HFT, and is backtesting helpful or harmful?

[d08]

I think a lot of new guys put the cart in front of the horse. The idea of coming up with a magical system that the computer can test and trade for you makes new guys see dollar signs. My world isn't that big, but I don't know anyone who has succeeded by starting with system design before learning to trade first. I suggest everyone learns to trade with tiny size...there is no substitute for experience.

That has more to do with generations as mechanical system testing has become easy in the last 10 years or so, both the amount of tools available and the software cost. Nowadays, most people seemingly start with automation rather than discretionary chart watching - this isn't necessarily a good thing because I think there is a certain benefit to hand-drawing charts and over-analyzing a limited dataset.
I envy discretionary traders as they are more dynamic than system traders yet not necessarily more successful as they often can't take advantage of diversification (human attention span limitation).

Someone that is capable in automation could probably write systems based on what you do, not everything that you do can be turned into a signal based system but I suspect most could be with significant effort.

HFT is like the mafia, taking its "tax" from everything traders who are trying to establish real positions do. The only "solution" I've found is to initiate positions at random intervals with market orders, at least you won't be advertising your size for all the HFTs to properly abuse you (I'm being naive here).

 

[Sergio77]

Primarily the spread and commissions affect daytrader performance. I think this is a must read even if you do not totally agree. I found it very useful.

Backtesting is like a war PC game. Most people play only the game on their PC, only a few manage to survive in a real game.

 

[jb514]

Yes Dustin it's very beneficial to new guys. Our firm has put a handful of guys who new next to nothing about trading on backtesting software and they were able to create strategies that were/are profitable in live trading. I was skeptical of backtesting in general mostly because of what I had read on elitetrader but seeing it pan out into profitable trading has really changed my mind.

 

[actionzip54]

Yes Dustin it's very beneficial to new guys. Our firm has put a handful of guys who new next to nothing about trading on backtesting software and they were able to create strategies that were/are profitable in live trading. I was skeptical of backtesting in general mostly because of what I had read on elitetrader but seeing it pan out into profitable trading has really changed my mind.

What if the systems start to fail as a market environment changes? Will the new traders have really learned anything...

 

[StarDust9182]

1) Has your trading been consistently profitable?

2) If so, for how long and how profitable (specifics, please)?

No offense, but I've never met a discretionary short-term trader who could prove their profitability via account statements or results from a third-party site. At least not for the long-term. Backtesting can have pitfalls and be worthless without proper safeguards, but it's far better than thinking you can outsmart the market on your own. Pretty much every discretionary trader I've met is a walking example of behavioral finance fallacies (overconfidence, anchoring, confirmation bias, etc.).

1) When I went for consistency, I ultimately dropped all backtesting and holy grail searches. In my early years, consistency was not what I sought, but I did change my thinking to consistency methodology about 15 years ago. I will let you decide if I am consistent.

In three decades of trading, I have had only 2 calendar years in a row with losses, 2008 was a loss (about 3% - due to professionals helping me! my own share was very tiny profitable) and one losing year that I remember around 1990. That is 4 years of 30. Every other calendar year has been profitable. My worst losing year was 4% (2008) and my best winner was 30% plus. My two years in a row losing were caused by a traumatic event with a private equity investment and my refusal to stop completely and fix that issue first. (ie my own psychology). (None of my numbers contain interest or dividend income and are all after fees) Excluding options I have never used any leverage. Position trading mostly but all time frames and multiple methodologies.

It is funny now but in the 1980s, nobody told me that this could not be done. My own ignorance worked out well. It wasn't until 1999 that I dug deep into systems and theory that I started getting into trouble, but I learned and became what I am today. I no longer tell people about my returns since nothing makes them angrier than when I make money and they lose it. If I lose it and they make it, they love that. So that is what I tell them, since I don't wish to argue with them.

I will never prove this to anyone. However, if I did, I would simply show my income tax records. That would nail it PERIOD. They go back for decades. My goals have changed over the years and now I have dropped my returns and increased my consistency by doing so. That is a conscious choice BTW.

Curious, that I have never met a retail system trader that was profitable over a decade either. Perhaps we can teach each other things. Your commentary is not how I look at things. I have met hundreds of traders and watched them die for different reasons in the markets. Most of them knew more than me.

Is that consistent?

 

[Grandluxe]

1) When I went for consistency, I ultimately dropped all backtesting and holy grail searches. In my early years, consistency was not what I sought, but I did change my thinking to consistency methodology about 15 years ago. I will let you decide if I am consistent.

