Great Quantitative Traders with 'Buffett' like returns

[CanOz]

As a continuation to a lively discussion with my good mate Rainman, i thought i would start a thread instead of clogging up the other one that was being dragged away off topic...

If a mod would be so kind as to move some of the discussion it would be greatly appreciated.

So besides some of the Tutlle Traders, i thought i would present a 'Hedge Fund Market Wizard' from Jack's Schwagger's book of the same name.

Edward Thorp is a quantitative trader that used statistical arbitrage as well as trend following strategies throughout his long career...

To Quote the Book

"Thorp's original fund, Princeton Newport Partners, ran for 19 years and had an annualized compound gross return of 19.1 percent. It is not the return, that sets Thorp apart. PNP compiled a track record of 227 winning months and only 3 losing months (all under 1%) - an extraordinary 98.7 % winning percentage."

Hows that for a minimal drawdown. It was not all trend following though.

Bill Dunn also compiled a great track record as a trend following trader.

Mulvaney Capital as well, but none could compete with Thorp's draw-downs....

More From Michael Covel's site:

Ed Seykota: Seykota is a trend following trader featured in Trend Following. He turned $5,000 into $15 million over 12 years in his model account (an actual client account).

Michael Marcus: Marcus is a trend following trader featured in Jack Schwager’s Market Wizards. He turned an initial $30,000 into $80 million

Kenneth Tropin: Trend following trader Tropin made $120 million in 2008 as buy and hold collapsed. Earlier in his career he led John W. Henry’s firm.

So as you can see, there are other ways to invest other than value investing. I totally agree that Buffets return are admirable, but not many can hold through the pain of the drawdowns, as illustrated in the Berkshire chart.

 

[craft]

Local boys
David Walsh & Zeljko Ranogajec

http://www.afr.com/opinion/columns/david-walshs-wisdom-beats-the-odds-20131213-ij8gn

 

[Ves]

Local boys
David Walsh & Zeljko Ranogajec

http://www.afr.com/opinion/columns/david-walshs-wisdom-beats-the-odds-20131213-ij8gn

I love reading interviews by some of the more reclusive hedge fund managers and professional gamblers. They always seem to be able to put a different perspective on things.

 

[CanOz]

I love reading interviews by some of the more reclusive hedge fund managers and professional gamblers. They always seem to be able to put a different perspective on things.

Agree! I came across that Tassie dude a while ago, very interesting.

Thanks for the links Craft!

 

[ThingyMajiggy]

So hard to find out details about this stuff though, like what actual platforms or languages they are written in, if it's a total custom written system, or if it's something else based on a platform that anyone can use.

Like I seriously doubt any top quant fund is using Amibroker, NT or Multicharts to create systems with?

 

[get better]

I listen to a podcast by Andrew Swanscott (http://bettersystemtrader.com/). He interviews system traders every month - very informative and great learning tool if you're interested in mechanical trading systems.

 

[Modest]

I listen to a podcast by Andrew Swanscott (http://bettersystemtrader.com/). He interviews system traders every month - very informative and great learning tool if you're interested in mechanical trading systems.

Thanks these are great!

 

[CanOz]

I listen to a podcast by Andrew Swanscott (http://bettersystemtrader.com/). He interviews system traders every month - very informative and great learning tool if you're interested in mechanical trading systems.

Agree, didn't know this existed, added to my trading podcast collection. Really enjoyed the one with Dr.Brett, came away with several ideas as well...well worth the listen!

 

[systematic]

Agree, didn't know this existed, added to my trading podcast collection. Really enjoyed the one with Dr.Brett, came away with several ideas as well...well worth the listen!

Agreed! Even though most (all?) those interviewed seem to run short(er) time frame systems and heavily use price data...I have got something valuable out of virtually every episode. I encourage any systematic trader (regardless of time frame or style) to take a listen.

 

[sinner]

So as you can see, there are other ways to invest other than value investing. I totally agree that Buffets return are admirable, but not many can hold through the pain of the drawdowns, as illustrated in the Berkshire chart.

AFAIK though, nobody has approached those returns over the sort of duration Buffet has.

Out of a large enough sample size there will always be someone with fabulous returns so one has to weight the returns against their consistency to weed out statistical bias and errors.

Not saying the return is poor or disputing its possible to maintain consistently for another 20Y, but we won't know until they get there.

Over long time periods, for those pursing "strategies" which aren't actually investing (I wouldn't really call Buffets strategy value) in businesses, the risk of ruin approaches 100%.

So hard to find out details about this stuff though, like what actual platforms or languages they are written in, if it's a total custom written system, or if it's something else based on a platform that anyone can use.

Like I seriously doubt any top quant fund is using Amibroker, NT or Multicharts to create systems with?

Who cares as long as the functional outcome is the same? Professional quants mostly test/develop/maintain systems and algorithms in mathematical languages, MATLAB is very prominent and R (what I use mostly these days) is an open source alternative gaining some strong following in the quant blogosphere. A lot of it is in Python.

But to reiterate, it absolutely doesn't matter which platform/language/system you implement a functionally identical system on.

 

[ThingyMajiggy]

Who cares as long as the functional outcome is the same? Professional quants mostly test/develop/maintain systems and algorithms in mathematical languages, MATLAB is very prominent and R (what I use mostly these days) is an open source alternative gaining some strong following in the quant blogosphere. A lot of it is in Python.

But to reiterate, it absolutely doesn't matter which platform/language/system you implement a functionally identical system on.

