Can Expectancy be positive and CAGR negative?

Discussions about the testing and simulation of mechanical trading systems using historical data and other methods. Trading Blox Customers should post Trading Blox specific questions in the Customer Support forum.
Post Reply
bobsyd
Roundtable Knight
Roundtable Knight
Posts: 361
Joined: Wed Nov 05, 2008 2:49 pm

Can Expectancy be positive and CAGR negative?

Post by bobsyd » Mon Mar 15, 2010 10:15 pm

Noticed some test results with negative CAGR but slightly positive expectancy - initially thought formula/calculation must be wrong but now believe it is possible as shown in the attached simple example.

Any thoughts on why it is possible?
Attachments
CURIOUS EXPECTANCY RESULT.xls
(23.5 KiB) Downloaded 196 times

sluggo
Roundtable Knight
Roundtable Knight
Posts: 2986
Joined: Fri Jun 11, 2004 2:50 pm

Post by sluggo » Mon Mar 15, 2010 11:02 pm

Your math is correct. Your hypothesis, "the quantity Van Tharp calls Expectancy is useful," perhaps not.

For another perspective, take a quick look at the last 3 sentences on page 34 of Ralph Vince's The Mathematics of Money Management. The decision variable that R.V. suggests you rely upon, is negative for your spreadsheet example. As you had hoped.

bobsyd
Roundtable Knight
Roundtable Knight
Posts: 361
Joined: Wed Nov 05, 2008 2:49 pm

Post by bobsyd » Mon Mar 15, 2010 11:30 pm

Unfortunately p34 wasn't one of the preview pages and from what I was able to preview I'm afraid that book is beyond me!

DeanoT
Roundtable Fellow
Roundtable Fellow
Posts: 52
Joined: Sat Apr 15, 2006 7:37 am
Location: Melbourne, Australia

Post by DeanoT » Tue Mar 16, 2010 3:22 am

According to the Kelly formula, your optimal risk for the scenario you demonstrated is:

Kelly = Win % - ((1 - Win %) / (Avg Win/Avg Loss))
= .5 - (( 1 - .5) /(1.02/1.00))
= 0.0098 or 0.98%

Your example bet more than twice the optimal bet size (2%), hence the negative overall result.

It might help to understand this by running your example again, but this time risking a much larger amount, say 50% of your account:

i.e.

Start with $250,000 and win 1.02 of the amount risked ($125,000) and you end up with $377,500.

Risk 50% of your $377,500 and lose, and you are down to $188,750.

Also re-run your simulation on a bet size less than 0.98% and you should see a profit.

http://en.wikipedia.org/wiki/Kelly_formula

added by Moderator: the Kelly formula gives the betsize which maximizes growth in a Bernoulli Trials (2 outcomes, fixed probability) game. Betting less than Kelly gives a slower growth rate, as does betting more than Kelly. If you increase betsize very very slightly above Kelly, bankroll growth diminishes very very slightly. If you continue to increase betsize above Kelly, bankroll growth continues to fall; eventually, growth becomes negative. In the example discussed here, the Critical Overbetting Point, where bankroll growth falls to zero and begins to turn negative, occurs at a betsize of 0.0196078. If the trader in this example bet 1.96078 percent of his bankroll on every trade, his bankroll growth would be zero. Plug in some numbers and try it.

Post Reply