hi - very interesting - and well done on the successful backtest - it sounds like we're taking a similar situation approach - I'd be interested to find out what sort of metrics you're getting from the system (max drawdown, yield etc) - most important of all in my experience with hedge funds is sharpe (ratio) - how the results correlate to a straight investment in the relevant security - there are a few other more complex ones, but shapre seems to be the equivalent of gross margin on a balance sheet!
I have some contacts at hedge funds and trading firms - but in my experience - unless you're doing something really geeky - the algo funds don't need models - their position is if a model works in real life - building a capital base is fairly straightforward - the hard part has been done!
They'll always require at least a year or so of live trading as well - many systems (including many previous versions of my own!) look great on paper - but practical issue derail them
Slippage and timings are the obvious ones - but other mundane stuff also becomes an issue - biggest one for me last year was overnight (and over weekend) spread / financing charges - GPB/USD spread on IG is 3 for example but if your system (like mine) - holds positions for 10 days - then that charge rises to around 12 pips - per trade!
My system is slightly different to most in that its based on trader psychology. So I use modified indicators (CCI, Trix and others) to create buy and sell signals at market opening across a wide range of securities (forex, commodities and indices). I think of it as the "google" model of trading (as in google's standard spec, but high volume server farms) - in that so long as a trade has a 30% chance of winning I can bet it - with 4:1 win / lose ratios (its not a million miles away from the original turtles model - but I actually didn't read their rules until last summer!)
Drop me an email if you like - or if other people are finding the conversation useful - I'm more than happy to post my backtests here for others to have a look at
Hi SoP
Thanks for writing back. It's great to bounce some results off someone who is on the same page (and Gamma seems cool too!).
I hope you will let me pose some questions regarding your trading after I have posted this.
I call this program Bondi because it catches small waves in both directions. The first time I ran it on probacktest it blew me away. I'd been following it with pencil and paper for a year or two but finally worked out how to explain it to a computer after my summer hols this year. It worked just fine, taking the edge to the market over and over again, and returning with a profit over time, even considering transaction costs etc. The problem was that probacktest plays the FX market in lots, not spot prices, and I found it difficult to get results for, say, risking 1% per bet. Rather, it would increase the number of lots bought which is not the same thing. So, long story sort of short, I had to import each trade into excel by hand.
This involved 6800 trades from EUR/USD, GBP/USD and USD/JPY on daily charts right back to 1975. There many thousands of other trades over 2, 3 days, weeks, months etc. Also over other liquid pairs and indices, commodities, Eurodollar futures etc but for now I leave them to one side.
There are two systems running concurrently within each currency pair. Wave capture and Reverse Wave capture. Each signal generates a trade which will risk 1% of total capital. Hence the maximum exposure can be only 6% long or short the USD.
The systems can oppose each other, ie hedge long against short, as they operate independently of each other.
Here's the results:
BONDI BEACH short wave capture program
-Trading the daily chart from 1975 to current day.
6800 trades in total
Wins: 36.7% of trades
Trade expectancy: +16 pips (including 2 pip transaction cost)
Average profit: 167.79 pips
Average loss: 70.31 pips
Average time in market: 3.4 days
Average exposure: 3.3% (long or short only 2.8%)
Maximum exposure: 6% (less than 1% of the time)
Average annual return: 46.7%
Average monthly return: 3.96%
36 month CALMAR: 1.13
5 year annualised Sharpe ratio (3.5% RF): .89
Worst month: -15.4%
Best month: 60.2%
ROR per trade over the past 20 years is >10%
$1000 invested in 1975 would now be worth $343,252,656
(with 1% reinvested on each trade and an average 2 pip spread)
Thanks for reading. It's easy to get lost in the figures in my office alone. I would really appreciate any questions or comments. I have spent the past weeks trying to shoot holes in the system and results so please feel free to help.
Kind regards
Bluewater18