**BarroMetrics Views: When Is A Profit, A Profit? 2 (Success Series)**

In the last issue, I considered the question: when do we add profits to capital? When do we deduct losses from capital? I introduced Ryan Jones’ approach.

(An aside.

Speaking of Ryan Jones, he has a sale on for his course starting Tuesday, April 19 and running for two days: see Fixed Ratio on Tuesday.

His course is normally USD 797.00; on the sale, his price will be, USD 197.00. And ‘no’ to those who will ask, “I am not receiving a fee for referring the course’. In fact, Ryan does not even know I have mentioned the saving).

Turning back to my subject……

Ryan’s approach does not suit me. I prefer to use another method – not better, just one with which I am more comfortable. My take requires we know:

- Number of historical consecutive losses
- Avg consecutive losses
- Loss Rate
- Population Size
- Avg pa return on capital (Avg ROI)

Let’s take my trading. Taking my trading results since 1990 (when I began the private limited partnership fund), we have the following results:

- Historical consecutive losses: 16
- Avg Consecutive losses: 4
- Loss Rate: .485
- Population size: 2210 (i.e. number of trades over the period, about 7 trades per month)
- Avg pa return on capital: 27.63%

Firstly, the data allows me to calculate the theoretical consecutive loss: 10.64 (say 11). So, I would not consider the consecutive loss sequence of 16 as an outlier. (If I did, I would need to reassess the data. For example does the data show a period of extraordinary losses?) With a run of 16 losses, if I risk 2% per trade, I’d risk a loss of 32%. Since my average profit is only 27.63%, I am not prepared to run that risk.

What if I risked 1%? My possible loss would be 16%. Still a little high – I am aiming for a risk of around 50% of average ROI (14%). So, I’d bring down the risk per trade to around 0.90% which I find an acceptable risk; I treat the 0.9% as my ‘normal’ size.

Secondly, when to add to capital? I’d add on any increase above 50% of average ROI = 14%. And finally, when to subtract losses from capital? I’d subtract losses whenever I’d lose 50% of the increased value (i.e. 50% of 14 = 7%).

So far we have determined what I would consider ‘normal size.’ I have one more step to determine my position size for the current trade: *I have to decide whether my trading is currently in Ebb, Flow or Normal State. *

More on this tomorrow.

Apologies if you have posted comments.

We migrated to a new platform. In the move, we lost your comments Do please post again.

Thanks

ray

re

“5.Avg pa return on capital”

I would like to confirm my understanding of your post.

By capital you mean the $ amount that you are basing your return calculation on at any given time and the $amount used for position sizing calc as opposed to the actual account balance.

For example, the actual account balance could be $107,115 but the “capital” figure being used for position sizing and return on capital calculation might be $102,150.

Re

“Firstly, the data allows me to calculate the theoretical consecutive loss: 10.64 (say 11)”.

How was that calculated?

Hi Chris

Best explain by way of example.

Let’s say I start with $100,000.00 – that’s my capital. I make $8,000.00. That’s less than 14% so, my capital is still $100,000.00.

If I now lose 7% of $100,000.00, I reduce my capital

to $93,000.00. The $8,000.00 is held in reserve as it were. I would now need to increase my $93,000.00 by 14% (including the $8,000.00) to add profits to the $93,000.00

But, let’s then say instead of losing, I make another S7,000.00. I’d this profit to the $8,0000.00. I have now made 15%. So, I add the $15,000.00 to $100,000.00.

I know have $115,000.00 as my capital base.

Thanks Ray.

Perfectly clear.

chrij posted this comment. I approved it but the post has not published:

““chrisj said, in April 19th, 2016 at 7:33 am

Re

“Firstly, the data allows me to calculate the theoretical consecutive loss: 10.64 (say 11)”.

How was that calculated?”

Hi Chris

The formula I use:

(log of runs)/(-log(winRate)

Can you provide a working example using the stats in your original post as inputs to the formula.

Hi Chris

I’ll try – all I can do is give you some inputs, Excel does the rest for me.

Let’s say my population is 2210 and my loss rate is .485. The formula I use is:

(log of runs)/(-log(loss rate) = longest losing streak.

The quotient gives an occurrence of 11.

I have seen another formula: (log(1-DC)/log(1-PW)

Where DC = Confidence Level (usually .99). The result gives an occurrence of 7 where the LossRate is .485.

I prefer to use the first formula.

A Google search will throw up probability calculators. I usually don’t use them unless they disclose the formula they are using.

For any one else who may have gotten confused this may help.

The earlier post had:

(log of runs)/(-log(winRate)

Later corrected to:

(log of runs)/(-log(loss rate)

i.e. loss rate.

I did a little spreadsheet and with the revised formula I get the same result as in Ray’s original post.

I will send my spreadsheet to Ray so he can upload it as an attachment if he is okay with doing so.

Chris

Hi Chris

Thanks.

I’ll upload tomorrow.