High Probability Trading

How to get a sane high-probability trading system

1. Do you know your win/loss ratio?

Sit down and add up every winning trade and every losing trade over the past year. Seriously—sit down and do it. Divide the winners by the losers to get your win/loss ratio. Let’s say your trading generates a win/loss ratio of 2:1, meaning a $2 gain for every $1 lost. This is pretty good, right?

Yes, but it doesn’t go far enough. The win/loss ratio doesn’t link the gains and losses to the amount of capital you put at risk and doesn’t help you figure out the expectancy for the next trade. Maybe you made one trade in the year that netted $200 on a capital stake of $10,000 and one that lost $100 in the same year on the same $10,000. You still have a 2:1 win loss ratio but the rate of return on capital is tiny and the frequency of trading is not high enough to deduce expectancy for the next trade.

2. Do you know your expectancy? Here’s the formula:

Average profit per winning trade x winning trades as a percentage of total trades

minus

Average loss per losing trade x losing trades as a percentage of total trades

Let’s say you find a set of indicators that would have generated this profile:

($800 x 35%) – ($400 x 65%) = $280 – $260 = $20

In other words, the indicators deliver a home run of $800 on 35% of the trades but big losses on a majority of trades, so your expected gain per trade on the next trade is a mere $20.

Here’s a more realistic expectancy:

($400 x 55%) – ($200 x 45%) = $220 – $90 = $130

The expectancy formula above indicates you had 55% winning trades and each gained $400, with 45% of total trades being losers but at a cost of $200 each. Using these indicators, your expectancy for the next trade is a net $130. This is a whole lot better than $20, but $130 on how much capital and over how many trades?

3. How much capital is at stake and how often do you trade?

If you are trading only four times per year, you might have a high probability of making 4 x $130 = $520, but if you are doing it on $10,000 starting capital, it doesn’t pass the “So What?” Test. The “So What? Test asks how much money you want to make as a multiple of starting capital.

Capital Goal = Starting Capital + (Expectancy x Capital Stake per Trade x Total Number of Trades)

Let’s say you have $10,000 to put into technical trading and you want to double it over a year to $20,000. You know how to fiddle with the arithmetic:

$20,000 =Starting Capital + ( $130 x number of trades)

If you traded the full amount of $10,000 on every trade (and the expectancy remains stable at $130 per trade), you would have to make 76.9 trades per year to double your money:

$20,010 = $10,000 + 10,010 ($130 x 77)

Two problems: you should NEVER, EVER put all your money on the first trade. This should be obvious to a second-grader. What if the first trade is one of the losers and you know you can expect to lose 45% on any losing trade? Your capital take is now $10,000 minus $4500 = $6500. Now you are no longer seeking to double your stake but to more than triple it.

Second, the number of trades per year is largely dictated by the system you select. Most indicators have a built-in sell signal and in addition, you should have a stop-loss regime in place that exits trades when conditions turn against you. If you pick a system plus stop-loss regime that trades (say) 35 times per year, you can expect to make $4550, not $10,000. To get to a $10,000 capital gain in one year, you have to pick a different set of indicators and a different stop, which now changes your expectancy and you have to start over at Step 2 above.

Final Lesson: Technical analysis offers two big benefits over traditional “value-based” trading. First, you are following the market, not trying to dictate to it. Second, by adopting a rational, systematic approach, you do not have unrealistic expectations. Drilling down to expected total capital gain on the basis of your trading system’s expectancy and trading frequency plus your starting capital stake is the sanest of all possible approaches.