Our "system trading" group which does extremely well, use a "system" of sorts, but have had to add a great number of actual trading bodies to effect trades successfully. The "system" cannot (at least has not to date) perform successfully without the human intervention. It seems to be more than a programming issue, or a communications issue, and more like a "reaction" issue (I hope that makes some sense)....I know what I mean, but I may not be explaining it correctly. I am working with two seperate groups to try to define parameters that can quantify some of the reasoning involved....and am actually trying to prove myself wrong (in my assumption that no system will work on its' own),,,stay tuned!!
Our "system trading" group which does extremely well, use a "system" of sorts, but have had to add a great number of actual trading bodies to effect trades successfully. The "system" cannot (at least has not to date) perform successfully without the human intervention. It seems to be more than a programming issue, or a communications issue, and more like a "reaction" issue (I hope that makes some sense)....I know what I mean, but I may not be explaining it correctly. Ahh... good! You were scaring me for a second there. I thought you were saying that systems trading does not work, period! With regards to needing a person to handle executions, that is an essential element of trading today. As technology improves, and automated execution becomes more reliable, maybe 100% complete systems trading will be more feasible. For example, right now you can automatically execute orders through an API to patsystems. There are Tradestation modules that will automatically execute trades for you. But you would be wise to monitor each execution to verify correct placement, fills, etc. You need a person to deal with such situations as computer crashes, network connection going down, etc. I am working with two seperate groups to try to define parameters that can quantify some of the reasoning involved....and am actually trying to prove myself wrong (in my assumption that no system will work on its' own),,,stay tuned!! Just don't overleverage! Read "The Predictors" and some post-mortem analysis about LTCM. When you become the next Bill Gates don't forget to remember the little people.
The "little people" are the traders who have the drive to actually put into practice their strategies, and who then become "big people" in the economic sense.
It's a bit easier for me to reply to the various questions and posts in one place, so here goes: 1. "Discretionary vs. Mechanical" - We have noticed over the years that it really takes a bit of both to do well. The mechanics come into play when entering orders (opening strategies, "easy side" pairing, "print anticipation" and the like), and the discretionary aspect comes into play when "legging out" of pairs, or reading the tape to close trades. We all wish that there was a simple answer, but there isn't. 2. "Trader Skills" Well, yes they are important, in part due to what I said in number 1 about exit techniques. A good trader will run in the 75% plus category for winning vs. losing trades, and this is much better than the random "red vs. green" discussion (we use the heads vs. tails metaphor). 3. "Bright vs. Echo" - well, Duh!! But to be serious, let's try not to make this "firm specific" in any analysis, let's all ask the hard questions of any firm and make logical, informed conclusions. 4. "Sub-Penny" - the Brokers are still upset (as are the Institutions) about going to pennies to begin with. Although it has been great for our traders, we can see why the retail firms are concerned. They are used to making money on both sides of trades (firm vs. customer/customer vs. firm out) and want to keep that teeny difference at minimum. The concern we have is the "jumping in front of orders" that is taking place on many ECN's - it gets pretty pathetic after while. Since we trade mostly listed, we have not had any real problems in that regard. Let's keep it at a penny if we can. IMHO
Don, I've been looking into pairs more recently and had a couple more questions for you on this technique. You were using AOL/VIA.b in your example, so I'll continue with that. I'm only using end of day data right now, just to get a feel for how to work out the mechanics of picking a trade. I'm tracking the spread in excel and using a 20 day moving average of the closing price as a mean to base decisions off of. I've noticed that over the last two months, the spread has been getting gradually wider, with VIA outperforming AOL. How do you deal with that? Do you just look at a very recent average, or do you look for fundamental reasons why this might be changing? I understand why putting on the short is the hard side of the trade. If the stock is moving down, it's hard to get a fill, and hard to figure out what the spread will be by the time you do get filled. How do you decide what side to put on first? Or are you strictly looking at each stock and trading it off the tape, only putting on the pair when it moves against you? The price correlation on these two stocks is about .85. Wouldn't you want to account for this when deciding position size? For example long 1200 AOL/short 1000 VIA, rather than 1000 each? Do you use a standard deviation amount, or only look at the price spread when deciding what's a good risk/reward? Thanks for your help on this topic. BTW, the spread got out to about $7.40 today, did you take this trade? Corey
Proof in the Pudding------ Don Bright, I wonder if you could tell me how many of your offices around the country are actually profitable. I mean like more than one trader in each office making a living trading at your offices. Since you've indicated the Las Vegas Office is doing well we will count that as one. Any others. Those trading at any Don Bright offices may also shed light.
After the approximately 1 year "learning curve" (which we all must go through), we have over a 70% success rate with all of our traders. Every office has successful traders, and nearly every office has new people getting to that point. Since we are subject to regulatory scrutiny, I cannot even post numbers like this unless they are true. I am pretty confident that our Calendar year 2001 numbers will be even better (I will be happy to post them when all the accounting is done). We added approximately 250 traders during 2001 (so far), and even with that number of new people our winning percentage is up there, and as I said, I will post up 2001 numbers early next year.
You miss quite a lot when you use only the closing prices to evaluate pairs trading. A big chunk of the profits are made by trading intraday. We often only trade the "strong" or "weak" side of the pair in relation to momentum, leaving the other side as a "crutch" only (when your entry goes against you). I may only add to the expanding position 3 or 4 times before starting to close when it is going against me, but I will trade it perhaps 10 times in and out, scalping dimes and so forth. In the AOL/VIA.B example, I have traded it from about 4.00 to about 8.00 in the last 3 weeks or so. I like to sell the VIA first when it gets above 6.50 or so, and then trade out of it. My last opening trade today was to sell it at 7.66 (I had covered my 7.30 trade (rom yesterday) at around 6.80 this morning at the opening). Hope this helps!!
Since a couple of more of you asked about the NYSE order book "going public" - All I can say for sure now is that it will be a while before we see the actual depth of the Specialist's order book. It won't really be of much help, since the professional traders will use it much like they manipulate the Level 2 screens (showing big bids/offers just away from NBBO). The NYSE is testing the 10,000 share minimum depth within 10 cents on selected stocks now, and I think that this is more important than showing some "possible" orders below the NBBO. Stay tuned as they say!