Risk of Ruin
Author: abegy
Creation Date: 5/17/2015 2:46 PM
profile picture

abegy

#1
Hi Eugene,

I see in the website myfxbook http://www.myfxbook.com/members/fxgrowthbot/forex-growth-bot/71611, the information "Risk of Ruin". Is it possible to add something similar in WLD ?

You can find formula here http://www.automated-trading-system.com/resources/risk-of-ruin-and-drawdown-calculation-tool/.

Thanks in advance ;-)
profile picture

Eugene

#2
Hi Alexandre,

Please refer to our previous conversation in this thread and to my post #10 in particular:

NEW MS123 Performance Visualizers and MS123 Scorecard

Is this formula any different?
profile picture

abegy

#3
Yes Eugene : you are right. This is the same subject.

In our previous discussion, I asked to have it as a metric. Now, what I see in the website myfxbook looks good for me : a table with different steps).

it could be an interesting visualizer extension in my point of view. Indeed, this is more simple to read and to understood that the monte carlo result.

The formula in my previous post is different that this one because there are different formulas. I cannot tell you which formula is the best one.

Tell me if you want additional information.
profile picture

Eugene

#4
Do you also see that MyFxBook's formula implementation is quite funny? That "Forex Growth Bot" system has had a -94.5% drawdown yet the risk of a 90% ruin is estimated at below 0.01%. I think it's rather on the opposite end of being close to 100%. ;-)

Let me see if it's feasible.
profile picture

Eugene

#5
After researching the subject, I have one serious concern. Both Jez Liberty's Javascript code that powers his RoR calculator and Ralph Vince's original formula (as cited in Chapter 22 of Mr. Kaufman's book referred to by Jez Liberty), focus on a fixed amount to come up with risk of ruin figure i.e. percent-based "bet size" (Liberty) or fixed-dollar "investment" (Kaufman).

While it may be fine for a web-based risk calculator, the formula as we know it does not seem directly applicable to a trading system. How do you find out the "Risk Amt (%)" aka "bet size" of your system being backtested? It may be evident with a percent equity sizing only. (Entering an arbitrary figure will not work out.) I wonder how MyFxBook does it.
profile picture

abegy

#6
Eugene, I would like to clarify the concept of the "Risk of ruin". The idea is to know what is the probability for your equity to loss x% (or x$). It's that why you must know some parameters of your trading strategy (%win, win/loss ratio...). This approach is a simplification of the monte carlo simulation concept. Now, you can understand that the x is not the way you calculate your sizing position.

Regarding the difference with the approach discussed in the "futuresmag" article is to add another axis of analysis. The idea is to know what if your probability to loos x% (or %$) of your equity if there is a deviation of your expected return. It's that why the formula introduce another parameter : standard deviation.

is it more clear now the concept ?

PS : in this link http://seor.gmu.edu/projects/SEOR-Fall09/ISG/Investment_Optimization/Resources_files/Risk_of_ruin.pdf, you will see graph putting in the original article in "futuresmag".
profile picture

Eugene

#7
QUOTE:
The idea is to know what is the probability for your equity to loss x% (or x$).

If only the RoR formula accounted for equity. Problem is, it does not.

QUOTE:
Now, you can understand that the x is not the way you calculate your sizing position.

I understood that. Without the x resulting from the system's actual position sizing, your RoR does not really belong as a performance visualizer. It's rather a calculator tool like on that web page that will return similar values for any given system of a certain profile (e.g. counter-trend system with high win % + low WinLoss ratio, or a trend-follower with low win % but a high Win/Loss) with similar position sizing applied.

The "Risk Amt (%)" aka the x should not be an arbitrary value if we're on the same page. Maybe you envisioned it as an addin rather than a visualizer?
profile picture

abegy

#8
After reading this article http://bettersystemtrader.com/riskofruin/, I understand why you are speaking of position sizing.

Let me time to come back to you with a propisition
profile picture

abegy

#9
Eugene, I would like to propose you the following strategy to implement the Risk of Ruin.

As the idea of the "risk of ruin " is to calculate the probability to loss your capital regarding the risk that you take on each trade, what do you think to apply it :
- with it's original definition when you select the position size option : max percent risk
- for the other position sizes, select the percentage of the largest loss regarding the equity at this time to approximate your maximum risk
profile picture

Eugene

#10
Alexandre,

That makes sense in general. I'll enter your request in our product backlog.
profile picture

abegy

#11
Thanks a lot Eugene :-)
profile picture

Eugene

#12
Alexandre,

1. I'll take Ralph Vince's RoR formula (from Kaufman's book) which accepts "Investment is the amount invested (e.g., $10,000)". As this usually is a variable, I plan to take arithmetic average of the position size (i.e. average trade size). Makes sense?

2. Do we include open trades in the calculation or consider closed trades only?
profile picture

abegy

#13
Hi Eugene, I fully agree with your proposal. it looks to be a good solution.
profile picture

Eugene

#14
Answering my question #2: open trades will also be included in the RoR calculation.

Risk Of Ruin is to appear in the upcoming version 2015.07 of the library, on the Performance+ tab. For each column (L+S, L, S, B&H), it'll show risk of ruin for -10%, -25%, -50% and -75% of equity. I think it's going to be sufficient without crowding the tab.