Position sizing is the process of determining HOW MANY contracts to trade when a trading system gets a signal. It is one of the strongest concepts available to traders and yet regularly the least accepted. Position sizing should manage risk, boost returns, and improve robustness thru market normalization. Position sizing can end up being far more significant than where a trader buys or sells! But most trading systems and testing platforms either ignore position sizing or use it illogically.
A serious problem with many trading systems is that they risk way too much of a trader’s equity on each trade. Most professionals agree that a trader should never risk more than 1% to 3% of his equity on any given trade. This same concept applies to the total risk for each sector. As an example, if a trader is risking 2% of his equity on each trade in some highly correlated markets such as 2 year bonds, 5 year bonds, 10 year bonds, and 30 year bonds, this is essentially the same as risking 8% in the same trade. Though over trading in this fashion can produce phenomenal looking results with returns of 100 percent or more, this is mostly merely a case of using too much leverage, taking too big a share of risk on each trade ( or sector ), and / or "cherry picking" the best starting date ( for example, right before a sequence of winning trades ).
When running a study of the worst-case scenario at those high-risk levels, it becomes clear that the chance of ruin climbs dangerously high. A sequence of losing trades, or even just beginning on the wrong day, could cause an investor to lose everything ( or at the least have a big drawdown ).
The final analysis is that when putting on a trade, a trader should know what proportion of his equity he will lose if he is wrong. This should only be a small part of his available trading capital. This also implies that he should know the risk he is taking on when entering a trade. Some trading systems, moving average systems, for example, don’t even know how much risk they’re taking. This is because the trading system doesn’t know how far the market needs to move to trigger an exit. We think it is perilous to trade this way and do not endorse it.
Another large problem is the lack of market normalization ( such as single contract based results ) in trading systems. We do not think it is logical, for instance, to trade one contract of natural gas with a standard daily volatility of about $2,000 for one Eurodollar contract with an average daily volatility of almost $150. To do that would imply that the natural gas market is more important than the Eurodollar market. If the Eurodollar market trends, we wish to give it just as much weight as the natural gas market ( or any other market ). In the prior example, a trader could simply take away the Eurodollar from the equation and get virtually the same performance. In essence, the results are unintentionally biased ( curve fitted ) toward natural gas. A $150 average winning trade in the Eurodollar is not going to nullify a $2000 average losing trade in natural gas!
We endorse trading a basket of commodities for diversification, but if traders do not normalize the info and almost all of their profits and losses arise from only a few of the markets in their portfolio, that’s clearly not diversification. The problem is that as time goes forward, traders are going to be dependent on that tiny scattering of markets to perform. It is way better realizing that all markets have the potential to perform at an equal level instead of being dependent on only a few of the markets in the portfolio.
Most software packages design work on a one contract basis. It is ( likely ) for that reason that most trading systems ignore position sizing or use it illogically. Of the many back testing products on the market for sale ; we are only conscious of two programs that can correctly perform position sizing and money management testing. Although there are lots of products which claim to do it, we’ve found that nearly all these products aren’t able to perform position sizing and money management correctly ( there are lots of reasons for this, please be at liberty to get in touch with us for details ). We use Bob Spears’ state-of-the-art testing software Mechanica for most position sizing based research and testing ( it sells for $25,000 a copy ).
Other Problems include sellers that only report smaller drawdown numbers like "closed trade" drawdowns or "average annual" drawdowns. There also are problems with position sizing ideas like "optimal F" or "Fixed Ratio." We feel that both these concepts are just a perilous sort of hindsight biased curve fitting.
Another common misconception says that traders should find their "best" single contract based trading system FIRST and THEN apply position sizing to it. This isn’t the right approach. Position sizing can change the risk-to-reward profiles of any single contract based trading system, a trading system that may have looked brilliant, with a smooth equity curve when on an one contract basis, can look much less interesting when all markets are similarly weighted for robustness.
For all the reasons cited above, we here at DH Trading Systems develop trading systems with correct position sizing logic. We think this not only raises the robustness and significance of the testing results, but can also help to bypass the inadvertent optimizing that may happen with other kinds of position sizing / money management based testing software.
