Intraday Pair Trade

Quote from Rehoboth:

You can definitely make money intraday trading pairs. A few things to consider:

1. Dollar Neutral is not the best ratio for pairs trading
2. Wait for the move to start back in rather than catching a falling knife
3. Focus on sectors that have the vol right now.


Long term pairs is a lot more tricky. Earnings, buybacks, secondaries, dividends, and other corporate actions all come into play. But those 3 things above still matter.
More...

Really appreciate your advice.

For 1. I guess I can try position sizing according to ATR of each stock
For 2. This is actually I do right now, for example if the bollinger band is (200, 2), i first wait for the ratios to go above/below 2.2 standard deviations (give it 10% margin), then I confirm they broke the boundary, and will take trade once it falls back within 2 standard deviation. This has to be done with a prop firm account coz I am unable to take too much unrealized loss intraday.
For 3. I will definitely look into that.

Two more questions
a) do you think plotting ratio works instead of plotting spread, on an intraday basis? By ratio I mean (stock A / stock B) and by spread I mean (stock A - stock B)
b) do you think 1-5 min bars are appropriate?

Thank you for your help.
 
Quote from ar1zona:

Sorry I forgot to mention this I have prop firm account so there's access to real time tick data. And although I did backtesting with 1-5 min bars (from Qlink) the execution is intended to be reading tick data real time from Sterling T&S.

The main reason I used 1- and 5- min bars is because tick data have too many noise. And its impossible to have ticks for both stocks at each specific second. Using bar closing price at least there's a consistency in data. I actually do find 5 min worked a bit better than 1 min, and heard people talking about even 1-5 min bars have too much noise and its better to use 10-20 mins if you have to trade intraday. Do you think its worth to test with a smaller timeframe (subminute) and what kind data should I get for that? Thank you.
More...

PM me a pair you're watching. I'll record a day's worth of TAQ data for you w/ congruent time stamps.
 
Quote from ar1zona:


a) do you think plotting ratio works instead of plotting spread, on an intraday basis? By ratio I mean (stock A / stock B) and by spread I mean (stock A - stock B)
More...

I think it varies depending on the group. When trading across currencies, the pairs I construct require construction on a currency adjusted spread basis (A-B). Same w/ out the currency adjustment, ie: spy vs /es.

Ratios should be used for gold/silver, gold/plat, crude/ng, etc.
 
Quote from ar1zona:

Really appreciate your advice.

For 1. I guess I can try position sizing according to ATR of each stock
For 2. This is actually I do right now, for example if the bollinger band is (200, 2), i first wait for the ratios to go above/below 2.2 standard deviations (give it 10% margin), then I confirm they broke the boundary, and will take trade once it falls back within 2 standard deviation. This has to be done with a prop firm account coz I am unable to take too much unrealized loss intraday.
For 3. I will definitely look into that.

Two more questions
a) do you think plotting ratio works instead of plotting spread, on an intraday basis? By ratio I mean (stock A / stock B) and by spread I mean (stock A - stock B)
b) do you think 1-5 min bars are appropriate?

Thank you for your help.
More...

a. The chart / calc I use is always (Stock A * Ratio) - Stock B Where the ratio is what you found to be appropriate for the day.

b. the cleanest shortest time frame you can get is always the best for this. For me it varies stock to stock b/c of the volume and prints of the thinner ones.


Just using a BBands isnt going to work either. A stock in the news is always going to generate trigger. Look for a non news pair, in a vol sector, that has unusual movement for the pair where both stocks are the reason the pair is out of whack. B/C if its only 1 stock causing the pair skew then you might as well just do it against the etf of the index.
 
Quote from Rehoboth:

a. The chart / calc I use is always (Stock A * Ratio) - Stock B Where the ratio is what you found to be appropriate for the day.

b. the cleanest shortest time frame you can get is always the best for this. For me it varies stock to stock b/c of the volume and prints of the thinner ones.


