Sirloin Stock Information (SM)
July 30, 2005
Four Key Components To Building A Trading System
By James Brumley
Need some insight on what you should really be striving for when you're building a mechanical trading system? When it comes down to it, there are really only a few criteria that are used in judging the merits of a trading system. The most obvious one is profitability - does the system work? But really, there's more to it than just that. The number of wins versus the number of losses is important too, but there's a lot of latitude there if the profitability is high. The size of the average win versus the size of the average loss tends to be held as important, and it is. However, that criteria is correlated to the number of wins and losses, so again, there's a lot of leeway there. The one thing that is too often overlooked is the consistency of a system. The fancier term for this is 'drawdown', but it's just a matter of consistency.....you'll see why below. Each of these four components is examined below, and then some of the common mistakes made when folks start building trading systems are discussed.
1) Profitability. You wouldn't think this would be tough to figure out, but building a system that actually works over a long period of time isn't easy. But what you really want to make sure of is that your software is running a hypothetical portfolio the same way you trade. Your software should allow you to specify a dollar amount for your total portfolio, and a dollar amount or a set number of contracts for each trade. That allows you to allocate just a portion of your portfolio, say 10% per trade, into the trading system to give you some real-life trading results. The thing you absolutely must do is factor in commissions into your trading. Most software can do that, but if yours can't, then do it manually. Once that's done, the final test is this.....does your system beat the market, or would you be better off in an index? Or, if the market is losing ground, is your system at least profitable to some degree.
PITFALLS: Many system builders run a hypothetical trading system over a long period of time (like the last five years) to make sure the system is an 'all-weather' type of system. Rather than run a system over five years, run it over five separate one-year periods. Why? You may find that one of the years is VERY profitable, and the other four years are losers. Your system can't be a one-trade-wonder. It has to be profitable in many environments.
2) Win/Loss Ratio. This is just an extension of the pitfall mentioned above (about systems applied to a long-term timeframe). One winning trade and nine losing trades may have been (net) profitable if your win occurred in the red-hot tech rally in 1999. That one win was the fluke though. The other nine trades are most likely what you're going to experience on an ongoing basis. So what should your win/loss ratio be? Some new traders think you need to win on at least half of your trades to make it worthwhile. Others think you need to win at least 2/3 of the time. If only!
The reality is that even the best traders win less than half of the time....it's just that their winners are much bigger than the losers (we'll get to that in a second). I'd say shoot for a system that wins about 40% to 50% of the time. Is your tested system showing wins more than 65% of the time? That's great, but I'd be skeptical of those results. We've been doing this a while, and when the success rate of a system starts to outperform everybody else's by that much, there's usually something very unique about it.....and it's usually something that won't be part of the equation going forward. In other words, if it's too good to be true, it probably isn't. This is often the case when a system is tailor-made for a certain timeframe or certain chart. All the criteria and parameters of a system are optimized for all the little nuances and unusual movements that occurred during that specific period. Those nuances and movements, though, may never occur again. If you're winning 40% to 50% of the time, and you're doing so in several different timeframes (as mentioned in the 'profitability' comments), then you've got a good system.
PITFALL #2: An acceptable win/loss ratio and average win/average loss ratio are inter-dependent. If you can win up to 50% of the time with your system, then you may not need to have your winners be enormously bigger than your losers. If you're winning less than 40% of the time, you'll probably need your winners to be three times a big as your losers. If you're serious about building a system, you have to know and respond to both numbers. (be sure to see below)
3) Average Win/Average Loss. How big is the typical winner compared to the typical loser? Obviously, winners need to be bigger than the losers for the system to be worthwhile. At a minimum, your winners should be at least twice as big as your losers. That may sound easy, but it's not.
PITFALL #3: A lot of traders have high win/loss ratios and strong average winner/average loser ratios with their systems. Unfortunately, they may only get to trade about twice a year. Unless they're putting their entire portfolios into that one trade (which is crazy), the system doesn't do them much good. Make sure you're getting a high enough trade count to fit your trading style and desired activity level.
PITFALL #4: Make sure you understand that most of your winning trades will be very small wins. You'll only have a handful of mega-winners, but they will significantly pull up the size of your average winner. That's ok. Even the best of systems can't predict how big the win will be - they can only guess as to which direction the market will take. Even if the system doesn't result in a homerun on a particular trade, as long as it doesn't wipe you out, it's a good system. You only want your system to get you in a trade when there's a chance of a big win, and it should get you out of the market when there's little to no chance of a big move. Most trades will just be mediocre.
4) Consistency (Or drawdown). This may be one of the least mastered components of system trading. In a nutshell, 'drawdown' just refers to the biggest string of dollars lost at any given time using the system. For example, say you started with a $100,000 account, and built it up to $160,000. Along the way, say you took the balance from $150,000 back down to $120,000 before it went up to that $160,000 mark. Your drawdown would be $30,000 ($150K minus $120K). Or, in terms of percentages, it would be a 20% drawdown ($30K/$150K = 20%).
Why is that important? Trading gurus disagree on the issue. Some would argue that you have to limit your drawdown as a defense against losing any capital - a mathematical rationale. However, if you've created a system that is (1) proven to be profitable, (2) has a good win/loss ratio, and (3) the winners are a lot bigger than the losers, than the drawdown shouldn't matter. After all, a good system will always overcome short-term losses. The reality is that the most important reason to understand drawdown is inside your head. How much loss can you stomach before you give up on the system?
There is bound to be some disagreement about this, but you should worry less about the degree of drawdown, and more about the total number of consecutive losing trades the system will probably produce. This recognizes that even with trading systems, which are designed to take emotion out of the decision, there's still an emotional impact. Even if your losses and your drawdown are small, how many losing trades are you really going to accept before turning the system 'off'? Four? Five? Ten? Try three. Yes, three. There's something about the number three that humans seem to respond to (three strikes in baseball, The Three Musketeers, "three's a crowd", etc.) If your system results in three consecutive losing trades, odds are that you'll abandon it. For that reason, I recommend striving to limit your total number of consecutive losers in your backtest to two. THIS WILL BE TOUGH TO DO! If you stick with the system, then the profitability will take care of itself, but you have to make sure it's a system you can tolerate. Two losers is the limit for most people.
As a review...
1) Systems should be profitable in several distinct timeframes
2) Between 40% and 50% of your trades should be profitable
3) Average wins should be at least twice as big as average losses
4) Worry less about dollar drawdown, and more about limiting consecutive losers to two
Hopefully we've given you a specific set of criteria to shoot for. If you're not yet using a trading system, you should consider applying one. It will take your trading success to the next level, if applied properly.
James Brumley is the chief analyst at Bluegrass Portfolio Management. After spending time as a broker, he established an independent investment research firm. He now manages portfolios, and you'll find his market commentary and analysis on several financial websites.
Article Source: http://EzineArticles.com/