Today's trade was a winner, recovering a bit more than half of yesterday's loss. This is going to need a lot more research. Especially around the relationship between volatility and returns. I suspect that it's not worth trading when volatility is low and maybe there is an upside limit to how much volatility to tolerate. Now I have 13 months of data (from Barchart) nicely organized. Just need to merge all the spreadsheets together and test hypotheses.
Tuesday, February 02, 2021
2nd Day of Trading
Subscribe to:
Post Comments (Atom)
3 comments:
Is there a difference in how your trading strategy is implemented when back-testing on historic data and how it is actually implemented when 'live'? Basically, if your model/rules/criteria are automatically applied when tested against historic data, but you have to implement parts of your strategy manually when trading 'live', then the human element/behavioral biases might be making your strategy work better 'on paper' (back-testing) then when implemented for real.
I do manual trades on IG but I know that they have a couple of different automated trading platforms available (MT4 and PRT) that might let you fully automate your 'system'.
Apparently Plus500 is cheap to trade, but lacks much ability to do automated trading systems: "Advanced traders who rely on incorporating third-party analytical and automation tools in their trading process will be disappointed as none of these tools can be integrated directly with WebTrader. Moreover, back-testing functionality is not available either."
I'm using Plus500 right now because you can take a smaller position than the full size SPI futures contract. This is especially important when the size of the trade needs to be modified for risk. The human elements here are two - they won't let you specify an automatic entry price that is very close to the most recently traded price. So then you have to do it manually and there can be slippage either way. Also, trades that aren't stopped out need to be manually closed. So the exact minute you close the trade affects the result whereas the backtest assumes that it was closed exactly at 4:30pm each day (I only downloaded hourly data). With Interactive Brokers there are good after time trades which can place market orders at specific times. These can also be made conditional on the position still being open (not stopped out). So if/when I go to the full size trades I will be able to use that without building a fully auto system. But if it's successful I likely will try to build a fully auto system at some stage.
I watched a youtube video last night about setting up a trading bot using simple python coding of the algorithm as platform QuantConnect. Its apparently free to try coding a trading algorithm and back-testing, and for live trades IB was one of the available accounts. Seems to be US centric, but that might suit you if you're interested in trading S&P500 etc.
The algorithmic trading looked fun to dabble with:
https://www.youtube.com/watch?v=s8uyLscRl-Q
Browsing through QuantConnect they also appear to sell subscriptions to particular quant trading algorithms (for those who can't or don't want to develop their own). If your rocket science trading strategies work out, you might be able to monetize it as a 'black box' people can subscribe to emulate in QuantConnect. ;)
Post a Comment