Just using a BBands isnt going to work either. A stock in the news is always going to generate trigger. Look for a non news pair, in a vol sector, that has unusual movement for the pair where both stocks are the reason the pair is out of whack. B/C if its only 1 stock causing the pair skew then you might as well just do it against the etf of the index.
More...

Could you expand a lil bit on that ratio please? You are the first one I heard calculating them this way. What exactly is the ratio used for? bring the spread to a certain range so it fits your chart or? And how do you go about determining the ratio, is it a different value everyday?

I totally agree on the news pair and some of the biggest losers in the backtest were the ones where one stock just keeps running away from the other. I can read the news in the morning and see if there's any significant stuff going on but its quite difficult to incorporate this into the backtesting.
 
Quote from Rehoboth:

a. The chart / calc I use is always (Stock A * Ratio) - Stock B Where the ratio is what you found to be appropriate for the day.

b. the cleanest shortest time frame you can get is always the best for this. For me it varies stock to stock b/c of the volume and prints of the thinner ones.


Just using a BBands isnt going to work either. A stock in the news is always going to generate trigger. Look for a non news pair, in a vol sector, that has unusual movement for the pair where both stocks are the reason the pair is out of whack. B/C if its only 1 stock causing the pair skew then you might as well just do it against the etf of the index.
More...

This is to reply your PM, from my understanding you prefer to check this thread over PM so I'm posting the questions here.

It is very encouraging to know you and others are profitable with intraday pairs. At least this gives intraday-pairs-wanna-be people like me a boost in confidence. Now its about digging hard enough to get it to work...

My ideal system was to find a small basket of elite pairs, like 10-20 of them, who are consistently profitable over 6-12 months with a certain parameter (i.e. 100, 2), and on average generate 0.5-1.0 trade per day. The advantage is its easier to check news/anomalies on a limited number of stocks in the morning.

You suggested to apply more filters to increase win rate and improve quality of each executions. Do you mean to come up with, for example, 50-100 pairs but each of them only have a very small chance to be triggered on any given day? I can see why having a more strict criteria helps and as long as you have enough pairs in your basket everyday on average you still have a couple of trades. But what kind of timeframe do you use to backtest this? I mean if I backtest with 6 month, and got 10 trades with a 90% win rate and profit factor of > 1.0, it does look very nice but how do you avoid the risk of data mining?

One of the reasons I wanted to come up with an elite small basket of pairs is to have enough historical trades to confirm the pattern found is not a coincidence. I am not in any ways questioning your method here in fact I am working on improving my sheet to be more efficient so I can test a wider universe of pairs as you said. I just want to know how you go about overcoming these concerns. Again, really really appreciate your help.
 
Quote from ar1zona:

Could you expand a lil bit on that ratio please? You are the first one I heard calculating them this way. What exactly is the ratio used for? bring the spread to a certain range so it fits your chart or? And how do you go about determining the ratio, is it a different value everyday?

I totally agree on the news pair and some of the biggest losers in the backtest were the ones where one stock just keeps running away from the other. I can read the news in the morning and see if there's any significant stuff going on but its quite difficult to incorporate this into the backtesting.
More...

The ratio is used to remove the market movement. So if the market has run 2% that does not affect the spread. Otherwise all you are doing is trading the market. Everyone I know has a different method for coming up with this ratio.
 
Quote from Rehoboth:

The ratio is used to remove the market movement. So if the market has run 2% that does not affect the spread. Otherwise all you are doing is trading the market. Everyone I know has a different method for coming up with this ratio.
More...

Yes I have heard quite a few ways to calc this ratio. This is probably the biggest difference between everyone's models. The rest of the story (take trades when the ratios are out of whack) are pretty similar for all pair traders of all timeframes.

Have never thought about adjusting for market movements but it does make sense. I will definitely modify the sheet to try out different ways to calculate the ratios and get back to this post if found anything. Thanks again for the guidance.

p.s. by market gone up 2% that's based on yesterday's close right?
 
Quote from ar1zona:

Yp.s. by market gone up 2% that's based on yesterday's close right?
More...

That is entirely up to you. I use open to close. pre-close to open. open to high etc...
 
You must log in or register to reply here.
Back
Top