The pain of News Corp

Of course, the worst month(s) come as soon as I have scaled up the capital haha.

Oh the joys of trading !

(August doesn't look crash hot either)

Time to scale up

June was disappointing, with a result of -2.17%, bringing our first monthly loss since January.

So overall for the half year we are +34.23%.

Starting from next month, I will be scaling up the system, so from July 1st, starting capital will be US$60,000. I'll probably start the trade log and %chart again, but the monthly profit chart will be able to show percentage performance regardless of the capital base.

I have also made a change to the position sizing of the system, to make the system more aggressive. We are now allocating more capital to each trade, 25% instead of 20%, which means we are maxed out at 4 positions.

May performance

Quiet month in terms of number of trades, but we hit our numbers. +4.53% for May which brings us to +36.40% YTD.



I've also been busy designing a new system, called TPO, another US stocks, mean-reversion based automated system, which is currently being papertraded. I couldn't trust any of my backtests with this one as it really does require intraday data to be accurately tested. So we'll run it for 6-12 months on IB's papertrader and see how the numbers stack up.

April performance

April was a steady month, without being spectactular, with a result of +4.30%.

There were 20 trading days in April (US markets were only closed on Good Friday), the system had some activity on 8 of those days. From these 8, only on one occasion were we fully invested.

Six months in

+8.53% for March and +27.57% for the first quarter. We've hit the 6 month mark so I've added a few more charts.

While results have been promising, I would expect the payoff ratio to even out over time and the hit rate to edge closer to 70 than 60%.

On March 15th, IB's margin risk controls didn't seem to kick in so I found myself holding 21 positions, well beyond my maximum of 5 or 6. So I just closed all positions (out of panic) 3mins into the session and ended up +$1499 for that day.

Monte carlo analysis shows that performance for March 15th, had the number of positions been restricted as per the system, could have been anywhere between -$200 and +$2258 depending on which stocks I bought, with the median being $799.

So while the broker error didn't leave me much to complain about(!), it did show that automated trading systems still need monitoring. Especially if your system buys weaknesss and you can see the futures are down 2% in the pre-open!

Correlation between volatility and performance

As a system (AGT is what I call my system) that trades intraday, its no surprise that it performs well in times of volatility where the intraday range expands, which is what typically happens during bearish times.

So it has got a bit of an insurance mechanism built-in to protect from Black Swans. But unlike the techniques employed by funds like Universa, AGT is not too shabby during uptrends either.

The peak of the GFC, which is from September 15th 2008 (Lehman Brothers collapse) until March 9th 2009 which was the bottom, when the VIX went crazy, is also the time AGT went nuts, and tacked on 83% in under 6 months.

When it rains, it pours

January performance

December performance