Pair Trading Strategy Journal
- Thread starter jonnysharp
- Start date
Just a quick update from the UK.
I've now been using PTF since April (blimey, doesn't time fly) and things have been going extremely well. I only trade UK stocks (although did have a disasterous foray into a US pair that wiped out 20% of my profits in one go) and as I've filtered them out I now have a 74.7% success rate and a 1.54% ROI which I'm pleased with.
Over the past 20 years or so I've looked at all forms of TA and have never been able to get anything to work over a period of time, possibly because I'm a rubbish trader but I'm finding pairs trading seems to work for me. I did a course with Rob Friesen a while back and things have been looking good since I was introduced to the subject of statistical arbitrage.
Previously I was a very successful sports arbitrageur but now can't get on with bookmakers because they have me labelled as an arber and restrict my bets. My main income now comes from Betfair and I run automated algorithms that take advantage of inefficiencies in some of their markets but I doubt these will last much longer as their markets are becoming more efficient by the day.
I'm working my stock statarbs up to a point where the income they generate will be able to take over from the dwindling Betfair income. At the moment the ROI is fine but I'm just not getting sufficient trades to create an income as yet. Because of my filtering I'm only getting 3 or 4 trades a week, so I need to build up my trade size in order for me to get the returns I need. And I've always found I need to do this slowly otherwise I don't make the best decisions. Something I learned long ago when I was a blackjack player was that your bets (trades) need to be at a level such that a loss has little impact on your future decision making process.
So, the upshot is that things are still going well and I'm slowly building my bank and my confidence.
I've now been using PTF since April (blimey, doesn't time fly) and things have been going extremely well. I only trade UK stocks (although did have a disasterous foray into a US pair that wiped out 20% of my profits in one go) and as I've filtered them out I now have a 74.7% success rate and a 1.54% ROI which I'm pleased with.
Over the past 20 years or so I've looked at all forms of TA and have never been able to get anything to work over a period of time, possibly because I'm a rubbish trader but I'm finding pairs trading seems to work for me. I did a course with Rob Friesen a while back and things have been looking good since I was introduced to the subject of statistical arbitrage.
Previously I was a very successful sports arbitrageur but now can't get on with bookmakers because they have me labelled as an arber and restrict my bets. My main income now comes from Betfair and I run automated algorithms that take advantage of inefficiencies in some of their markets but I doubt these will last much longer as their markets are becoming more efficient by the day.
I'm working my stock statarbs up to a point where the income they generate will be able to take over from the dwindling Betfair income. At the moment the ROI is fine but I'm just not getting sufficient trades to create an income as yet. Because of my filtering I'm only getting 3 or 4 trades a week, so I need to build up my trade size in order for me to get the returns I need. And I've always found I need to do this slowly otherwise I don't make the best decisions. Something I learned long ago when I was a blackjack player was that your bets (trades) need to be at a level such that a loss has little impact on your future decision making process.
So, the upshot is that things are still going well and I'm slowly building my bank and my confidence.
Good to hear Doc!!!
I'm long NMR short CS last week which was an absolute disaster..... My first proper black swan. Still in trade will re-evaluate this week...
Have you got anything to add re: The stop debate?
I'm UK based and, am not surprised by how few trades you are finding. If you want liquidity and pairs than you really have to play the US market. Your foray into the US represents a data point of one and, has no statistical significance whatsoever. Why not try again with the US?
I'l interject and throw a couple of comments forward...
I've found that a stop loss (whether volatility or fixed % based) will dampen down returns, but it shoud also smooth out volatility. I'm coming round to the view that it is necessary, even if the backtesting suggests its not optimal. Why?
1/ Concerns over the 'pre-selective' nature of the universe of stocks backtested.
2/ Avoiding the black swan
3/ I accept that it sharp bull markets, mean reversion strategies are likely to suffer. Therefore this strategy may be directional. The stop loss will likely help limit losses under this scenario. This month is a good example because we saw a strong bull market. Here are my numbers so far for trades opened in Sep (n=32)...
2% stop loss W/L Ratio=.91 Prob=.59 (positive)
3% stop loss W/L Ratio=.59 Prob=.59 (negative)
4% stop loss W/L Ratio=.6 Prob=.63 (slightly positive)
No stop loss W/L Ratio=.52 Prob=.63 (negative)
So we see that, whereas previously the results were linear with the no-stop loss being optimal, this month this relationship breaks down.
It's not enough data to conclude anything meaningful, but I'm beginning to think that a stop loss is necessary to mitigate losses in unhelpful market conditions.
I think some sort of stop loss is completely necessary. By the definition of the strategy we are aiming for a fixed profit and while the winning ration might be 75%, the losers will tend to be slightly bigger or even at best. Since most likely in the long run we will not be capturing positive black swan events and more likely than not try to ride out the negative ones, it's no surprise that this strategy is sometimes compared to picking up nickels in front of a steamroller. At some point you have to cut off a losing trade or you can wipe out a month worth of gains, or worse. Unless you have a good fundamental reason, it does not make sense to fight any trade that was based on statistics and theoretical probability of that trade being successfull when stocks aren't normally distributed. As they say, keep flipping those burgers all day.
Quick follow-up to my post on 9/17. This appears to be a decent area to take gains in PEP/KO and COP/CVX. DE/CAT hasn't done much and should be looking for an exit before they report earnings. WYE/PFE convergence continues and that should narrow further now that we are likely getting close to the deal closing. Same applies to SGP/MRK.
One that appears attractive at current levels is long CVS, short WAG.
Walt,
I have been going over some of your comments, looking for the comment about the % difference between the ratio and the average ratio. I think this is key for me, as the difference between these two represent the maximum profit for the given trade.
Not sure if you have a solution to your comment above, but I have found it quite easy to determine ahead of time what the spread price will be when the average ratio and the ratio will converge. Using that, I just enter a GTC order at that spread price.
Adrian
I just put together a pair system that makes around 18% per year (unleveraged).
How much returns are you guys making in actual trading? And do you use margin/leverage?
Also what is your return/drawdown ratio? That is Average Annual Return divided by maximum expected peak to valley?
I want to see if there's room for improvement in what I've come up with.
How much returns are you guys making in actual trading? And do you use margin/leverage?
Also what is your return/drawdown ratio? That is Average Annual Return divided by maximum expected peak to valley?
I want to see if there's room for improvement in what I've come up with.
Hello Guys, I have been trading pairs for awhile, with a return in over 1.5 year of 54%. I'm puting them at the virtual trading on WEALTHLAB, (PAIRS-ARB)
www.wealth-lab.com/Community/VirtualTrading/Rankings.aspx
I'm using pairtrading finder software but them I use resistence and support for the candidates, This week I have two picks but still undecided wich one to use: TIN:VCP and MR:ACL
Regarding a program that automatic can close a pair when a target have been optained, I believe it can be setup with http://traderxl.com/ withan account at TD-Ameritrade
www.wealth-lab.com/Community/VirtualTrading/Rankings.aspx
I'm using pairtrading finder software but them I use resistence and support for the candidates, This week I have two picks but still undecided wich one to use: TIN:VCP and MR:ACL
Regarding a program that automatic can close a pair when a target have been optained, I believe it can be setup with http://traderxl.com/ withan account at TD-Ameritrade