By Van Tharp

1. Whenever you enter into a position, always have a predetermined exit point at which you will concede you were wrong about the position.

This is your risk (R), and if you lose this amount, you have a 1R loss. Even if you are a buy-and-hold investor, you should have some point at which you will bail out of an investment because it is going against you (e.g. a drop of 25%). This rule essentially sets up all position sizing rules.

2. The golden rule of trading is to cut your losses short (1R or less) and let your profits run (more than 1R, i.e. a multiple of R).

Let’s say you buy a stock at \$50 expecting the price to go up \$10, a 20% gain. You decide in advance to exit if the price falls by \$1. Now assume that you have four failed breakouts (i.e. 4 x 1R losses) before you have your \$10 gain (in this case a 10R gain). You were right only 20% of the time, but your losses totaled minus 4R and your profits totaled plus 10R. Your total gain was thus 6R, six times your initial risk.

3. When the total sum of your R-multiples for all of your trades is positive, you have a ‘positive expectancy’ system. You must have a positive expectancy system to make money in the market.

Expectancy is the sum of your R-multiples divided by the total number of trades. Thus, if you have 50 trades which give you a total R-multiple of 20, then from your 50-trade sample, you would estimate your expectancy to be 0.4. In other words, over many trades, on average, you will make 0.4 times your initial risk on every trade.

4. A low risk idea is an idea with a positive expectancy that is traded at a low enough risk level to allow for the worst possible contingency in the short term so that you can survive to achieve the expectancy over the long term.

This basically means that ‘how much’ you risk on any trade is critical. ‘How much’ is what we call position sizing. In my opinion, aside from personal discipline, it is the most important factor in your trading.

5. Anti-Martingale position sizing strategies work.

Martingale strategies do not work. Martingale strategies are strategies that have you risking more after you lose, such as doubling your risk after a loss. Because people tend to have long streaks against them, they do not work. Eventually you will go broke. In contrast, anti-Martingale strategies, which cause you to increase your position as you win, tend to be very successful. In general, strategies which are based on increasing your bet size as your equity goes up are anti-martingale strategies and they work well.

6. A simple strategy that will work for everyone is to risk a small percentage of your equity on every trade, such as 1% or less.

If you have an account that is worth \$100,000, then risking one percent would mean risking \$1000. If your stop (i.e. 1R risk) is \$5, then you would buy 200 shares (i.e. 1000 divided by 5 = 200 shares). Furthermore, if you applied a 1% risk to the example given in Rule 2, after 5 trades you would be up about 6% since you would be gaining 1% per each R-value. You would be up exactly 6%, since you would only be risking 1% of your remaining equity on each trade.

7. You need to know the R-multiple distribution of your trading system to determine your position sizing strategy.

We frequently play trading simulation games in our workshops in which the R-multiple distribution of the potential trades are known but the value of each individual trade is unknown because the trades are selected randomly from the sample (i.e. a bag of marbles) and replaced. People can become very good at determining their objectives and achieving them in this sort of game.

8. Strategies that are designed to achieve only the maximum return (such as optimal f; the Kelly criteria, etc) are foolish and usually result in huge drawdowns.

For example, if you trade a system that is 55% 1R winners, 5% 10R winners, 35% 1R losers, and 5% 5R losers, then the percentage risk that will achieve the highest average return is 19.9%. With this percentage, you could achieve a huge return if the right sample occurs (i.e. all 10R winners), and this would also give you a very high average return, but you would generally lose a large amount of money on most samples. In other words, you might get one sample in which you make a total of a billion dollars, and many samples in which you lose money. If this were the case, you would have a high average ending equity (because of the huge return in one sample) even though most samples lost money.