Pair Trading Strategy Journal

chiefraven · Feb 27, 2009

the PQ ROSE trade...

just looking at an overlay charts of the two stocks.... there's not much deviation to exploit ....... in addition...... these two pairs are about to cross... meaning that the higher one is crossing to become the lower one.... and the higher one is dropping down and becoming a lower one...... and when this happens..... when you try to pairtrade at this area... most of the time it will fail.... just a heads up

mlsignups · Feb 27, 2009

Johny:

I'm trying to replicate your correlation chart in my pairtradefinder for this and i can't get the correlation to match. Can you share again your lookback period? At one point in the journal you indicated I think it was 100 days. Is that what you still are using?

Also, i don't know if the program is supposed to do this but there are two lines of data in your chart for 2/23... i don't have that in mine...i don't know if that could be why mine is different than yours.

Mike

tekka · Feb 27, 2009

I was wondering if using IQ feed is helping traders make better profits. Seems to me that often trade execution is crucial to the profits. Comments from IQ users really appreciated.

Another question is on the volatility graph. Is anybody actually paying attention to that one? Is showing the volatility of the actual ratio (similar to RSI), or something else.

And finally, does anybody have and idea when the next version of PTF is coming out?

jonnysharp · Feb 27, 2009

Quote from saico:

Hi Johnny,

for me I decided to pass signals on penny stocks for several reasons. First of all, those stocks are a subject for many gamblers in market. They can be moved very easyly because of their low price. Even more when that stock is a low volume stock. Just see the examble with the pair you just posted. If you would have taken the signal on feb. 19, your position would have seen a massive drawdown. Even its a dollar neutral trade. Those stocks are that low for a reason and the selling pressure is usually higher than in higher priced stocks. I think there are a lot of other signals out there on pairs that include solid stocks. The risk is not worth it to take these ones imo. Just my 2 cents, Johnny.

Saico.

Valid reason and I agree, however Im no stranger to risk and a bit of gunslinger so I like these trades, I only place a small % of my a/c in each trade so in the event of a black swan I won't be wiped out.

jonnysharp · Feb 27, 2009

Quote from mlsignups:

Johny:

I'm trying to replicate your correlation chart in my pairtradefinder for this and i can't get the correlation to match. Can you share again your lookback period? At one point in the journal you indicated I think it was 100 days. Is that what you still are using?

Also, i don't know if the program is supposed to do this but there are two lines of data in your chart for 2/23... i don't have that in mine...i don't know if that could be why mine is different than yours.

Mike

Yeah I use 100 day lookback, the 2 lines could be because there is 2 layer signals in one day and thus its shown as 2 lines.

jonnysharp · Feb 27, 2009

Quote from tekka:

I was wondering if using IQ feed is helping traders make better profits. Seems to me that often trade execution is crucial to the profits. Comments from IQ users really appreciated.

Another question is on the volatility graph. Is anybody actually paying attention to that one? Is showing the volatility of the actual ratio (similar to RSI), or something else.

And finally, does anybody have and idea when the next version of PTF is coming out?

yeah I use IQfeed helps a bit I think although its not that different to when I was using the free data feed as Im looking at real time quotes in my trading platform anyway. Yeah I use the volatility chart, you can size according to volatility, I like to see volatility making new highs aswell, as pair trading is similar to selling premium it helps to be on the right side of volatility. I think PTF v3 is due out later this year according to the guys there.

jones247 · Feb 28, 2009

When pair trading two stocks (preferably within the same sector), there's a MAJOR PROBLEM with the mean-reversion strategy. The assumption is that the two stocks will converge after diverging to a certain point. Oftentimes, layering-in is used to improve the performance of the mean-reversion plan (i.e. "it did not revert on the 1st entry, let's try to enter again - hoping that the reversion will occur after the next std deviation).

This is the crux of the problem: If fundamentals dictate that you long the strong stock and short the weaker stock, then one would think that the stock is more likely to continue to diverge rather than converge. To bet on a mean-reversion, one would typically need to short the strong stock and long the weak stock. In doing so, it appears that you're betting against market fundamentals.

In a nutshell, my theory is that it may be better to trade with the trend of the divergence than to try and anticipate a reversal (mean reversion) of the two stocks. In other words, once a pair start to diverge (especially if supported by fundamentals such as earnings or news reports), then enter the spread with the plan of the divergence continuing.

Perhaps mean-reversion would be better suited with indicies, such as the spy vs. dia...

your thoughts... Do you see a problem with my logic???

Thanks,

Walt

saico · Feb 28, 2009

Quote from jones247:

Do you see a problem with my logic???

Not a problem at all, Walt. But the fact is simply that mean reversion worked for the last 40 years. Thats the difference to many other directional strategies.

saico

tatankas · Feb 28, 2009

Quote from jones247:

your thoughts... Do you see a problem with my logic???

Thanks,

Walt [/B]

Walt, the strategy described on this journal, is for a short time-frame. Holding period is only 7-15 days.

I think fundamentals are much more important when your holding period is from 1 to several months.

total_keops · Feb 28, 2009

Quote from jones247:

When pair trading two stocks (preferably within the same sector), there's a MAJOR PROBLEM...
your thoughts... Do you see a problem with my logic???

With mean reversion as done here you are trying to catch a little fast irrationality. If you read the posts carefully you will see that trades does not occur when there is a trend but rather when at the boundaries of a trading range.
If you read the book Statistical Arbitrage, you will find that this strategy is not as profitale as it used to be (Merill started doing this in the 80 with a black box and other catched the idea). But still, there seems to be some opportunities for profit when done properly. (I dont trade them yet but am working on a scanner).