assetHistory has a mode for hourly data, but the manual only lists Bittrex and CryptoCompare for this mode. I tried using it with IB and while it looks like it kind of works, it produces a kind of corrupt history that looks like the image below. I'm assuming this means that this feature does not properly support IB? It would be a nice option considering minute data takes a day to download for a single asset due to their API limits.

I think it's more correct to say that IB has bad historical data.

Maybe, but is it THAT bad? Even downloading something liquid like QQQ, these artifacts still exist (though they're more rare). And when viewing the graph in the TWS, the data does not have these errors.

You assume that the data displayed in TWS is the same data output by the Gateway API.

Andrew, would this mean that TWS (probably) has an internal data correction feature?

It's a shame that this reputable broker, which was once a front-runner in trading technology, hasn't fixed these basic issues.

Neither bad data nor TWS correction, but IB hourly data is not probably not implemented in assetHistory. Load M1 and then convert it to H1, or post a feature request for IB H1 with assethistory.

If you are at least slightly familiar with Python you can use ib_insync and its ready-to-use notebooks for basic tasks: https://nbviewer.org/github/erdewit/ib_insync/blob/master/notebooks/bar_data.ipynb

While we're on the topic the IB api... Does anyone know exactly what their BID_ASK mode returns for the Open and Close? The doc states "Time average bid" for Open and "Time average ask" for Close but I'm not sure what that means and I don't find it explained anywhere in the doc. The bars it returns look strange. See below comparison of TRADES mode vs BID_ASK mode

Try this.

Outlier = 0;

'Time average bid' is the average of all bids during the bar; same for asks. Similar to what Zorro delivers with price().

The bid-ask spread is negligible relative to a 1-hr high-low range - so you see these marks as almost one on the chart.