I have a stock trading strategy in mind which I would like to backtest and trade.
This proves to be rather difficult as I need to scan the entire stock market for a pattern on 1 min bars in real time.
Regarding the backtest it seems equally difficult to find the proper approach as a realistic backtest would cover the next smaller timeframe like seconds or ticks.
Has anyone tried something like this?
Is there any software capable enough so it wouldn't need to be coded entirely from scratch?
Here is what I found out so far:
real time trading:
backtesting:
Has anyone here ever tried something remotely similar?
I do have quite restrictive filters so the approach would only trade about 5-10 stocks a day.
Can't tell whether that is still way too demanding for the backtesting software or the data provider. Would a backtest like that crash the software?
Feedback very much appreciated if you have ever done something remotely similar especially regarding the backtesting process.
This proves to be rather difficult as I need to scan the entire stock market for a pattern on 1 min bars in real time.
Regarding the backtest it seems equally difficult to find the proper approach as a realistic backtest would cover the next smaller timeframe like seconds or ticks.
Has anyone tried something like this?
Is there any software capable enough so it wouldn't need to be coded entirely from scratch?
Here is what I found out so far:
real time trading:
- Multicharts could be capable
- polygon provides real time data but it doesn't seem like multicharts is working with them
- IQFeed restricts the amount of stocks that can be streamed so I would have to choose a subset of stocks that are most likely to pop up in my scanner
backtesting:
- polygon seems to have all the historic data - including one seconds bars - I need but it doesn't seem like multicharts is working with them
- IQFeed has very little data < 1min but enough data going back years on the one minute level + IQFeed has a great reputation
- it would be huge amount of data i.e. I would have to backtest a subset of the market e.g. 2000 stocks and probably test those stocks over a month and then backtest the next month and so on...
Has anyone here ever tried something remotely similar?
I do have quite restrictive filters so the approach would only trade about 5-10 stocks a day.
Can't tell whether that is still way too demanding for the backtesting software or the data provider. Would a backtest like that crash the software?
Feedback very much appreciated if you have ever done something remotely similar especially regarding the backtesting process.
