That makes sense and explains the small size of the candles. But probably not the best choice to use for backtests with longer timeframes. The average ask can be very different from the actual closing ask of the candle which is when you would actually buy the asset.

Yes, i guess the use of it is for benchmarking execution algos.

Have a look at ib-insync.

I'm familiar with that package, I used it for the barplot in the previous post.

Unfortunately IB is returning the same nonsense for this asset using python as it does in Zorro. TRADES mode seems to work, but BID_ASK or BID or ASK returns tons of bogus outliers.