I've struggled with whether or not to say, as on the one hand it feels like criticising your friends baby, but equally not saying anything feels equally uncomfortable.
I grabbed the historic GIEW yearly gains you posted last year, tagged on your 2010 gain and then compared those 1999 to 2010 returns with the Coffee House Portfolio using Simba's backtest spreadsheet - and saw this
Doing another comparison with my own figures, we come out at very similar yearly averages (mean) since 1999, but your standard deviation is way way higher - which results in your annualised falling behind.
Looking a little deeper I get the impression that when you get sales often you immediately reinvest those funds in other candidates, maybe starting new AIM's? Which maintains a higher average 'at-risk' amount, but does bear a greater impact when a large decline is encountered.
I've an online calculator at http://www.jfholdings.pwp.blueyonder.co.uk/cagr.htm in which you can plug yearly mean gains and standard deviation to see the approximated annualised gain that that produces. As an example plug in a 11% mean and 40% standard deviation and out pops a 3.5% annualised (CAGR). For the same mean but with a 10% standard deviation and out pops a 10.5% annualised.
At the end of the day its the annualised that is the real money we get to keep/spend and your total fund value volatility over time appears to be killing all of your trading benefits and perhaps more.
Just a friendly well intentioned nudge towards something that you might like to look into.
Market Data 
Markets 
AI