I know it makes no difference, it's just I'm curious as to what methods/languages/systems they actually use to develop and trade their ideas, and it never get's mentioned. I know just looking around on the net there's Matlab, R and Python. I would think the HFT mobs might be more C/C++ for speed though.

 

[sinner]

I know it makes no difference, it's just I'm curious as to what methods/languages/systems they actually use to develop and trade their ideas, and it never get's mentioned. I know just looking around on the net there's Matlab, R and Python. I would think the HFT mobs might be more C/C++ for speed though.

AFAIK it doesn't really work like you seem to think at all.

Outside of HFT, nobody develops, tests and maintains a system plugged in directly to the market, especially if the turnover is low.

You will use the language to develop, test and maintain the system which generate signals from data, the signals are passed onto the trading desk for optimal execution by algorithm or human. The algorithms can be thought of as simple intraday execution systems, if you know you're going to be buying all day they might be doing something simple like waiting for a 1-min bar to close down then place a limit order, or buy below VWAP, or whatever.

Risk management kicks in once the execution is completed and is a different system again that will feed signals back to trading desk to increase or reduce risk.

It's rarely all connected together in one piece of code, because the signal is not merely "LIMIT BUY 100 LOTS @ $100", when you are trading in size it is best to pass off that sort of signal to a trader or algorithm that can get you optimal execution, which might mean executing 100 seperate buy orders for 1 share over the course of a day or week.

Even as in HFT, you don't bootstrap your system in C or C++ or whatever. It's all done in MATLAB or similar, then you build a very fast representation of the algorithm, optimise it even more, and feed it into a trading engine which is actually the thing sending orders to the exchange.

 

[ThingyMajiggy]

AFAIK it doesn't really work like you seem to think at all.

Outside of HFT, nobody develops, tests and maintains a system plugged in directly to the market, especially if the turnover is low.

You will use the language to develop, test and maintain the system which generate signals from data, the signals are passed onto the trading desk for optimal execution by algorithm or human. The algorithms can be thought of as simple intraday execution systems, if you know you're going to be buying all day they might be doing something simple like waiting for a 1-min bar to close down then place a limit order, or buy below VWAP, or whatever.

Risk management kicks in once the execution is completed and is a different system again that will feed signals back to trading desk to increase or reduce risk.

It's rarely all connected together in one piece of code, because the signal is not merely "LIMIT BUY 100 LOTS @ $100", when you are trading in size it is best to pass off that sort of signal to a trader or algorithm that can get you optimal execution, which might mean executing 100 seperate buy orders for 1 share over the course of a day or week.

Even as in HFT, you don't bootstrap your system in C or C++ or whatever. It's all done in MATLAB or similar, then you build a very fast representation of the algorithm, optimise it even more, and feed it into a trading engine which is actually the thing sending orders to the exchange.

Yeah that's the stuff, would love to have a run down of how it all works and is linked together by someone on the inside, not like that would ever happen though, they seem fairly secretive.

 

[Trendnomics]

Yeah that's the stuff, would love to have a run down of how it all works and is linked together by someone on the inside, not like that would ever happen though, they seem fairly secretive.

Would be interesting, but shouldn't really concern a retail trader - plenty of actual achievable edges out there for retail traders.

 

[ThingyMajiggy]

Would be interesting, but shouldn't really concern a retail trader - plenty of actual achievable edges out there for retail traders.

Haha can't a guy just be interested in a technical area these days? Both assuming I'm talking about strategies and bottom line, I'm just curious about the actual setup(hardware, software, tech, systems used, not trading strategy).

As if they would share their actual trading edge! Sole retail guys don't even want to do that, let alone HFT firms. Usually their "edge" is simply being the first on the buzzer anyway isn't it? Not an actual edge like a retail trader would look for(market patterns etc.)

 

[Wysiwyg]

Would be interesting, but shouldn't really concern a retail trader - plenty of actual achievable edges out there for retail traders.

They are and evidence of them please?

 

[Trendnomics]

They are and evidence of them please?

• Trend following
• Mean Reversion
• Pairs Trading
• Dividend Momentum
• Portfolio Re-balancing

Plenty of evidence out there (google) validating the edges of the above mentioned - also see my trend following diary on here with actual trading history.

 

[Wysiwyg]

• Trend following
• Mean Reversion
• Pairs Trading
• Dividend Momentum
• Portfolio Re-balancing

Plenty of evidence out there (google) validating the edges of the above mentioned - also see my trend following diary on here with actual trading history.

Trading methods.
There are countless "results" published but what are the edges in these trading methods. Obviously "everyone" would be trading an "edge" if there was one. If we drill down, the edge is in position sizing, risk amount and ultimately the position closure reasoning. Oh and throw in some luck that you hit a sensational run.

Also, let's not discount the "edge" of a bull market.

 

[Trendnomics]

Trading methods.
There are countless "results" published but what are the edges in these trading methods. Obviously "everyone" would be trading an "edge" if there was one. If we drill down, the edge is in position sizing, risk amount and ultimately the position closure reasoning. Oh and throw in some luck that you hit a sensational run.

+ Strong emotional intelligence and trading psychology to be able to consistently execute the required trades (this is where most people fall flat).

 

[Wysiwyg]

+ Strong emotional intelligence and trading psychology to be able to consistently execute the required trades (this is where most people fall flat).

Yes there is more to it than meets a Google search.

• Trend following --- trends end at any time (possibly the best method based on trade closure)
• Mean Reversion --- price reverts beyond mean (indefinitely?)
• Pairs Trading --- don't hear much about
• Dividend Momentum --- does happen
• Portfolio Re-balancing --- discretionary risk adjustment