Unloaded the whole portfolio this morning. I was down at an uncomfortable level and I had the opportunity to get some at a profit and some at a loss so I covered for a loss on the whole protfolio. Sold ELD.TO 10.63$ Bought AU 37.82$ -5.7% Sold FNFG 11.01$ Bought FMER 17.86$ Worst one, a sector play that turned not to be a good idea. -11.86% Bought EWW 27.93$ Sold EWG 15.39$ +1.56% Sold MT 19.12$ Bought TS 19.77$ +6.11% Sold ITW 28.69$ Bought SWK 29.60$ +0.09% All % after fees. They are still considerale for the size I play but I dont care, I learn.
Nothing really big. I should be more torough on the analysis. Take losses when I realise I'm wrong. I try to improve my patience in trading so this effort sometimes comes in conflict with what I should really do. I just get my feet wet so it's cool with me. There will be better trades.
Hi Keops, Do you think that you would have been profitable overall if you traded for the divergence, instead of the convergence (mean reversion)? In other words, buy the strong stock & sell the weak stock after they seem to begin the pattern of divergence from each other. thanks, Walt
Yes, They all went against me at some point but on some I would have had to take some profits quickly because some reverted as expected. I'm only one trade down, no big deal (3 loss, 2 wins, 1 flat). I dont think it's a smart idea to trade the divergence with a convergence set-up. But I think divergence can be profitable and I should implement it. I just dont have the time to take care of it right now. It would require me to write some new code for that. As I pointed out earlier on I think it's best to have 50% convergence, 50% divergence. There is also some intraday that can be done both styles but I dont get into that now.
keops, I remember you saying you don't use pairtrade finder and you programmed your own screener, may I ask how you are determining your entries and exits?
Hi Jonny, Many thanks for this thread, it's very inspirational. A couple of questions, if you don't mind: 1. Can you disclose which approach are you using, and which you would have chosen were you to start anew? Pros and cons? 2. Is the Pair Trade Finder software restricted to run on one computer only? I.e., can I use it on my desktop and laptop (not simultaneously)? 3. From your experience, what is a comfortable number of pairs to hold at any given time to be sufficiently diversified without overcomplicating the portfolio maintenance? What's the average number of pairs you hold? Many thanks!
I wrote my own scanner in C. I update the data with Yahoo (Yahoo and Google historical quote downloader and an Excel sheet for the intraday update). I run the program once at about 15:15 and scan manually looking at the charts in Matlab. Scanner entries criteria: Stocks above 5$, average daily volume over 500k, market cap above 500M. NYSE, AMEX, NASDAQ, Toronto. Ratio above 0.25 and below 4 to have stocks that are about the same price. 50 day correl > 90%. Correl not in a downtrend (correl 5 days ago minus correl today > 0.05 or correl 10 days ago minus correl today > 0.1). Ratio close above 2 * 30 days rolling stdev. I then keep 3 types of pairs; industry/industry, ETF/ETF, Sector/Sector (ETF can be combined with wathever sector). Chart entries criteria: Look at the distance from mean in stdev chart and ratio chart. Not in a trend. Somehow subjective. In a trading range. Not in a breakout. Good risk reward ratio (try to get above 1) Manual entries criteria: Dividends? Play a stock only in one pair. Look at the individual stock charts (no breakout...). Earnings? News, takeover potential? Market cap in the same range. Exits: Somehow subjective. Fixed profit target. Predetermined stop. Think it wont do it (sort of time stop). Come back flat after a drawdown. News.