I realize a few people have posted similar issues before here:
IQ Feed for real-time scan
Real life execution not matches backtest results
How to Manage Streaming vs Backfill Discrepancies (5-second charts)
Has anyone been able to improve the accuracy of their real time data scans so that pricing data better matches backtests? For clarity, I am streaming RT data for (1) SPY ATM call option and (2) SPY ATM put option. When I compare the prices (logged on a 1-minute periodicity) to backfilled historical data, they are dramatically different. This has led to many false trading signals.
I have:
- Synched my computer’s clock
- Tried IB data and IQFeed. IQFeed, given the time stamps, is somewhat more accurate but far from perfect for me
- I do pre-scan of relevant tickers
- I have enabled ‘wait for backfill’ on scan project
My database is currently set to 1-min timeframe, but I have used 1-second and tick DB timeframes before while still using 1-min periodicity to generate signals.