Kelly Betting Journal

[donpiano]

My trading approach has a 2:1 win loss ratio with a 40% win rate
The Kelly formula says the optimal bet fraction for my trading system is 10%

Starting account size: £3760
Starting bet size: £376

 

[donpiano]

The system I trade is a 'go with the trend' type of system. I do not trade against the trend or momentum.

My system gets on average 17 trades a month. So not every day.

Also just to clarify my 2:1 win loss ratio is not fixed, it is an average. On some trades I lose less than 1R and sometimes i make less than 2R, and sometimes more than 2R.

Slippage aside, I never lose more than 1R. I always have a worst case stop loss in the market at 1R.

 

[donpiano]

Long losing streaks are common with a 40% win rate system. 5,6 and 7 losers in a row happen often. 10 losers in a row are bit more rare but should be expected.

Sometimes this kind of very nasty pattern can happen:

7 losers in a row followed by 1 winner followed by another 9 losers in a row.

My system has been fully backtested going back many years, this should in theory give me full confidence to stick with it during long losing streaks.

I have a 9R monthly stop loss in place for the month, If i am down 9R or more, i do not take anymore signals for the month. I never want to lose more than 10R in any one calendar month. According to my back tests this kind of losing month doesn't happen often, around once every couple of years.

 

[donpiano]

"Jumping the system"

I will use this term to describe any action that causes a trader to deviate from following a trading system correctly.

There are many ways a trader can 'Jump' his or her system. But the common ones i suffer from are

• Not taking a valid entry.
• Taking a loss early or taking a profit early.
• Not trading at the correct position size (eg betting £150 on the next trade, even though the system says bet £300)

All of these are based around the Fear of a loss, we often get a strong feeling that the next trade is going to be a loser.

During a losing streak we fear more losses. Even after a long winning period we can also become more fearful and don't want to give back our hard won profits.

This is rooted in the gamblers fallacy bias. Which is a common bias experienced by human traders. We often want to predict what our equity curve will do next. Just like we try and predict what the market will do next!

Really it is just over thinking and over complication on the traders part.
The trading process is relatively simple, but we complicate it with our emotions and biases.

My trading can also suffer from another emotional issue. If i am in really bad mood one day, for a reason completely unrelated to trading. For example, i didn't get enough sleep and so i am feeling agitated or feeling depressed. The chances of me jumping my system are greatly increased.

 

[donpiano]

Back to the Kelly position sizing concept which is the title of thread.

Apparently John Kelly chained smoked his way to an early grave (aged 41, in 1965). But he did leave the world with his formula for optimal betting.

His formula tells me to bet 10% of my account on each trade.

This is the 'Full Kelly' amount for my trading approach.
Some traders like to use 'Half Kelly' instead of full Kelly. With half Kelly the drawdowns are much lower. The profits are also lower, but still very good.

I don't like to do things by halves. And i can take the drawdowns that come with Full Kelly because my account isn't very big.

If i had a much bigger trading account, i would not bet anything like 10% or even 5%, I would bet 2% and eventually 1% and eventually 0.5% when my account was at £10 million.

 

[donpiano]

In my post above i said i have a 10R stop loss for the month.

How does this fit with my bet sizing?

Well it means if I hit the stop loss for the month, I will lose about 65% of my account in one month. I will be down to just 35% of what i had when i started the month.

If i make a 10R profit, with compounding effects, i will make about 150% for the month.

But my upside is not capped, i could make much more than 10R and so make well over 150% for the month.

The probability of making 15R, 20R or even 30R in one month, is much lower than say making 5R for the month. But it can happen.

Any month over 20R is an outlier month. These don't come around often based on my trading approach. If i took more signals my chances of making 20R would be higher. But i can only take the signals that my system gives me, which is about 17 signals per month.

 

[donpiano]

Two losing trades since last update.

Happy to report running at 100% flawless execution. No jumping the system or loss of discipline.

This get harder the longer the losing streak lasts.

Current account size: £3308
Current Betsize: £330.8