Very few trades in 1 minute data, but many trades in 30 minute data
I downloaded data on the ES (and other CME eminis) from Kinetick. I subscribe to their RT data, so could get minute data on the ES back to 2007 from them. My assumption, maybe incorrect, is their historical would be good, I’ve used their RT data, no issue, for ages.
When I test the same strategy on the 1 minute data back to ~2007, with simple strategies, a self created exp. moving average/MACD cross over/ with volume, nothing exotic, using the 1 Minute data (exported from Ninja Trader via Kinetick then imported to SQ), sometimes I get only 2 trades or so, and many, many, many times, I’m getting 0 trades, on both random and genetic build. It does this on all 4 CME eminis.
When I run the exact same strategy on the sample 30 minute sample data (both the ES and TF, the 15 minute sample) included in Strategy Quant, I get hundreds of trades even though the samples are longer time frame and don’t include as many years.
I thought it must be the data, but I checked it, re-downloaded, re-exported, re-checked headings, etc..etc.. and it’s pretty much all default NT settings, it seems fine…I visually checked the headings make sense, so the OHLC colums all seem labelled correctly.
What else can I check to figure out why the same strategy only generates a couple of trades in almost 10 years of 1 minute data, but hundreds of trades on 3 years of 30 minute data? After several days of running Genetic and Random instances on 16 core machines, on the 1 minute data, I have very few strategies that past my (pretty loose, tbh, filter requirements). If I run the same strategy on the sample data, with the same filter requirements – bang many, many qualifying strategies and trades in seconds….
Must be the data right? But it looks ok… (but my background is RT, not experienced in back testing, etc..), what can I check…?
NVM, I did figure it out, it was exactly data errors in export from NT/Kinetick and import to SQ. 🙂
Viewing 2 posts - 1 through 2 (of 2 total)