I've now been trading the ASX200 systematically, mostly on Plus500, for two weeks. I'm only ahead by about AUD 500, but that's not surprising. Actually, it's a good result given the small position size so far, and the potential for drawdowns. I've done more backtesting and improved the algorithm. About half of total returns would come from holding winning positions (during the day) overnight and I haven't done that yet. I also updated my soybean trading model with the last year or so of data and the algorithm I was using would continue to be profitable, so I am adding that. By contrast with the ASX200 model, this is a trend following model that doesn't trade very often. Finally, during the past week I did some trading of Treasury Wine around the high volatility in the stock with the earnings release and broker upgrades and downgrades. This involved far larger fluctuations in value than the systematic trading.
Sunday, February 21, 2021
Trading Update
Subscribe to:
Post Comments (Atom)
5 comments:
Aside from using algorithms that have been developed and back-test from historic data, is there some fancy statistical analysis of recent trading data that could alert you when market behaviour is becoming atypical? The biggest trading blow-ups seem to occur when there is a sudden change in what is 'normal'. Getting closed out repeatedly by stop-loss orders, and having a significant run of losing trades is probably not the best way to find out that a trading model is no longer applicable/working well.
I know people monitor market volatility (and presumably individual stock price volatility) to see when market conditions change, but perhaps there are some other signals that could provide quick notification if things are starting to deviate from historic norms. Maybe the skewness of the distribution of daily price tick movements or some such?
In backtesting when these models don't work they just don't make money, so it becomes like just rolling dice. They don't really lose money systematically. Really the method is rolling dice while trying to take advantage of momentum and reduce losses with stop losses. I don't think you can actually predict anything.
BTW these sort of models make most money when volatility is high. I am thinking of adding some measure of longer term trend to the ASX daytrading model. When there isn't much trend you probably don't want to trade. This is the next research project.
I decided last night not to trade TSLA or GOOGL and instead place a long-term trade for a small amount of ASX200 index to basically just gear into the normal long-term rise in share markets (I'm not entirely sure what the overnight interest charges on the position will be pa - it might not actually be worth doing even if the market returns 9% pa in the long term). Of course as soon as I had that long position in place the ASX200 dropped about 2% today ;) Fortunately it is such a small position that I only lost about $35 overnight, and the stop-loss won't kill the position unless I'm down by about $150 (I think that requires about 10% drop in the index). I tend to muck around with trading for fun rather than take a systematic approach - for example I only found out this morning that my iShares MSCI Australia Index Fund CFD order had been filled and traded overnight. Seems that this is some weird US-based trading of the ASX200 when out market is closed! Given the Australian market dropped a lot today the CFD trade may lose even more overnight. But in the long run it should (hopefully) track the ASX200 index (although it might be also affected by the AUD/USD exchange rate if it isn't hedged into AUD...)
Yes SPI trades from 5:10pm to 8:00am and then from 9:50am to 4:30pm each day. So, it just closed now on Saturday morning. This is essential to my strategy :)
Post a Comment