Let's discuss academic research on mean-reverting trading strategies...

[dtrader98]

People should think of this as more of a probability game. Whether you want to deal with stat arb, mean-rev, or whatever game you play-- it all comes down to a game of expectations; with real systematic trade-offs like slippage, commission bias , and st taxes the more frequent you trade. Unfortunately, if you are on your own, all of the above negative biases work strongly and consistantly against you.

If you have somehow surmised that net expectation is in your favor (taking all of the above into account) and most of your criterion are satisfied, the next question to ask is:

is there any way to minimize the impact of a potential black swan?

hint: 10:1 leverage is likely not the answer.

Throw some black swans in your scenarios and see what happens. If the net + is gone, then you need to rethink your approach, because if you don't: then at that point you are merely playing a game of chicken. A lot of these institutions can afford to play chicken, because uncle sam is on their side; unfortunately, unless your name is goldman sachs, you cannot.

Too many neophytes think offense, when the real key is defense in this game.

 

[propseeker]

Quote from dtrader98:

Too many neophytes think offense, when the real key is defense in this game.

so, so true.

 

[warrenmurdoch]

A good and readable explanation of the differences between correlation and cointegration can be found here http://www.tradingmarkets.com/.site...ntegration-is-not-the-same-as-correlation.cfm

 

[intradaybill]

Quote from warrenmurdoch:

A good and readable explanation of the differences between correlation and cointegration can be found here http://www.tradingmarkets.com/.site...ntegration-is-not-the-same-as-correlation.cfm

Is anyone else having problems accessing the tradingmarkets.com website with DEP enabled in IE?

When I enable DEP, IE won't open the website meaning that it is trying to execute software on my computer, or what?

 

[Ms Varima-Garch]

Quote from dtrader98:



is there any way to minimize the impact of a potential black swan?

hint: 10:1 leverage is likely not the answer.

Throw some black swans in your scenarios and see what happens. If the net + is gone, then you need to rethink your approach, because if you don't: then at that point you are merely playing a game of chicken. A lot of these institutions can afford to play chicken, because uncle sam is on their side; unfortunately, unless your name is goldman sachs, you cannot.

Too many neophytes think offense, when the real key is defense in this game.

if the black swan come it kill all the cointegrations and all price analomies

 

[virtualmoney]

Quote from Ms Varima-Garch:
if the black swan come it kill all the cointegrations and all price analomies

Exactly, the key to successful mean-reverting strategies is to know when to cut loss...any suggestions using some quantitative rules?

Is there like a Co-integration indicator for 2 instruments?

 

[colonelangus]

Quote from virtualmoney:

Exactly, the key to successful mean-reverting strategies is to know when to cut loss...any suggestions using some quantitative rules?

Is there like a Co-integration indicator for 2 instruments?

for a basic idea, look at where the spread has been over the past year with a ratio that puts the pair as close to zero as possible. from that, you can have a better idea as to where your stop on the pair should be. i generally have it just outside the previous high/low of the spread.
for my cointegration indicator, i divide the pair's 15 day ATR by the last 6 month's range of the pair. the higher the number, the more cointegrated the pair is likely to be(>.07) the lower the number, the greater the likelihood for the pair to trend.

any other formulas on this are welcomed with big, greedy hands

 

[Ms Varima-Garch]

think it's also important to do actual backtesting on various pairs do get a feel for max wins/losses, over various time intervals.

i was thinking of getting a free trial of pairtradefinder, because they have the backtest feature (but i couldn't launch it because of some stupid problems with SQL or whatever it's called)

in any event, i think it's good to do actual trade backtests to get a feel for any given pair.

ratios, correlations, cointegrations are probably tools in the toolbox, not any one of them is 100% bullet-proof.

also, there may be fundamental (including, macroeconomic) reasons why a certain relationship is behaving in some "non-normal" way. you can make a bet that the price analomy will vanish, but it may persist.

the market can remain inefficient longer than you can remain solvent

so i DON'T think you can trade on the strength of ratio relationships alone. that said, my advice does not carry much weight because i'm rookie in this.

i've attached an article by Dr. Chan about how one - ONE! - spread killed the Amaranth fund.

you have to control your emotions, commitment commitment commitment. you have to stay focused, like a warrior, you have to unleash your various inner, and just take it from there. discipline is very important. you must always control your emotions and your ego. you really do have to stay centered, as they say, in a sentiment driven market. do some breathing exercises, go for a walk. don't let those winners overpower those losers. it's about domination and control. don't try to dominate the person on the other side of your trade. don't fall into the tempation of being subdued by that person either.

don't try to get personal with the person on the other side of your trade. he or she doesn't know who you are, a total stranger.

your ego must be taken out of the equation completely, it's like zen in the art of archery, you already know, before you shoot, your arrow will hit the target.

none of the above, thankfully, applies to spreads. in spreads, you just have to relax, and let it run, unleash your various inner, and just take it from there, basically .. .

there are tons of free stuff on www.gummy-stuff.org. amazing stuff. this once again confirms FREE is often better. if you take a good look at it, there are excel add-ins for correlations, pairs trading and related stuff. it's probably not immediately tradable, but it's free and that outweighs everything else.

MarketTrack is FREE, finviz is FREE.

There is free stuff out there, you just have to have an open mind, and be ready for the free stuff. It's all about psychology, most of us, because of some psychological trauma, cannot open ourselves up to free stuff. We keep punishing ourselves, only reinforcing our feeling of guilt by using non-free stuff. By holding on to our losers, by not allowing our losers to bascially do battle with our winners. On their own. We really need to be the arbiter in this, we can't take sides just like that.

FREE is often better. I'm shocked by this paradox but it's true.

 

[Cutten]

Quote from dtrader98:

People should think of this as more of a probability game. Whether you want to deal with stat arb, mean-rev, or whatever game you play-- it all comes down to a game of expectations; with real systematic trade-offs like slippage, commission bias , and st taxes the more frequent you trade. Unfortunately, if you are on your own, all of the above negative biases work strongly and consistantly against you.

If you have somehow surmised that net expectation is in your favor (taking all of the above into account) and most of your criterion are satisfied, the next question to ask is:

is there any way to minimize the impact of a potential black swan?

hint: 10:1 leverage is likely not the answer.

Throw some black swans in your scenarios and see what happens. If the net + is gone, then you need to rethink your approach, because if you don't: then at that point you are merely playing a game of chicken. A lot of these institutions can afford to play chicken, because uncle sam is on their side; unfortunately, unless your name is goldman sachs, you cannot.

Too many neophytes think offense, when the real key is defense in this game.

The alternative is to formulate a strategy that predicts Grey Swans (IMO there are no Black Swans in financial markets), and switch from mean reversion to Grey Swan trading when the signals flash.

 

[Cutten]

By the way, why is it called "stat arb" when there is no arb? Ditto for risk arbitrage.