The point is that it's market neutral. For example, they might have gotten the same 12% even if the S&P returned -12%. Also, these guys are pure academics. They viewd the stocks as time series and that's it. I think you can make more than 12% if you include other things like EPS or something and actually see every trade you're making and look at every stock you're buying/selling. They just made a program that took a ton of data and analyzed it. I also think they assumed comissions to be $15/trade which is more than double what Scottrade charges.
Spread (pair) trading question
- Thread starter IIAce
- Start date
The point is that it's market neutral. For example, they might have gotten the same 12% even if the S&P returned -12%. Also, these guys are pure academics. They viewd the stocks as time series and that's it. I think you can make more than 12% if you include other things like EPS or something and actually see every trade you're making and look at every stock you're buying/selling. They just made a program that took a ton of data and analyzed it. I also think they assumed comissions to be $15/trade which is more than double what Scottrade charges.
The whole problem is that correlation is defined by the instantaneous returns. What time scale? That is up to you. Cointegration is a long term measure, a great example of highly correlated but non-cointegrated pair is SPY and QQQQ
Quote from RedManPlus:
From a Google search it seems to me...
That corelation and cointegration are 2 similar ways of establishing the same thing for pair trading purposes.
Give me an example of a pair that might have 0.900 corelation ** for a rational reason **...
And not be "cointegrated".
Or give me a specific example where calculating "cointegration" is superior to corelation.
And while building computer systems that track and rank 400 stocks in real-time is complex...
The underlying or core idea for profitable hedging strategies is usually quite simple.
So when I see people talk of tick-by-tick "data mining"...
I just roll my eyes.
rm+
They make a great daytrading pair... THEY ALL DO... COINTEGRATION is a relative measure. Personally, I prefer the ETFs due to their tick valuation (ie. a tick is a penny) whereas things are a bit different for their futures since the tick values are different. HOWEVER, appropriate ratios usually suffice...
Regards,
MAK
Yeah same thing, great pair but not cointegrated... e.g., it is highly correlated but the amount of correlation decays as the timescale increases.
The QQQQ can be viewed as a 30% subset of SPY...
And has recently had a high 90s corelation...
But I would not view this as a tradeable pair...
Because of the kind of occasional, but regular and dramatic, divergence we saw a few years ago.
There is absolutely no problem with using corelation properly...
Using whatever the term one decides is appropriate from experience.
You've been trading for 6 months...
And are all tied up in semantics.
I'm trying to say this nicely...
But geniuses quickly discover shortcuts that slash through the chaos...
While grinders waste their time on semantics or spend years flogging dead end strategies.
The pro/wannabe split here at ET is about 5/95...
And it's quite easy to spot both the pros and the wannabees.
rm+
That's an extremely oversimplified description of pairs trading.
But I would add some basic preconditions:
(1) The 0.9 pair corelation has to be rational, explainable. The securities must be similar... not just corelated.
(2) No one waits until "the ratio returns back" (reverts to historical mean)... you trade in for hours or days... and you trade out. Endless loop.
(3) Baskets work much better than pairs.
(4) You have to be good enough to take money away from ** professional traders **. For pairs trading purposes... it's a zero sum game less all transaction costs including the value of your time. Money is not magically created... you take it away from a pro... because that's who's in there at the bid/ask spread.
rm+
I'm not tied up in anything. And QQQQ/SPY is a very tradeable pair, even with the divergence, maybe you just aren't smart enough to see the chaos.
Quote from RedManPlus:
The QQQQ can be viewed as a 30% subset of SPY...
And has recently had a high 90s corelation...
But I would not view this as a tradeable pair...
Because of the kind of occasional, but regular and dramatic, divergence we saw a few years ago.
There is absolutely no problem with using corelation properly...
Using whatever the term one decides is appropriate from experience.
You've been trading for 6 months...
And are all tied up in semantics.
I'm trying to say this nicely...
But geniuses quickly discover shortcuts that slash through the chaos...
While grinders waste their time on semantics or spend years flogging dead end strategies.
The pro/wannabe split here at ET is about 5/95...
And it's quite easy to spot both the pros and the wannabees.
rm+
Yes I agree that the pair sp500 vs R2K seems to work for daytrading (or intraday changes in spread)...would you say that the pair will work for multi-day/multi-week timeframes? I find it that ER2 (the Russell electronic future) doesn't have much depth (size) to trade intraday to scale up in size. I haven't looked into the ETF for R2K...does it have decent volume?