# Risk of 20% account loss

The difficulty I have with the Risk of 20% account loss is it does not seem to be consistent. With my SMA signal the risk is now 23.8% relatively high. It is very seldom that high. I know it is recalculated every weekend. Why would the high reading on this signal come when the signal itself is at an all time high? The problem is when the risk goes over 20% you get booted off of the best signal list and best stock signal list. So last week the risk measured 9% and I was on the best list. This week with the signal on a new high I am not showing on best list? Seems inconsistent to me.

Rick Haines

this indicator should be replaced with "Risk of Ruin" indicator. It is an industry standard and far more valuable for telling the risk of a system.

Index how do you calculate Risk of Ruin?

Rick

here are a couple of items, but many more on the web.

http://en.wikipedia.org/wiki/Gambler’s_ruin

http://www.poker-tools-online.com/riskofruin.html

Risk of ruin is applicable to gambling systems where the odds are defined and the amount that can be won or lost are about even (a binary distribution). If the "house" has a small edge, 52 percent versus your 48 percent, on aq given hand and you either win or lose your bet, risk of ruin can calculated.

In trading, almost no system either wins or loses a given amount. The distribution of trade outcomes is usually normal, or lognormal with "fat tails" on either end. Risk of ruin is not a straight forward calculation for this type of distribution. To get around that, Monte Carlo analysis is usually done to get an approximation of the probability of occurance of a drawdown of x percent or higher.

The problem with Monte Carlo analysis is that it ignores the serial correlation between trades (winners or losers come in bunches) or the correlation between groups (grains and meats, or related stock groups). The answer given back by a Monte Carlo analysis ignores the Money Management a trader wraps around his system.

Keith,

Thank you for your very enlightening post. I have been missing them for quite some time now.

Traditional Monte Carlo analysis not only ignores relevant data as you mentioned, it makes many false assumptions on the data provided.

It’s better than nothing in that it teaches caution, but beyond that is not particularly useful.

How does the Random Equity Curve Generator fit into this?

Rick

2 Keith Fitschen

Correct and very good point. Gambling (poker) RoR isn’t quite applicable for trading.