How to Get Started in Automated Trading
01/03/2013 1:00 pm EST
Kevin Davey is one of the best people to talk about automated system trading. He's won several high-profile trading competitions and is open and frank about what it takes to make money with automated trading. In this interview, we talk with Kevin about how he recommends beginners think about programming their own system ideas, things all traders need to keep in mind when doing so, and how he evaluates systems when deciding to put real money in them. We also discuss the pros and cons of buying off-the-shelf automated trading systems.
Play in Windows Media format
Play in RealPlayer format
Play mp3 stream
Direct link to mp3 file
Tim Bourquin: Hello, everybody, and welcome to another interview. My guest today is Kevin Davey, and we're talking about systems trading and discretionary trading and the combination of both and what he's doing these days and maybe some ideas for you to add to your own trading repertoire in 2013.
So Kevin, first of all, thanks for joining me on the phone today.
Kevin Davey: Well, thanks, Tim. Thanks for having me.
Tim Bourquin: Alright. So what are you up to these days in terms of systems trading? You've been doing this about as long as anybody, and what's working for you these days?
Kevin Davey: Well, you know, just a constant reinvention of new systems and trying to plug in new systems to a portfolio and remove ones that are no longer performing as well as I'd like. What seemed to really be working good now lately in the past couple of years is I've done a lot more diversified type portfolio. So instead of, for example, trading one system with the euro, for example, I might trade five systems with the euro and gold and crude oil and a couple of other markets. And that seems to work a lot better because it smoothes the equity curve a lot of times.
Tim Bourquin: Alright, so you're running these systems simultaneously, then, in different markets?
Kevin Davey: Yes, yes. It's usually where I've had the most success is running different strategies, completely different strategies, so different entry types, different exit types and then different markets and then also a lot of times different timeframes. So I might be running a few systems on shorter time period bars and some of them on daily bars all the way out to, I think, the highest they go is weekly bars.
Tim Bourquin: All right. So how do you initially come up with systems? I mean, I don’t know a lot of discretionary traders out there that have an idea and they're just not sure where to start. Where's a good place to get going?
Kevin Davey: Where I get most of my ideas, at least the start of them, is just by keeping in tune with what's going on in the industry by reading any new trading books that come out or some of the trading magazines. Now a lot of them will give systems out or at least give ideas for systems or indicators and that type of thing, and typically once it goes out in the public domain there's a reason for that and it's usually because the author doesn’t think it works as well anymore because I guess the old saying or the old thinking there is if he thought it was so good, why reveal it to anybody?
So that will be the start. So I'll look at just about anywhere I can for trading ideas and some of these ideas actually come from different fields. So for example, I think it was in the last year or so I was looking at game theory, which has some applicability to futures markets, but it got me thinking. And I actually created a system that uses some of the aspects of game theory, which is kind of a little off the wall type approach.
Tim Bourquin: And when you've come up with a new idea for a system, how far back tested do you take it to decide whether or not you want to try it with real money going forward?
Kevin Davey: Well, usually, what I'll do is I'll try to go as far back as I possibly can. So let's just take the mini S&P, for example. I'll go back to '97, which is when that contract started. I won't go back further when they were still doing the pit trading, but I will go back to the start of electronic contract. Now, on some other markets, it really depends on the timeframe.
So for example, if you're going to be using one-minute bars or two-minute bars there are certain markets where maybe five years ago there really wasn’t a lot of volume in them and you'll end up getting a lot of minutes or two minutes, let's say, where no trading takes place and that can mess up if you're using any kind of indicators. It can also mess up any kind of patterns that you might be using. So a lot of times it's not set in stone what I'll do, but in general it's the more dated, the better.
Next Page: Criteria to set as a floor|pagebreak|
Tim Bourquin: And are you looking for a specific win/loss ratio with a specific drawdown that you say, "Here's my minimum. I want it to win 60% of the time and I want the max drawdown to be 2%" or something? Do you have kind of criteria that you'll set as a floor?
Kevin Davey: Yeah, there's a few criteria and it's usually unfortunately at the very end of the whole process of developing a strategy of when I have my actual go/no go criteria and that is based on some simulation, Monte Carlo simulation where you take all the past trades and you put them in a hat and basically pull them out to create an equity curve and you do that a few thousand times and then you'll get an idea of the probabilities of a certain drawdown for a certain return on your capital.
