How does the pareto principle apply to trading?

A place to discuss anything.
Post Reply
arbitrage16
Posts: 532
Joined: Tue Feb 14, 2017 7:27 pm

80% of losses coming from 20% of your trades/setups...?
Jukebox
Posts: 1576
Joined: Thu Sep 06, 2012 8:07 pm

Didn't you at least ask yourself that question first?
User avatar
Derek27
Posts: 23468
Joined: Wed Aug 30, 2017 11:44 am
Location: UK

If 80% of your losses come from 20% of your trades, your total losing trades (100% losses) will be greater than 20%.

The 20% figure that you credit for 80% of your losses will depend on whether you've taken the smallest losses (and possible left out one that amounts to the remaining 20% of losses), or take the largest losses and leave out 20% of the smallest losses.

Here's an example. A guy has total losses of £100, 9 X £3 and 1 X £73, and a total of 10 winning trades, 20 trades in total.

You could say 73% of his losses came from 5% of trades your you could say 27% of his losses came from 45% of his trades, so it's not very meaningful.
User avatar
marksmeets302
Posts: 527
Joined: Thu Dec 10, 2009 4:37 pm

arbitrage16 wrote:
Sun Aug 05, 2018 4:06 pm
80% of losses coming from 20% of your trades/setups...?
That's an interesting question. I've been thinking about this myself lately. Pareto distributions are examples of power law distributions. They are nasty in the sense that they have a mean, but no well defined variance. Well they can, but usually don't. Sometimes even a mean is too much to ask for. This implies that you don't know the impact of outliers in advance and these could potentially kill you. What is normally done is that you clip the outliers (you hedge against them) in order to survive these events (*). The resulting distribution of returns is still not gaussian (there's a bump at the clipping point), but something called a truncated pareto distribution which does have a defined variance.

The reason I was thinking about this was because I saw a hedge fund advertise some wildly optimistic sharpe ratios. It got me thinking how easy it is to manipulate these numbers. Sharpe ratios are defined as the mean minus a risk free rate divided by the square root of the variance.Therefore the whole concept of a sharpe ratio to compare investment proposals hinges on the question whether the underlying distributions have a finite variance or not. I'm pretty sure that wouldn't hold up for these guys.

(*) if you work with other people's money you don't hedge against these events because that will lower your returns and nobody wants to invest with you. If the shit hits the fan you just say sorry to your customers and start a new fund. In many cases the sorry is omitted.
Post Reply

Return to “General discussion”