Hacker News Reader

Comment by crucialfelix | Hacker News Reader

I did it for a while in the very exciting year of 2008. I see-sawed back and forth and was profitable for months at a time, but in the end I was down. I strictly kept my daily losses limited at $50/day. So really small risk. At the end of that year my day trading losses were about $1500 and otherwise I was in cash just watching the world burn. My long term funds lost more than that in a day.

I've lost out on more money by not being in the market long term going up.

Of course I thought I might get good at it (everybody dreams), but the main reason I did it was to learn to confront fear and to keep logical and strict to the trading plan in the face of it. I am still risk averse though, so I didn't really learn thoroughly.

The other main thing I learned is that if you aren't a robot then you don't have any business out on that killing floor. The age of human day traders is over.

View on HN

I traded a little during the crisis until I got hit with the pattern day trading rule. I cashed out up 28%. Maybe they were doing me a favor cause I probably wouldn't have done so well on a longer timeline.

So if you went in with an "algorithm" which baked in your own rules for trading and left it for a year, would it have performed better?

Does a greater proportion of non-HFT bots make money?

Developing that trading strategy is a continuous job; that's what successful traders do. They don't play stocks, they play strategies. There are going to be computers involved no matter what. Your indicators, your trading platform, what analysts you trust.

So what is a bot ?

A "bot" suggests to me something just trading in and out of some security at some average frequency: weeks, days, intraday, minutes, seconds and then there is HFT.

These things are all just feeding off of the movement of whales: major flows of money that cause those waves that small technical traders think they can figure out just by watching the shape of them.

There is really an amazing amount of intelligence involved in playing the day's game and you are taking in a lot of data, watching other markets (oil versus airlines, currencies, futures, sovereign debt) and using this to either realize the moment is now or to just scare the shit out of yourself and end the day up frazzled and $30 richer. That was my experience.

Successful traders reduce their risk even further by choosing what they are trading, shortlisting things that might pop and being able to jump on those (manually or automatically) in less than a heartbeat. That's where the bots will rip an entire sector up before an amateur trader has any chance to understand what is happening let alone why it happened. But those bots are just extensions of the trader.

Anyway, the whole thing is retarded and I have more fulfilling things to do on the planet.

> would it have performed better?

Definitely. Discipline is tough when somebody just demolished your chinese solar stock and you are down a week's earnings. The bot would've cut and run.

Thanks - I like the idea of strategies being an extension of the trader. Until we have fully independently trained NN-based trading algorithms, I guess there will always be domain knowledge being applied. (Even with NNs the domain knowledge will be applied in choice of layers etc.)

>Definitely. Discipline is tough when somebody just demolished your chinese solar stock and you are down a week's earnings. The bot would've cut and run.

What stopped you from cutting and running in that case? Was it the hope that the stock would swing back?

> What stopped you from cutting and running in that case? Was it the hope that the stock would swing back?

I believe you have to experience this yourself, since this is about emotions, bias, and insecurity. Everybody only has a sample size of one with this.

I can say that I found it very similar to playing Poker. In both cases you need to conquer your emotions and biases and simply follow the damn strategy. In both cases, the edge cases will make you constantly doubt your strategy.

Isn't it the case that essentially no traders beat the performance of index funds long-term?

It isn't, there's fanous people like Warren Buffet of course, but I personally know of 3 individuals in investing forums who post all their trades and handily beat indexes over long periods.

Since you mentioned him: http://finance.yahoo.com/news/buffett-most-mportant-investme...

> Nearly a decade ago, Warren Buffett made a million-dollar bet: that by investing in a completely unmanaged, broad-market low-fee index fund, he could beat the gains earned by a high-powered hedge fund with a team of managers at the helm. His opponent was Protege Partners, LLC, a New York City hedge fund with $3.5 billion in assets under management.

[...]

> His simple Vanguard S&P 500 (VFINX) fund has delivered returns more than 40 points higher than those of the hedge fund. “I believe this is the most important investment lesson in the world,” he said.

By the way, the bet is online at http://longbets.org/362/

Warren Buffet has so much money in his control that he cannot beat the market - his very investments change the value of the good companies he invests in - and not in his favor. He can make small bets of companies, but even if the bets all pay off (they will not - even great investors make mistakes) it isn't enough to be very significant.

His scale also works in his favor for some transactions like municipal bond insurance or paying cash to acquire Fortune 100 companies.

So he beat one particular hedge fund, that hardly proves that "essentially no traders beat the performance of index funds long-term."

http://www.cnbc.com/2015/06/26/index-funds-trounce-actively-...

> Pity the active fund manager.

