Mean Reversion Strategy

[svobifx 10]

I would like to know if there is a good trading system based on mean reversion? Can I find an expert advisor for it and something I can trade on short and long terms. I really like the idea behind mean reversion trading but I can’t find good material about it.

[MightyMouse 12]

There was one used a while back by LTCM. It worked for a while, but they sort of ran out of capital. Thank goodness for Greenspan.

[SIUYA 11]

There was one used a while back by LTCM. It worked for a while, but they sort of ran out of capital. Thank goodness for Greenspan.

 

When a UBS banker friend of mine told me the same thing - "the only thing wrong with LTCM was that they ran out of money - if they could have kept going they would have been ok" - he probably ended up being head trader somewhere with that great insight.......

 

The problem with most mean reverting strategies is precisely that - not saying that there is not value in them - insurance companies do it - but the problem is you might run out of money.

 

Svobfix -

No I dont know of any mean reverting systems that work - i am sure they exist - but like everything many work until they stop. Plus just asking such a question is unlikely to ellicit many positive responses except those trying to sell you something.....

 

why do you like the idea behind mean reversion?

[TheNegotiator 13]

Depends which product you trade, what the product is currently doing and over which timeframe you trade it. Just like any other strategy.

[BlueHorseshoe 10]

I would like to know if there is a good trading system based on mean reversion? Can I find an expert advisor for it and something I can trade on short and long terms. I really like the idea behind mean reversion trading but I can’t find good material about it.

 

Forget about "expert advisors". If it were that simple, we'd all just go out and buy an expert advisor and we'd all be rich . . .

 

To trade successfully you will need to find a way to understand the general behaviour of the market you want to trade, and then find a specific way to exploit that behaviour which you can apply with consistency.

 

The very first question you need to be asking is "does the market I want to trade display mean-reverting behaviour?" This needs to be the case in a very general way (the more general the better) before any specific mean-reversion strategy will be viable.

 

Another thing that may help to focus your research is to understand that the terms are essentially used in two different (though closely related) ways:

 

1) The way that I think SIUYA is using the term applies to the quantitative statistical arbitrage strategies which involve trading a cointegrated pair long and short on the assumption that the difference will decay over time. This was most commonly practiced in stocks - a great book to read would be Ernie P Chan's - you can find his very informative blog here: Quantitative Trading

 

2) The other way in which the term is used is to refer to a single instrument reverting to its own mean. Pretty much all price movement is the reversion of an instrument towards some mean - the problem is always knowing to which mean it is reverting at any given time. By definition, large price moves involve reversion to longer term means, and small price moves to shorter term means. Strategies to exploit this often involve identifying a 'target' mean and then fading price movement away from it at 'extreme' levels. I think that the best insight that I can offer here is that reversion to a short term mean is most predictable following reversion to a longer term mean.

 

Hope that's all of some help to you!

 

BlueHorseshoe

[zdo 20]

Just some thoughts…

 

The “mean” is not really a very good ‘central tendency’ to look for reversions to.

 

The most safe mean reversion systems go very light load on normal excursions. They only load up size on exhaustions of extreme outliers…

 

Instead of saying they “ran out of money”, I think it’s more accurate and instructive to say that LTCM

1) refused to stop out and

2) decided to ‘double down’

… ultimately the description “ran out of money” has same result, but this way points more to the avoidable ‘mistakes’ involved

[equtrader 10]

I would like to know if there is a good trading system based on mean reversion? Can I find an expert advisor for it and something I can trade on short and long terms. I really like the idea behind mean reversion trading but I can’t find good material about it.

 

I know of plenty. Any system you design with a high win rate that is curve-fitted or a statistical fluke may serve as a good candidate for a mean reversion strategy. Take any tool that designs trading systems, for example Price Action Lab. Set parameters so that the systems it finds have high win rate but low statistical significance. The win rate of such systems will have to drop towards 50%. Take the opposite side. Risky but magnificent.