So really what I use there is a criteria for the return divided by the max drawdown. That becomes probably my biggest criteria, but I've gotten to the point now where I can do some preliminary testing way back when at the beginning. So, for example, if I was doing crude oil and it ran five years or ten years and it was averaging about $5,000 a year just in some simple optimization with a single contract, it's probably not going to cut it. I know it won't cut it later on so then I just abandon the idea and move on to something else.
Tim Bourquin: How many systems are you comfortable running in any one time? And are they automated completely or are you actually just getting the signals and then hitting the buy button or sell button on your own?
Kevin Davey: Right now I'm running, I think, somewhere between five and seven systems and some of them are completely automated. One, for example, is trades five different markets. I consider it one system because I do it as a whole and on separate account and that one is completely automated. Some of the other ones, one, for example, goes back to when I wasn’t working full-time or when I was working full-time I should say and not trading full-time so I needed a system that worked just on day bars. So that one doesn’t need to be automated. It's easy enough and orders at night.
So it's kind of a mixed bag of some automated, some not. The issue obviously becomes probably what you're asking is once you get up a number of systems, it becomes really difficult just to keep on top of them all. So it's very easy to let a trade slide that you don’t see. Maybe your software crashes and then in a while you're rebooting it. You miss the trade and you don’t even know it till the next day.
So the one thing I do is keep a log of all my errors and errors to me are anything that doesn’t match what the historical testing is. So entering late or entering a buy instead of a sell and so on and so forth. So I keep track of that and that helps me to really focus since obviously what you want to do is eliminate all errors which is never going to happen, but if you're making 20% or 30% errors, you in effect aren’t even trading the systems you thought you're trading and that's not good.
Tim Bourquin: So a trading log is important even to somebody who is a systems trader, not just a discretionary trader, it sounds like.
Kevin Davey: Right, right, absolutely, because it's something that can help keep you focused that you're running the systems as you should. Even with automated systems, I use software that automatically does it to the brokerage and probably 90% of the time it works great, but every once in a while an order will pop up and it will recalculate and it will say, "Do you want to sell, or do you want to get out of this position?" Like, what? You have to make a quick decision and so knowing that you're doing the right thing under a little bit of pressure is pretty important.
Tim Bourquin: So even in an automated system you're saying that once in a while the system will hiccup and it won't automate the trade and execute what you want to do.
Kevin Davey: Yeah, absolutely. And part of that is just me probably the way I automate my systems. I don’t have a lot of code that does the automation because typically most of my systems are pretty simple as far as entering at market close, selling at market, try to stay away from limit orders which sometimes have some troubles. But yeah, that's basically it.
Next Page: Systems trading or automated trading?|pagebreak|
Tim Bourquin: Let's talk about the idea of high frequency trading and its effect on discretionary trading. Do you think systems trading or automated trading is the answer to that for the retail trader?
Kevin Davey: Yes. I mean, where there's, just to take the mini S&P, for example, where there is a ton of volume and there's no range. It's only a half point range or a three-quarter point range.
Tim Bourquin: Alright, Kevin, I know that a lot of discretionary traders these days in the retail area are concerned about high frequency trading because it doesn’t allow them to get as much transparency in the market. They see orders come and go quickly and they're having a tough time with that. Do you think automating their system and systems trading, in general, is the answer to that? Does it get rid of that problem that retail traders seem to be having with HFT?
Kevin Davey: Well, it can. It really depends on how you trade and your trading style. If you're going to automate a system that’s a scalping system, I would think you're going to be hit pretty hard with the automated trading with all the high frequency traders. It's going to be a lot tougher. My personal experience, I like the idea of having a scalping system that's maybe in and out ten times a day and gets a quarter point here and there and at the end of the day you have basically like a little cash machine.
But what I've found especially more recently is that's harder and harder to do in any kind of historical testing. It might have worked five years ago or maybe even one year ago, but now it's not working. And those kinds of strategies just don't work, and what that pushes me to do is I always end up going longer term. And so a lot of my trades end up being days to weeks where the HFT people probably aren’t having such a big influence. I'm sure they're still getting a little bit on the spread when I get in and out, but what I've found is trading less is one way different about them.
Tim Bourquin: Right, yeah, lengthening it out and if you're trying to scalp, no question you're competing with a lot, a much different crowd, I guess, I should say than if you went long term. So are you using TradeStation? What are you using to automate your systems and then connect it to the brokerage?
