Yes
I created custom strategy with only one indicator - qtpylib.awesome_oscillator. After that I ran:
python3 ./scripts/plot_dataframe.py -s CustomStrategy -p ETH/BTC --indicators2 ao --timerange 20190524-20190526
and opened user_data/freqtrade-plot.html.
The awesome oscillator on freqtrade-plot has a lot of missing values(but of course tradingview.com has them). I get this issue for all pairs, and I just can't buy due to this.

awesome oscillator in qtpy uses the exponential weighted function with default length of 34:
def awesome_oscillator(df, weighted=False, fast=5, slow=34)
That's what you see on the graph: first value is printed after 34 starting candles. These are not 'missing' values, it's a starting period for calculation of the first value of the mean function. If you have used EMA(100) in the strategy, you would get first value of EMA only after 100 starting candles.
Your time range is just 2 days, i.e. only 48 1h-candles. Try to use a longer time range.
In addition to the above:
Tradingview has the full length of the indicator because it has data from before what you see - however you should be able to archive this as follows:
You could also run backtesting seperately (with a longer timerange) by using --export trades --export-filename user_data/backtest_export.json and then plot a smaller timerange (not at the beginning of the backtest period) to avoid this problem (don't forget to add --export-filename user_data/backtest_export.json to the plot-script so it's using the exported trades.
@xmatthias "with a longer timerange" 馃槑
Thanks for example of partial plotting of the results, I did not think of such a use case.
This probably should be fixed somewhere in the docs in a section like "Efficient usage of freqtrade and its tools".
Many thanks for the explanation! Also, I have another question, more related to the ta-lib, but I think you can help me with it too:
https://github.com/mrjbq7/ta-lib/issues/266
Most helpful comment
In addition to the above:
Tradingview has the full length of the indicator because it has data from before what you see - however you should be able to archive this as follows:
You could also run backtesting seperately (with a longer timerange) by using
--export trades --export-filename user_data/backtest_export.jsonand then plot a smaller timerange (not at the beginning of the backtest period) to avoid this problem (don't forget to add--export-filename user_data/backtest_export.jsonto the plot-script so it's using the exported trades.