In three decades of trading, I have had only 2 calendar years in a row with losses, 2008 was a loss (about 3% - due to professionals helping me! my own share was very tiny profitable) and one losing year that I remember around 1990. That is 4 years of 30. Every other calendar year has been profitable. My worst losing year was 4% (2008) and my best winner was 30% plus. My two years in a row losing were caused by a traumatic event with a private equity investment and my refusal to stop completely and fix that issue first. (ie my own psychology). (None of my numbers contain interest or dividend income and are all after fees) Excluding options I have never used any leverage. Position trading mostly but all time frames and multiple methodologies.

It is funny now but in the 1980s, nobody told me that this could not be done. My own ignorance worked out well. It wasn't until 1999 that I dug deep into systems and theory that I started getting into trouble, but I learned and became what I am today. I no longer tell people about my returns since nothing makes them angrier than when I make money and they lose it. If I lose it and they make it, they love that. So that is what I tell them, since I don't wish to argue with them.

I will never prove this to anyone. However, if I did, I would simply show my income tax records. That would nail it PERIOD. They go back for decades. My goals have changed over the years and now I have dropped my returns and increased my consistency by doing so. That is a conscious choice BTW.

Curious, that I have never met a retail system trader that was profitable over a decade either. Perhaps we can teach each other things. Your commentary is not how I look at things. I have met hundreds of traders and watched them die for different reasons in the markets. Most of them knew more than me.

Is that consistent?

We are talking about short term active traders who trade for a living.

 

[jb514]

What if the systems start to fail as a market environment changes? Will the new traders have really learned anything...

I think your question applies equally to automated/discretionary traders. They really aren't in a different position compared to discretionary traders. Everyone is going through the same process of analyzing the previous results, but with backtesting/automation it's a little more cut and dry. It's also not much of a sit back and watch game. People are constantly tweaking strategies. At least for some strategies the line between discretionary and automation is starting to blur.

 

[Satan's Helper]

Excuse me??



What mood??



What general mechanics?? Markets go up, down and sideways, that's it. That's all they have been doing for centuries.



Give your backtest sofware 20 years of historical data and it will immediately tell you if your trading idea is profitable or if you are just kidding yourself (and making your broker richer), simple as that.



Sure you can do that too, but then you need to demo test for at least 10 years before you can draw any statistically accurate conclusions about the profitability of your system. Do you really want to demo test until 2024?

Because only total amateurs demo test for a month and then say "Hey look, my system is working".

Back testing only works well if your software takes into consideration the bid/offer data to give a accurate idea of fill prices. Many times a strategy looks great on a back test but in reality the slippage during those events eats all of the profits.

An example of mood is a crises brewing overseas and trading desks pushing any strength to get the prices high enough to to short to their customers. What looks like a strong move is a trap set up to get liquidity.

Mechanics are how order desks execute institutional orders. Understanding the system and how it works is a huge advantage over statistical analysis. So many people get caught up in analyzing price and they forget it is people who really drive the price. Trading systems only magnify those moves.

If you understand order flow then you don't need to back test or run a strategy for a long time. That is why the best traders typically worked on a order desk or back in the day the trading floor. That is where you learn the fundamentals of trading and how things work.

 

[StarDust9182]

I think your question applies equally to automated/discretionary traders. They really aren't in a different position compared to discretionary traders. Everyone is going through the same process of analyzing the previous results, but with backtesting/automation it's a little more cut and dry. It's also not much of a sit back and watch game. People are constantly tweaking strategies. At least for some strategies the line between discretionary and automation is starting to blur.

Perhaps the HFT issue is about the size and amounts of shares traded. Since my positions sizes are larger than most retail traders accounts, it does impact me but for smaller accounts perhaps not. I do trading research every year and in 2006 I joined a day-trading methodology group. Through 5 years of watching about 300 named traders, only 1 of them is still trading for sure. He trades more like I do though. We had lunch last week and went through the lists of names. He knew his city and I knew mine. Gone, all gone. Some of the loss sizes of these people are astounding to me for disciplined day trading. (75000 of 100000 trading account for example) Their biggest issue IMO is the belief that any system can work long term in a dynamic chaotic environment.

 

[xelite777]

Here are the backtest results of a simple moving average crossover system. It has been tested with 11 years of historical data on Apple (AAPL).

Amateurs who started trading Apple with this system from 1998 to 2000 would have naively concluded that this is indeed a fantastic system ("hey, see, my system is working, I am making money!"), while in reality it is a losing system, as you can clearly see.

I can show you hundreds of examples like this.

Still thinking that backtesting a system is a waste of time and not necessary?