Kevin Davey: I use TradeStation, and I've also in the past used NinjaTrader really as a conduit for getting the signals from TradeStation and to Ninja to a broker. I don’t do that currently, but a lot of people always tell me, "Hey, you need to come over to Ninja and start programming in that." For me it just becomes a matter of do I want to learn another trading language that maybe there's some. I'm sure there's some nice things it does the TradeStation doesn’t do and vice versa, but it's just a matter of time. I only have so much time. It's only me. And I like creating new systems because you really have to keep just a constant flow of systems that are possible things you might trade.
Tim Bourquin: Now, you program your own I take it, but most traders out there can't or don’t have the ability to do their own programming. Are there third-party people out there that you can get on at Elance.com or oDesk or some of those things where you can have somebody program it for you?
Kevin Davey: Oh, absolutely. And there's even in the TradeStation forums, I know there's a lot of "certified" people, it's people who have discussed it with TradeStation and TradeStation gives them some kind of seal of approval of being a reliable person to program things. So there's definitely a lot of sources, but what I tell people usually if you're in this for the long haul and you're really serious about five to ten years of creating strategies, it's probably worth your while to take the time to actually create your own systems by learning the language.
Tim Bourquin: Alright. One of the things I want to talk about is buying a system. Now we should probably preface this with the disclaimer that you do sell systems as well, but there are a lot of black box systems out there that they give off their percentages, wins, and their drawdowns, but you don’t know anything about how they're trading it. They just say, "This is an e-mini system, and here are some stats about it."
What are your thoughts about that? I mean, should I know what 's going into a system that I'm actually trading or is it okay if it's back tested and it's somehow confirmed that just go ahead and buy that and use it regardless that I know what's really going on underneath?
Kevin Davey: It really depends on who the vendor is. If you have some kind of personal relationship with the vendor or for some reason you trust this vendor, that might be an okay approach to actually buy a black box. But you're taking a leap of faith that what the vendor—the number—he gives are actually true but that he developed the systems correctly. A lot of people think they can get TradeStation or Ninja Trader and just fire it up and program, let's say, a simple moving average crossover, optimize it, oh, it looks good, and start trading it and be on their way. I come across people almost every day who do that kind of thing. It's just typically a beginner mistake.
Unfortunately, what a lot of people do is they realize that, wow, this doesn’t work. So maybe I'll become a vendor and sell it to somebody else. Buying a system is really difficult, and I personally wouldn’t buy any system unless everything was disclosed. But that's probably just me but I want to see exactly how it works.
Next Page: Does account size really matter?|pagebreak|
Tim Bourquin: And also something to keep in mind is that some systems are probably better for higher account sizes. I mean, does account size play into which system is best for me?
Kevin Davey: Oh, absolutely. And a lot of people don’t take the time to realize that. A lot of people who contact me, for example, they'll say, "Well, what kind of account size do I need?" And just with the government regulations, I can't really give any person a specific number. It's kind of hard because it depends on the person himself how much they're willing to lose, what kind of return they're looking for, all their goals and objectives.
But what I do see is I see a lot of people who will try trading a system with not enough money. For example, somebody will email me and say, "Well, hey, I have $2,500. What can I trade?" I write them back and I say, "Nothing," unless maybe you want to try the micro lots in forex, the super small ones. That might be a place. Of course, nobody wants to hear that and they usually get upset.
Tim Bourquin: Yeah, they want you to turn their $2,500 into $250,000 with some black box stuff, I would imagine, is what people look for Holy Grail just in a system.
Kevin Davey: Right.
Tim Bourquin: Alright, well, Kevin, this has been great. I mean, we've just scratched the surface here, obviously, in terms of systems trading. But give us your website so our listeners can go check you out and find out about how you program and do systems.
Kevin Davey: Okay. It's www.kjtradingsystems.com. I'm pretty good. It's just me because normally what I do is I'm a full-time trader and I just do this a little bit on the side because a lot of people ask just based on my trading history they know I can produce some good systems. So feel free to email me and we can talk.
Tim Bourquin: Yeah, one of the things we didn’t mention is that you've won a couple of trading contests with your systems in the past years, haven’t you?
Kevin Davey: Yeah. In 2005, 2006, and 2007 in the World Cup of Futures Trading Championship and that was the same contest that years before Larry Williams turned $10,000 into over a million. I didn't do quite that well, but I finished in first one of the years and second to the other years.
Tim Bourquin: Excellent. All right, Kevin, well, thanks for your time.
Listeners, you can check out his website by clicking on the link in the transcript for today's interview. Thanks for your time, Kevin.
Kevin Davey: All right. Thanks, Tim.