a number of small twists. (Yes, people will not believe you if you claim this. Rightly so.)

Now, how can these facts coexist? The answer is that the formula "sort of works". There are long periods where it doesn't work, and periods where the returns are phenomenal. If someone were to start a new fund, they would have a number of issues:

- It is just not quite believable that such a simple formula is profitable. What are those hordes of phds actually doing? (Squeezing a lemon actually). Surely someone else would have found it?

- Institutional investors have all sorts of issues with investing. It's done by committee. They need you to have a long track record. They want all the right boxes checked.

- What happens if you launch and you start with a fallow period?

- Even if a simple formula works, why does it work? Are you happy to trade something that you don't really understand?

I've been on the hedge fund algo side for years. You're spot on and it's what a lot of these guys on this thread are clueless about.

Basic or simple approaches can be combined with others and be statistically/mathematically sound and robust.

Some work for periods of time, some don't, some stop working. Algorithms degrade over time and that's continued optimization is a must.

Case and point: LTCM and the story behind black scholes algo running up against a black swan.

I'd highly recommend watching this for the others reading this thread that really just are familiar with the whole space:

https://www.youtube.com/watch?v=lmvxZgnwwD4

> black scholes algo

Black Scholes isn't an "algo". Sweet mother of god...

No need in explaining the difference between an algorithm that wraps around an equation or to call an equation a "model" or to call a model a "formula". Take your pick but focus on the meat of the discussion if possible.

Dude, I work in risk on the sell side. If anyone in banking seriously mentions TA they get laughed at at best. Bad idea to bring it up in an interview though.

The only field where a sort-of technical analysis may be mentioned is HFT, but that's due to the fact, that HFT works on such a microscopic time frame that it is difficult to come up with conventional data used in financial analysis (example: correlation). So it's more of a tech arms race than anything else.

Just because TA uses mildly advanced math, doesn't mean it's any good. Just like it doesn't mean that heroin is good for you, just because the producers use sophisticated chemistry equipment.

Technical analysis is mostly useless for the kind of services financial advisors provide - namely long-term investing.

Example: If, at some point, the advisor looking at Google, which just spiked 10% from $80 to $88/share, passed it over, he did his client a huge disservice, because the stock is at $600/share, say 10 years later.

Technical analysis may be somewhat useful in short-term trading (speculating, rather), but its overall value has been way overblown.

For more sophisticated investing, you need a multi-factor model - a mixture of fundamental analysis, industry/macro indicators, and (perhaps) a touch of technical analysis.

I agree, combining TA with Sharpe Ratios for example.

I've also protected portfolios from showing close to zero gains by using basic stats and TA principles available to anyone, to avoid creating positions in overbought stocks. While also sometimes overweighting on stocks that they are keen on buying, that have either spiked down or have shown significant retracement, oversold, days before.

Reply on news.ycombinator.com