There's some curious personalities that trade (at least claim to trade) based on news, fundamental metrics, alt data n stuff. I don't mean invest, I mean trade. Well that looks like a skill to be proud off, superstimuli always feels cool aye? Good thing tho there no real reason in doing it all.
The most precise term to explain non-market data is, well, everything that ain't have a direct involvement with what happens inside the order matching servers of a given exchange.
So open interest is in fact a great example of non-market data.
The one & only real purpose for using all this data is to know (not to guess/predict/forecast, not to even anticipate), but to understand when the ACTION is going to happen. If you think deeper, ultimately it's all about asset selection to satisfy whatever purpose you got. if you ever got caught yourself feeling fooled when media release a bad info but prices go up, or media release a good info but prices go down, it's ok. It doesn't work that way, direction of prices can't be affected this way. Direction of prices is the result of how buyers meet sellers which is based on +inf number of factors, where a non-market data is simply just one of these +inf factors. It exclusively provokes action, meat, hype, momentum, volatility, whatever you call it. What's happening is that things start to happen very fast. Without a trigger event, the trading activity would've been the same, it just would've take longer to unwind. News don't change the structure, they make it all happen faster, that's it.
Examples of non-market data that can be used to expect action:
1) Trading schedule, eg the US, EU opening times;
2) Economic releases;
3) Commitment of traders reports;
4) Significant news;
5) Changes in yield curves;
6) "Fundamental" stock data;
7) Open interest;
8) etc etc etc
One really important thing to add is that, just like trading activity is understood in context (other resolutions), sizing also includes context (equity control, market impact), the same way every non-market data event lives in the context (previous releases, other releases, overall economy). You're interesting not in a new per se, but rather in what does it mean in the world. For example, inflation reports don't mean much when the rates are low, but when the rates are high, they trigger significant activity.
That's the area where statistical learning, automated learning, "machine" learning, 'Really' starts to make sense business-wise. The ultimate goal is to create a system that will process every kind of data you have (NLP and TDA should help) and output the tickers with raising/already risen levels of interest.
The most precise term to explain non-market data is, well, everything that ain't have a direct involvement with what happens inside the order matching servers of a given exchange.
So open interest is in fact a great example of non-market data.
The one & only real purpose for using all this data is to know (not to guess/predict/forecast, not to even anticipate), but to understand when the ACTION is going to happen. If you think deeper, ultimately it's all about asset selection to satisfy whatever purpose you got. if you ever got caught yourself feeling fooled when media release a bad info but prices go up, or media release a good info but prices go down, it's ok. It doesn't work that way, direction of prices can't be affected this way. Direction of prices is the result of how buyers meet sellers which is based on +inf number of factors, where a non-market data is simply just one of these +inf factors. It exclusively provokes action, meat, hype, momentum, volatility, whatever you call it. What's happening is that things start to happen very fast. Without a trigger event, the trading activity would've been the same, it just would've take longer to unwind. News don't change the structure, they make it all happen faster, that's it.
Examples of non-market data that can be used to expect action:
1) Trading schedule, eg the US, EU opening times;
2) Economic releases;
3) Commitment of traders reports;
4) Significant news;
5) Changes in yield curves;
6) "Fundamental" stock data;
7) Open interest;
8) etc etc etc
One really important thing to add is that, just like trading activity is understood in context (other resolutions), sizing also includes context (equity control, market impact), the same way every non-market data event lives in the context (previous releases, other releases, overall economy). You're interesting not in a new per se, but rather in what does it mean in the world. For example, inflation reports don't mean much when the rates are low, but when the rates are high, they trigger significant activity.
That's the area where statistical learning, automated learning, "machine" learning, 'Really' starts to make sense business-wise. The ultimate goal is to create a system that will process every kind of data you have (NLP and TDA should help) and output the tickers with raising/already risen levels of interest.
Comment:
More precisely, the goal is to understand when the ACTION gonna be/gonna start happening. The goal is not to catch a spike/structural break, more like to ride the action while it lasts.