> More dollars have flowed to index strategies that track a market benchmark, such as the S&P 500 index, partly because such funds typically have lower costs than active funds and more investors believe that stock-picking managers can't regularly beat the financial markets.

> Now a new Morningstar study, released this week at the Morningstar Investment Conference, finds that actively managed funds lagged their passive counterparts across nearly all asset classes, especially over a 10-year period from 2004 to 2014.

This evidence all suffers from the problem that you are trying to prove a very difficult claim. Information about averages won't help you here.

"Essentially nobody" is admittedly not well-defined, but I don't take it to mean that a handful of guys doing better would disprove it (and in any case have trouble finding any data points in favor of the opposite position).

I claim that the top 20% (as specified in the headline) could not be reasonably described as "essentially nobody". And the article you linked two posts ago found that more than 20% of funds beat the market.

That's over the course of a year, which is not what I would consider a long-term measure. Over the course of the year, sure, you could easily have people who beat the market.

Warren Buffett is not a day trader. To paraphrase, "My favourite holding period is forever."

He is a good data point to demonstrate that long term investors can beat the indices. Most don't, however, and it is my understanding that day trading is even worse. Just as there are professional poker players that can consistently make money, I'm sure there are some day traders that do very well, but it is not a skill that most people have.

He is also typically actively involved in companies he invests in, which sets him apart from somebody sitting around on e-Trade in a pretty big way.

People actually misstate the argument regularly. It is of course possible to beat the index over long periods, but there is little evidence that there is any skill that allows you to do it. That is, could those 3 individuals write down a prescriptive algorithm that allows someone to replicate their results ahead of time?

Another way to describe the problem is, given the same number of bots, randomly picking tades, as members of a trading forum, how many would beat the index? Is the number more than 3?

I think you also have to define "the index". There are dozens of indexes on the NYSE alone, and many come with legitimate questions about how well they actually represent the market. In that sense, I think it is possible to choose (and not luck into) a better basket of stocks than what goes into the indexes. Of course, it isn't easy.

If there's no evidence that beating the market depends on any particular skill, that would seem to be saying it's essentially a matter of chance. If that is the case then it stands to reason that, as the period of time observed grows longer, it becomes less and less likely for anyone to beat the market.

Link?

An online friend just got into currency trading and went into some tutorials about reading curves. Apparently there are recurring curve patterns, which allows one to predict where a rate is going. If you can code up some pattern recognition algorithm, there might be a way to make money.

Although I'm very skeptical about those patterns, because there could be someone trading back and forth to create those trends, and screw you when they want.

Anyway analyzing data and gaining money with it makes really no sense at all. Surely you can make money to live, but there is no point to it until you really are able to win big, and nobody does. Investing money where real business create economic activities that are in effective demand, maybe. But trading numbers, that's really an awful thing.

If those patterns existed and were detectable, the moment people started trading on those patterns (or wrote bots to do so), the pattern would disappear.

This is one of the services the market provides. It creates uniform price conditions across time and space. If there is a discrepancy that doesn't make sense (like a recurring profitable pattern in stock prices), you can make money by fixing it.

"Trading numbers", as you call it, is (through a layer or two of indirection) helping to establish true prices, transmit market information, provide liquidity, etc. All of these things are critical for "real business" to operate efficiently in a large economy. The abstract markets create huge value for "real business" and its users charge a small percentage of the value created.

I am a technical analyst. Patterns do occur but ultimately prices are driven by fundamental events. Interest rate decisions, market data releases and similar. When that occurs no pattern can survive on the smaller timeframe. Technical analysis works but it is not written in stone.

What are your thoughts on A Random Walk Down Wallstreet? He pretty thoroughly picks apart the efficacy of technical analysis vs buying some indexes with supporting data.

> Technical analysis works but it is not written in stone.

Works 60% of the time 100% of the time.

Haha, kinda. My daily audit shows that I am on average 63% right. It comes up to cutting your loses early and riding out the profits.

I've tried trading on patterns for a short while.

Problem is, those patterns do exist, but they are chaotic behaviors, and transition probabilities are already priced in. So that, for a non-causal analyst looking at past data they look very reliable, but for somebody with no forward knowledge, they are useless.

How else beyond being a precog and insider trading could one have forward knowledge?

By getting your data with days after the fact, and looking at the patterns with all the data graphed.

Or, in other words, the way those market analysis courses and gurus teach people to read the market. The patterns are very reliable if you ignore causality.

I might be misreading what you are saying, but how is getting data days after the fact forward knowledge?

Past performance is no guarantee of future results.

Reply on news.ycombinator.com