mahkoh
1,065 posts
msg #112254
- Ignore mahkoh |
3/15/2013 4:23:35 AM
PTF also uses alphabetic order. This does address one of the issues I found with pair trading. If FLR:HES is outside 2 SD it does not necessarily mean that HES:FLR is as well.
|
pirate67
99 posts
msg #112258
- Ignore pirate67 |
3/15/2013 10:54:41 AM
OK, but how are you determining which stock to go long and which to short? The ratio is 2 SD's from the mean, but that is not saying that one of the pair is necessarily that far from the mean. I must be missing something here!
|
mahkoh
1,065 posts
msg #112260
- Ignore mahkoh |
3/15/2013 2:19:49 PM
You can draw a chart of the ratio and plot an MA and bollinger bands on it. It does not matter where the stocks themselves are in relation to their own Bollinger Bands (although one could take that in consideration), it is about the ratio of the two moving a number of standard deviations from the MA. And then hopefully revert to the mean.
|
Kevin_in_GA
4,599 posts
msg #112261
- Ignore Kevin_in_GA |
3/15/2013 3:05:07 PM
Here is an example of how this works.
http://stockcharts.com/h-sc/ui?s=HES:FLR&p=D&yr=0&mn=1&dy=0&id=p57762860282
The pair is HES and FLR (HES:FLR) over the last month. What is plotted is the ratio of HES/FLR, which has a set of Bollinger Bands applied to it (20 days, 2 sd in blue and 20 days,1 sd in red). Note that on March 1 the ratio closed ABOVE the upper BB(20,2). When this occurs you would expect the ratio to revert back to the mean so you would take a short position in HES and a long position in FLR. It doesn't matter where each individual stock is relative to the same Bollinger Bands, all that matters is the RATIO.
Both positions are entered at the open on 4 March. Over the next two days the ratio closed back below the BB(20,1). One then sold the FLR position and covered the HES position at the open on 7 March.
Was this profitable? Here are the prices for those stocks:
Short HES at the open on 4 March at 69.11 and cover at the open on 7 March at 69.88 (a loss of 1.1%)
Long FLR at the open on 4 March at 60.17 and sell at the open on 7 March at 63.41 (a gain of 5.38%)
The pair makes money even though one trade is unprofitable. You are still up overall on the pair trade, which is the goal. If the ratio drops below the lower BB(20,2), you go long HES and short FLR.
For this pair you can see examples of this type of trade on 2 Jan, 7 Jan and 25 Jan of this year, all of which are profitable.
|
leonsf
18 posts
msg #112268
- Ignore leonsf |
3/15/2013 8:56:25 PM
Hi Kevin,
Just to confirm that I got this right.....
So when the ratio closed above SD (standard deviation) 2, we go short the stock on the left side of the pair and go long the stock on the right side of the pair.
And on the other hand, if the ratio closed below SD 2, we go long the stock on the left side of the pair and go short the stock on the right side of the pair.
Do I get this right?
Again, thanks for sharing this useful info.
~Leon.
|
pirate67
99 posts
msg #112271
- Ignore pirate67 |
3/16/2013 12:15:43 PM
Thanks Kevin, how it works is clear
|
miko
68 posts
msg #112275
- Ignore miko
modified |
3/16/2013 2:10:13 PM
Thx, Kevin, love this stuff :)
In your HES:FLR example, there was follow through the following opening (esp for the entry), indeed trading the close looks to be a losing trade; is the follow through common? To ask it another way, does trading the following opening outperform trading at the close (or is trading the close too impractical, so not even tested/considered)?
One reason I ask this is that I have seen strategies that outperform open-open vs close-close enough to suggest that close-open may be a trade in itself (eg in your example, at the close of Mar 1, going long HES/short FLR and closing on the open Mar 4).
|
mahkoh
1,065 posts
msg #112276
- Ignore mahkoh |
3/16/2013 4:21:19 PM
I would say that this is quite random. HES gapped up on March 4 after announcing to divest their retail business and a stock buyback. If FLR had made a similar announcement instead of HES then the previous close would have been a better entry.
The reason for entries at the close is because it was backtested that way, PTF's backtest feature uses Yahoo's end of day data.
If you have PTF running it will show pairs in your watchlist that trigger intraday. It also has a feature to send you emails although I never managed to get that to work.
|
miko
68 posts
msg #112277
- Ignore miko |
3/16/2013 5:30:05 PM
mahkoh,
While I appreciate your response, I was hoping for something quantified, not a refutation based on a single trade that was given as an example.
My understanding is that the trades were done on the open, not the close, and backtested that way.
|
mahkoh
1,065 posts
msg #112279
- Ignore mahkoh |
3/16/2013 5:57:24 PM
I am not 100 % sure, but I have always assumed PTF uses the close for entries. If you want to be sure you'd have to check with the person that created the software. Unfortunately however, support for pairtradefinder is pretty nonexistent.
|